Turbulent Heat Transfer in Supercritical Fluids Under Transcritical Temperature Conditions

Boiling at the ambient pressure undergoes a critical heat flux in the nucleate boiling regime and it determines the thermal efficiency and the applicable range of heat flux (low/moderate heat flux, \u3c105 W/m2; high heat flux, 2.5×105–107 W/m2; ultrahigh heat flux, 107–109 W/m2). Since the regime beyond the critical heat flux shows a significant reduction in the overall achievable heat transfer rate, it is a reference for efficient heat transfer. We propose to investigate supercritical fluids to overcome such a limit. A supercritical state is reached when the fluid is at temperatures and pressures exceeding its critical point. The supercritical fluid has simultaneously a liquid-like density and gas-like diffusivity, without a distinctive phasechange or surface tension, with the potential of overcoming deleterious effects and limitations of classical boiling. In this study, we have performed direct numerical simulations solving the compressible Naview–Stokes equations for natural (R-134a, carbon dioxide, and methanol) and forced (R-134a) convection in transcritical temperature ranges in order to investigate and discuss the phenomena of pseudophase change with a specific focus on the heat transfer and turbulence structures/dynamics using structural and statistical approaches. In natural convection pseudoboiling at supercritical pressure in transcritical temperature ranges (ΔT =1 K, 5K,20K, and40KwhereΔT is bottom-to-top temperature difference), increasing ΔT leads to the higher recirculation frequencies, enhancement of heat transfer, and breakup of global circulating motion. Two-dimensional simulations have limitations to proper prediction of the flow dynamics and heat flux compared to three-dimensional ones. The proposed heat transfer correlation explains the natural convection pseudoboiling well, yet not perfectly. The real fluid effects cause the large thermodynamic gradients at the pseudotransition interface and increasing ΔT requires finer grid resolution due to unresolved length scales. Turbulent forced convection is studied with ΔT = 5 K, 10 K, and 20 K (Ttop/bot = Tpb ± ΔT/2). At these conditions, a pseudophase change occurs at various wall-normal locations within the turbulent channel from ypb/h = −0.23 (ΔT = 5 K) to 0.89 (ΔT = 20 K), where h is the channel half-height and y = 0 the centerplane position. Increases in ΔT also result in increasing wall-normal gradients in the semi-local friction Reynolds number. Classical, compressible scaling laws of the mean velocity profile are unable to fully collapse real fluid effects in this flow. The proximity to the pseudotransitioning layer inhibits the turbulent velocity fluctuations, while enhancing the temperature and density fluctuations. The latter reach peak values (relative to their mean) comparable to what is observed in a M =3.0 ideal gas isothermal-wall compressible turbulent channel flow. Conditional probability analysis reveals that the sheet of fluid undergoing pseudophase change is characterized by a dramatic reduction in the kurtosis of density fluctuations, hence becoming thinner as ΔT is increased. Instantaneous visualizations show dense fluid ejections from the pseudoliquid viscous sublayer, some reaching the channel core, causing positive values of density skewness in the respective buffer layer region (vice versa for the top wall) and an impoverishment of the turbulent flow structure population near pseudotransitioning conditions