Skip to main content

An Efficient wall-distance computation in RANS flow simulation using KD-tree

Chaewon Jeong, Haeun Lee, Jin Seok Park

2026-03-01

An Efficient wall-distance computation in RANS flow simulation using KD-tree

Chaewon Jeong, Haeun Lee, Jin Seok Park
Journal of computational fluids engineering, vol. 31(1), pp. 65–78, 2026

초록

In this study, a KD-tree-based wall-distance computation method is proposed to improve computational efficiency in compressible turbulent flow simulations on large unstructured grids. In Reynolds-averaged Navier-Stokes (RANS) turbulence modeling, accurate evaluation of the wall distance is essential for predicting near-wall turbulent behavior, while conventional Brute-force approaches suffer from rapidly increasing computational cost as the grid size grows. To improve computational efficiency, a two-step KD-tree search strategy is introduced to significantly reduce the number of wall elements involved in the distance calculation: candidate wall elements are first identified through a KD-tree-based proximity search, and the exact wall distance is then computed only for the selected candidates using a point-to-triangle geometric formulation. The proposed method is validated using the HB-2 configuration, the ONERA M6 wing, and the NASA CRM aircraft, demonstrating a substantial reduction in computational time compared to the Brute-force approach while maintaining reliable aerodynamic prediction accuracy.

BibTeX

@article{jeong_efficient_2026,
    title = {An Efficient wall-distance computation in {RANS} flow simulation using {KD}-tree},
    volume = {31},
    issn = {1598-6071, 3022-6252},
    url = {https://journal.kscfe.or.kr/articles/doi/10.6112/kscfe.2026.31.1.065},
    doi = {10.6112/kscfe.2026.31.1.065},
    abstract = {In this study, a KD-tree-based wall-distance computation method is proposed to improve computational efficiency in compressible turbulent flow simulations on large unstructured grids. In Reynolds-averaged Navier-Stokes (RANS) turbulence modeling, accurate evaluation of the wall distance is essential for predicting near-wall turbulent behavior, while conventional Brute-force approaches suffer from rapidly increasing computational cost as the grid size grows. To improve computational efficiency, a two-step KD-tree search strategy is introduced to significantly reduce the number of wall elements involved in the distance calculation: candidate wall elements are first identified through a KD-tree-based proximity search, and the exact wall distance is then computed only for the selected candidates using a point-to-triangle geometric formulation. The proposed method is validated using the HB-2 configuration, the ONERA M6 wing, and the NASA CRM aircraft, demonstrating a substantial reduction in computational time compared to the Brute-force approach while maintaining reliable aerodynamic prediction accuracy.},
    language = {ko},
    number = {1},
    urldate = {2026-04-17},
    journal = {Journal of computational fluids engineering},
    author = {Jeong, Chaewon and Lee, Haeun and Park, Jin Seok},
    month = {March},
    year = {2026},
    pages = {65--78},
    pubtype = {국내저널},
}