Surrogate-assisted multi-objective optimization of leaf-vein textured journal bearings under thermohydrodynamic lubrication

Surface texturing can improve the thermohydrodynamic (THD) lubrication of journal bearings, yet THD benefits depend strongly on texture topology. This study proposes a symmetric leaf-vein (SLV) texture and evaluates its THD lubrication performance using a three-dimensional CFD-THD multiphase flow model, surrogate-assisted multi-objective optimization, and tribological experiments. Relative to the smooth surface, rectangular and herringbone textures, the SLV texture promotes lateral lubricant transport and hydrodynamic pressure build-up, leading to lower friction and a reduced maximum oil-film temperature T max under the investigated operating condition. A parametric study shows that localized texturing in the converging wedge maximizes the dimensionless load-carrying capacity (LCC) (up to +23.36% versus the smooth bearing), whereas full-coverage texturing most effectively suppresses T max (up to −2.77 K). An optimal dimensionless depth of h p ⁎ ≈1.0 is identified; deeper grooves intensify lubricant entrapment and vortex-related viscous dissipation, degrading LCC and causing a temperature rebound. For design optimization, a TabPFN surrogate is constructed to map key SLV parameters ( L ⁎ , D 1 ⁎ , h p ⁎ , θ ) to THD metrics (friction f , LCC, T max ) and is coupled with NSGA-II; an entropy-weighted TOPSIS method selects a representative optimum. Compared with the smooth bearing, the optimized SLV design reduces cavitation area and dimensionless friction by 23.02% and 19.3%, increases LCC by 14.2%, and lowers T max by 2.22 K. Experiments corroborate these trends, with the optimized texture reducing the coefficient of friction by up to 26.7% and the measured oil-film temperature by up to 4.3 K. • Systematic analysis of 3D CFD-THD model for leaf-vein textured journal bearing. • Leaf-vein texturing reduces COF/ T max by up to 18.6%/2.77 K vs. the smooth bearing. • TabPFN surrogate with NSGA-II enables efficient multi-objective optimization. • Optimized leaf-vein texture reduces COF 35.9% and lowers T max by 2.22 K. • Tribo-tests confirm COF decrease and up to 4.3 K lower oil-film temperature.