Simulation of Mixing of Single-Phase Fluids in T-Junctions

The purpose of this study is to sample a procedure for numerical simulation and calculation of the processes of mixing in pipes of a T-junction (tee) of natural gas with the so-called “stripped” components, such as methane, hydrogen, and nitrogen, to obtain a mixture that can be used as a fuel at thermal power plants. The specific of fuel gas mixing is high Reynolds numbers of the simulated flows, which can be as high as Re = (5–10) × 10⁶. An analysis is presented of some experimental and modern computational studies of the processes of flow mixing in pipes and T-junctions. It is pointed out that the application of various well-accepted models for eddy viscosity or Reynolds stresses in the numerical simulation on the basis of Reynolds-averaged conservation equations yields a satisfactory agreement with experimental data on mixing flows in a T-mixer only with an unjustified decrease of the turbulent Schmidt (Prandtl) number to the value 0.1 or an increase of the known constant of turbulence models C_μ by a factor of 9. It can be concluded that eddy-resolving methods are unsuitable for the investigation of mixing processes in fuel pipeline joints due to high Reynolds numbers and a great length of the main pipe. An analysis of the predictions has revealed large fluctuations in the local ratio of the generation rate of the turbulent kinetic energy to the rate of its dissipation and a sharp decrease in its value averaged over the pipe cross section at a distance of several diameters from the starting point of mixing, which is not characteristic of pipe flows, mixing layers, or jets. An attempt was made to improve the predictive capabilities of the standard k–ε model for developed turbulence, while keeping the turbulent Schmidt number \({\text{S}}{{{\text{c}}}_{t}}\) and the constant С_μ within the substantiated limits. An empirical formula for \({\text{S}}{{{\text{c}}}_{t}}\) and a modification of the standard k–ε model, which takes into account the variability of С_μ according to the Rodi dependence carefully verified against data on various free flows, are proposed. Experimental investigations of isothermal mixing of air flows in a tee mixer, one of which contained tracers in the form of glycerin-based liquid microdroplets, were carried out. The profiles of hydrodynamic characteristics of the flow downstream of the tee were measured by the planar optical SIV method at a distance of 5.5D from the axis of the pipes' intersection. To verify the modified k–ε model, numerical simulation was performed of the mixing of gases and liquids in a tee mixer, and the predictions were compared with the experiment. The results are presented of the calculation of natural gas mixing in a tee mixer with a methane-hydrogen fraction from petrochemical facilities.