Mixing phenomena with shear-thinning fluid is frequently studied in industry. However since Lumley 1969 and DeGennes et al. 1986, few progress was done in terms of phenomenology of shear thinning turbulent mixing. In order to have a better understanding of properties of flow combining turbulence and non-Newtonian properties a very interesting configuration is the T-junction flow. Two flow entries generates one main shear layer when the flows from the two separate branches join. This leads to a zone in the flow where turbulence is generated far from the wall. This could help to clear what is the most important concerning interaction between polymer and turbulence phenomena, wall generated turbulence or bulk-flow turbulence. Furthermore such a flow generate an important recirculation zone upstream the junction that could lead to complexify the mixing of species transported by the two different branches. Such Recirculation zone lead to low mixing that have to be understand and compare with the main shear layer mixing zone.

 In so in this numerical work we have focussed on the T-junction flow study with different regimes corresponding to those imposed by the experiment of Nguyen et al. 2012 from 2400 to 8000. This experiment examined the behavior and the evolution of the velocity and concentration fields in the case of the mixing of one passive scalar specy transported by the perpendicular branch of the T flow. Water flow and Non Newtonian flow were studied.

We will show the results obtained from the resolution by DNS (direct Numerical simulation) of the Navier-Stakes equation coupled with variable viscosity following the Carreau-Yasuda law fitting experiment data on Xanthan gum which is known as a shear-thinning fluid solution. The cross section is square shaped. Our calculations are 3D ones using finite volume method via open source code OpenFOAM.

We used 20 million of meshes. Mesh spacing constraint is applied to wall boundary to ensure y+ ~ 0.5 at first wall cell. The following Nominal Reynolds numbers were imposed : 4800, 8000 for water; 2400 and 4000 for Xanthan gum solution. Temperature as a passive scalar was imposed on the inlet of the vertical branch.

For our Newtonian simulations, the flow is always turbulent, for all our non-Newtonian simulations, the flow before the junction has a very viscous core (about 100 times more viscous than water). Nonetheless, the flow can develop instabilities or turbulence downstream the junction : it depends on certain parameters

We will show that re-laminarisation is observed for Xanthan gum solution instead of a self-sustainable turbulence for the small Reynolds case (Re~2400). Whereas with pulsative conditions (St = 1, 5) for both entries and the flow becomes inherently unstable.

For higher Reynolds number (Re~4000), the non-newtonian flow is inherently turbulent and the turbulence peak is shifted downstream comparing to Newtonian situation.

Additionally, by varying momentum ratio between the main branch and the vertical branch and conserving Reynolds number at exit, we obtain two characteristic jet flows classified from Igarashi et al. 2002 as “deflecting jet” and “impinging jet”. We will show their flow structures as well as their impact on passive scalar mixing efficiency.

We demonstrate that in the non-newtonian case, the mixing efficiency at outlet is enhenced in the “impinging jet” comparing to “deflecting jet”.

This work was granted access to the HPC resources of CINES under the allocation A0042A10464 made by GENCI.