No abstract

No abstract

In this paper, a mixing of gases through square Jets issuing normally Into a CrossFlow (JICF) is investigated by means of both numerical simulation and experiment. The jets are emitted by two injectors mounted at the top and bottom of an Injector Frame (IF) which is installed at the center of an Eiffel type wind-tunnel. This jet configuration makes it possible to approximate an industrial gas mixer placed at the center of a pipe. Large Eddy Simulation based on the Smagorinsky model is used, enabling characterization of the mean and fluctuating velocities as well as the oscillating flow frequencies. Different diagnostic techniques, such as Laser Doppler Anemometry and Particle Image Velocimetry are employed for validating the numerical models, and a good agreement between prediction and experiment is obtained. In the numerical simulation, introduction of a passive scalar through the jet makes it possible to show three dilution phenomena. They are generated respectively by the wake of the IF, the jet/wake assemblage and the jets alone in function of the momentum flux ratio between jet and crossflow. Influence of the various parameters on the mixing process between the jets and the crossflow is identified. The numerical results show that if the IF wake is suppressed with the presence of a trailing edge behind the IF, classical formation of Counter-rotating Vortex Pair is found. NomenclatureC = Smagorinsky constant D = characteristic length (m) Esd = energy spectral density (m 2 s −1 ) F = frequency (Hz) g = gravity vector (m s −2 ) J = ρ jet U 2 jet ρ cf U 2 cf : momentum flux ratio between jet and crossflow M k = molar mass of the species k (kg mol −1 ) p = pressure (Pa) R = ideal gas constant (J mol −1 K −1 ) Re cf = U cf D jet ν cf : Reynolds number built on U cf and D jet 356 A. GOURARA ET AL. Re inj = U cf D inj ν cf : Reynolds number built on U cf and D inj Re jet = U jet D jet ν jet : Reynolds number built on U jet and D jet R v = velocity ratio between jet and crossflow S i j = large scale strain rate tensor (s −1 ) |S i j | = magnitude of the large scale strain rate tensor (s −1 ) Sc t = turbulent Schmidt number St = F D U : Strouhal number t = time (s) T = temperature (K) Tu = ( u 2 1 +u 2 2 +u 2 3 3U 2 cf ) 0.5 : turbulence rate U = (u 1 , u 2 , u 3 ): velocity vector (m s −1 ) U = bulk velocity (m s −1 ) x = (x 1 , x 2 , x 3 ) ≡ (x, y, z) : coordinates (m) Y k = mass fraction of the species k = filter size (m) = Von Karman vortex magnitude (m) ν = kinematic viscosity (m 2 s −1 )Y k 3 ): subgrid scalar transport flux of the species k (kg m −2 s −1 ) Subscripts and Superscripts 0 = operating conditions cf = crossflow cl = centerline CVP = counter-rotating vortex pair inj = injector frame (IF) jet = jet max = maximum sg = subgrid scales te = trailing edge VK = Von Karman vortices behind the IF Operators = filter ∼ = Favre filter = time averaging operator

Computational fluid dynamics based on Reynolds averaged Navier–Stokes equations is used to model a turbulent planar buoyant adiabatic wall plume. The plume is generated by directing a helium/air source upwards at the base of the wall. Far from the source, the resulting plume becomes self-similar to a good approximation. Several turbulence models based predominantly on the k-ε modeling technique, including algebraic stress modeling, are examined and evaluated against experimental data for the mean mixture fraction, the mixture fraction fluctuations, the mean velocity, and the Reynolds shear stress. Several versions of the k-ε model are identified that can predict important flow quantities with reasonable accuracy. Some new results are presented for the variation in a mixing function for the mixture normal to the wall. Finally, the predicted (velocity) lateral spread is as expected smaller for wall flows in comparison to the free flows, but quite importantly, it depends on the wall boundary conditions in agreement with experiments, i.e., it is larger for adiabatic than for hot wall plumes.