Nanofluid channel convection: Mechanistic analysis of corner-induced thermal boundary-layer modification

Efficient thermal management in compact channels requires maximizing heat transfer while controlling hydraulic resistance. Previous studies have coupled changes in projected area, blockage ratio, and streamwise location with shape, obscuring the independent roles of curvature and corner sharpness. This study employs a double multiple-relaxation-time lattice Boltzmann method to examine forced convection of a graphene oxide–water nanofluid in a two-dimensional baffled channel. Circular, elliptical, hexagonal, and teardrop-shaped heated inserts with identical projected areas are systematically compared over 50 ≤ Re ≤ 500 and 0% ≤ φ ≤ 4%, enabling a controlled equal-area comparison that isolates geometry effects on near-wall transport. Temperature contours, local Nusselt number (Nu) distributions, pressure penalty, and the performance evaluation criterion (PEC) are interpreted as successive manifestations of geometry-controlled thermal gradient renewal. Increasing Re compresses the thermal boundary layer, while increasing φ augments conductive-diffusive transport through effective property enhancement. The hexagonal insert produces the strongest corner-induced thermal-gradient renewal and achieves the maximum average Nusselt number at Re = 500 and φ = 4%, but imposes the largest hydraulic penalty. At Re = 400 and φ = 4%, the elliptical insert sacrifices only modest heat transfer while substantially reducing the pressure penalty, yielding the highest PEC, with the teardrop insert close behind. These results reveal a geometry-dependent inversion: the geometry that maximizes Nu does not maximize PEC. This inversion provides a physically based design criterion: sharp-cornered inserts are preferable when heat removal dominates, whereas streamlined inserts are more advantageous under pumping-power constraints.