In this work, flow databases from direct numerical simulations were used to study the physics of turbulent heat and mass transfer in non-reacting and reacting flows. Statistical analysis and visualizations were used to gather information about turbulent scalar transport to help scalar transport model development and validation.
Passive heat transfer in non-reacting flows was investigated using simulations of a channel flow and a plane Couette flow. Flow databases were created for each case, for a Reynolds number of 3000, based on the mean velocity and the channel half-width. The flows were bounded by isothermal walls with one wall at a higher temperature than the other. Near the wall, the heat transfer mechanisms are similar in both flows. In this near wall heat transfer, streamwise vortices play a significant role. At the centerline, however, the Couette flow has large scale flow structures, not present in the channel flow, that transport heat across the centerline. These structures are possible due to the non-zero turbulent production at the centerline of the Couette flow. For the channel flow, the lower turbulence levels at the centerline are not effective in breaking up packets of hot or cold fluid that originate at the walls. Instead, these packets convect with the mean flow resulting in a less efficient heat transfer than in the Couette flow.
To study heat and mass transfer in reacting flows, a statistical analysis was performed on existing databases of spatially developing, three-dimensional, reacting and non-reacting shear layers. Turbulent Prandtl and Schmidt numbers were calculated to characterize turbulent heat and mass transfer. The time averaged scalar equations and the budgets for fluctuating scalars showed interesting similarities between temperature in the non-reacting shear layer and oxidizer mass fraction in the reacting case. These similarities correspond to the similarity in the mean profiles for these quantities. In the budgets for fluctuating scalars in the reacting flow, however, the turbulent transport term is more important than in the non-reacting case and counters the peaks in turbulent production.