Heat transfer model for horizontal flows of CO^sub 2^ at supercritical pressures in terms of mixed convection

When a material crosses its critical point or a pseudocritical temperature at which the specific heat of the material at a constant pressure is maximal, the thermal and hydraulic properties vary significantly. Buoyancy, induced by a great density variation of near-wall fluid, results in asymmetric heat transfer coefficients at the top and bottom walls of a horizontal circular channel. However, flow acceleration, which occurs in the flow direction because of significant density variations, has an identical effect on heat transfer regardless of the flow direction. For this reason, only the acceleration effect can be investigated in terms of a turbulent shear stress variation for horizontal flows. Therefore, a study on the buoyancy effect for horizontal flows is required with respect to not the shear stress variation but different aspects between the top and bottom walls. In this study, semi-empirical and empirical heat transfer models were proposed based on a mixed convection. The models were evaluated using experimental data. The semi-empirical model has a mean absolute difference (MAD), the average error, of 21.73% for the top wall and 22.35% for the bottom wall. However, the empirical model has a MAD of 10.00% for the top wall and 10.44% for the bottom wall. The proposed models significantly improve the prediction accuracy of the Nusselt number at each wall, as well as for the average Nusselt number compared to the previous correlations.