Entropy production of emerging turbulent scales in a temporal supercritical n-heptane/nitrogen three-dimensional mixing layer

A study of emerging turbulent scales entropy production was conducted for a supercritical shear layer as a precursor to the eventual modeling of subgrid scales (SGS: from a turbulent state) leading to large eddy simulations (LES). The entropy equation was first developed for a real, non-ideal fluid using a validated all-pressure fluid model, and the entropy flux and production terms were identified. Employing a direct numerical simulation (DNS) created database of a temporal three-dimensional supercritical shear layer using the fluid model, the different contributions to the irreversible entropy production term were evaluated. Both domain averaged and root mean square (RMS) terms were computed at three different stages of the DNS, representing the timewise ascending, culmination, and descending branches of the spatially averaged positive spanwise vorticity. The unifltered and filtered databases were compared to evaluate the relative importance of irreversible entropy production from viscous, Fourier heat diffusion, and molar fluxes terms. The results show that the average entropy production is dominated by the viscous terms at all stages of the evolution: however, the contribution to the RMS of the molar flux term for both the ascending and descending branches is non-negligible. This latter result was traced to the molar gradients tending to be smeared by emerging turbulent scales. Based on this finding a physical picture of the layer evolution was presented involving competition between large scales entraining heavy fluid from the lower stream and forming strong density and mass fraction gradients at spatially varying locations with time, and small-scale turbulent structures evolving but being damped by contact with the newly formed strong density gradient regions which act similar to material surfaces. Analysis of the results showed that the primary contribution to the molar flux dissipation for both the average and the RMS is the mixture non-ideality.