Case 1 : Vf=50m/s

On the animation below you can see the evolution of the flow inside the injector with a fuel injection of 50 m/s. We can see that the flow presents a transcient flow at the beginning where the fuel begins to enter in the injector. After few milli-seconds, we can observe an steady state where fuel and air are mixed together. We can see that there are air-pockets which are carrying out inside the mixture at the exit of the injector. The mixture is homogeneous except those air-pockets which alternatively appear at the top and at the bottom of the domain.

Once the steady flow sat up, we have done an average on the value of the fuel mass fraction in the domain. Alternative air-pockets crossing outlet obliged to make an average. On the figure above we have the evolution of the fuel mass fraction along the terminal cross section of the injector (x=0,21m). Average field of fuel mass fraction

Fuel mass fraction repartition at the outlet shows that the mixture is well homogeneous (in opposition to Case 2 and Case 3 where fuel velocity were less important ). But there is a dissymmetry between the upper wall (scale vertical coordinate 1) and the lower one (scale vertical coordinate 0), which is not physicaly correct because all the datas (the geometry, the initialization and the boundary conditions) are symetric.

A formation of a structure can be noticed at the left of the lower intake of fuel as if the flow comes back to the left of the domain. This phenomenon is more important for the lower intake than for the upper one. Vectors velocity

These two facts can maybe be explained by the cells repartition of the grid which is not symetric ( it is an unstructured grid ), and near the lower entering the grid is wider. So maybe the vortice which appears near the lower wall is due to this fact : it is an numerical effect which is not physical at all. There is no vortice near the upper wall, while there's one near the lower wall but if it might appear by extended simulation time as the last picture of the animation seems to show.

An entering of fuel at the outlet can also be noticed at the end of the animation. Here again it is a numerical effect (which is not physical at all). This phenomenon seems to be due to the domain which is not long enought for the simulation, in fact too big structures go out from the domain to be well simulated by boundary conditions. Field of fuel mass fraction (iteration 77000)

Note:   AVBP calculates with a constant CFL, so time step is variable during the calculation that's why we prefer speak in number of iterations instead of time of simulation. 