50 m/s fuel injection

In order to obtain a mixing area in the exit of the injector we have increased the fuel velocity at 50 m/s. We start simulation from the initial condition until a time of 35 ms (80000 iterations).
In this case the flow is not symmetrical. We can see that on the shaded contours  and on the cross sections of fuel mass fraction.
The most important phenomenon which we can see is the existence of a large vortex which is coming undone from the down fuel injector (with a ? form). This sort of structure is not physical. Indeed, if we see the shaded contour of U velocity, we note that the recirculation area is in the outlet cross section. It means that there are back flows in this section. Even if it could be physically possible, numerically, this is not acceptable. We can also see the existence of two fuel pockets which are going back to the inlet. This is the same phenomenon that the previous case, except that there is another in front of the top injector. It is always numerical diffusion. In part, these fuel pockets could be the cause of the existence of the large vortex.
The mean fuel mass fraction have been ploted after the last solution (at 22 ms i.e. 500000 iterations) during 30000
iterations.  It shows that air and fuel are sufficiently mixed at outlet, but the mixing is not homogeneous.

In this case, we obtain a better outlet mixing area than the two previous cases. But here, results are opened to criticism. Shaded contours of fuel mass fraction at t=35 ms Cross sections of fuel mass fraction in x=0.1 (middle of the injector) and x=0.21 (outlet of the injector) Shaded contours of U velocity at t=35 ms  