An improved CFD wall heat transfer model based on a critical analysis in heavy-duty diesel engines

There is a growing demand for higher power density in diesel engines, making thermal management increasingly critical. 3D-CFD (Three-Dimensional Computational Fluid Dynamics) simulations play an important role in the thermal analysis of heavy-duty diesel engine pistons due to the complex heat transfer phenomena associated with sprays in diesel combustion. 3D CFD simulations rely on wall heat transfer sub-models to estimate heat flux at the combustion chamber surfaces, which are then used as boundary conditions to perform 3D finite element thermal analysis. Equilibrium wall heat transfer models are widely adopted alongside RANS and LES turbulence modeling approaches to minimize computational cost. In this study, fast-response surface thermocouples were used to validate local heat-flux predictions on the piston surface of a heavy-duty diesel engine. Comparisons between CFD-predicted heat fluxes using existing equilibrium models and experimental data revealed notable discrepancies and inconsistencies across various piston locations and operating conditions. A comprehensive theoretical and numerical analysis of the existing steady and unsteady heat transfer formulations was conducted to identify the root causes of their inconsistencies. These inconsistencies were primarily attributed to strong spatial variations in the near-wall flow field generated by spray plumes. Existing formulations generally use a two-layer approach consisting of the viscous sub-layer and the mixed/turbulent layer, which limits their ability to represent diverse local turbulence conditions. To address these limitations, an improved equilibrium heat transfer model is proposed. The model adopts a three-layer formulation in which the turbulent region is further discretized into two sub-layers, as shown in the equation below. This added resolution enables the model to better account for the spatial heterogeneity of combustion under varying engine loads. The proposed three-layer approach showed significant improvements in prediction accuracy and consistency compared with existing models across diverse local flow conditions. y + < 13.2 − ρ c p u τ q w T g − T w = Pr . y + y + > 13.2 ρ c p u τ q w T g . ln T g T w = 2.1 ln y + + 13.2 Pr m − 5.34 y + > 200 ρ c p u τ q w T g − T w = 2.1 ln y + + 13.2 Pr m − 5.34