Confined Impinging Jets Reactors (CIJRs) are interesting devices because of their ability to achieve high mixing efficiencies. Consequently they can be applied for the production of nano-particles through processes such as standard precipitation and solvent displacement in which the rate of mixing at all scales determines the Particle Size Distribution (PSD) and therefore the final quality product. In a precipitation process (e.g., precipitation of barium sulphate) small and mono-dispersed particles are formed when the super-saturation level is high and homogeneous, a condition achieved by rapid mixing between reactants. A similar situation is observed in solvent displacement processes, where a solution of a polymer dissolved in an organic solvent (e.g., poly-capro-lactone in acetone) is rapidly mixed with a non solvent (e.g., water). Since acetone is miscible with water, while the polymer is not, the solubility of the polymer in the solution changes, super-saturation is built up, and polymer nano-particles are formed. The possibility of predicting the PSD by means of a detailed precipitation model based on Computational Fluid Dynamics (CFD) is appealing because it would allow design, optimization and scale up of precipitation reactors with notable reduction of experimental activities. For the precipitation model to be effective, however, it is of fundamental importance the accurate description of the velocity and turbulence fields since they define mixing at the macro-, meso- and micro-scale levels. In several works (Liu and Fox, 2006, Gavi et al., 2007) an Interchange by Exchange with the Mean (IEM) approach coupled with a presumed Probability Density Function (PDF) method (i.e., the Direct Quadrature Method of Moments, DQMOM) was implemented in a CFD code to predict the interaction between mixing and fast chemical reactions. Modelling results and comparison with experimental data concerning the reaction course show that the choice of the turbulence model and the near wall treatment have a great effect on the final model predictions, and therefore an independent validation of flow and turbulent field predictions is needed. This validation can be carried out by resorting to Particle Image Velocimetry (PIV) a very powerful technique that has been used to investigate the flow field in a variety of reactor geometries: stirred vessels (Sheng et al., 2000), rectangular micro-channels (Li et al., 2004), planar jet reactors (Feng et al., 2005) and T-mixers (Hoffman et al., 2006). In this work micro Particle Image Velocimetry (microPIV) is employed to investigate the flow field in a CIJR. The experimental data are then used to validate the velocity field predicted by Large Eddy Simulation (LES) in terms of mean velocity, turbulent kinetic energy and turbulent dissipation rate. PIV allows to measure the instantaneous velocity field in a plane. Two images are captured for each realization and a cross-correlation technique is used to compute the velocity flow field.. Since every velocity vector measured by PIV is a result of spatial average of the velocity field in one interrogation cell, the resolution of PIV has to be high enough so that the point velocity can be approximated with the spatially averaged velocity obtained with PIV. Much more complicated is the problem of estimating the energy dissipation rate from PIV measurements. By resolving the flow field to the Kolmogorov microscale, it would be possible to directly compute the energy dissipation rate from the strain rate. Since this is not possible, it is raccomended that the spatial resolution of PIV is no more than five times the Kolmogorov microscale. Then the turbulent energy dissipation rate can be calculated from the fluctuating velocity, the integral length scale and a constant of order of unity (Feng et al., 2005). This approach can be applied to a turbulence field that is homogeneous, isotropic and in spectral equilibrium. Moreover, the fact that the integral length scale varies in different regions of the reactor has to be taken into account. A more rigorous approach to the dissipation rate computation is the large eddy approach, which assumes a dynamic equilibrium of the sub-integral scales. The turbulence dissipation rate is then equal to the flux of turbulent kinetic energy from large to small scales through the inertial sub-range. Consequently this method does not require to solve the velocity field down to the Kolmogorov scale, because the flux of turbulent kinetic energy is estimated at length scales within the inertial sub-range (Sheng et al., 2000). The micro PIV experimental setup is primarily composed by the double pulsed Nd:YAG laser (New Wave Research Gemini PIV laser), the inverted microscope (Nikon T-300 Inverted Microscope), and the camera (LaVision Flowmaster 3 camera). The flow was seeded with fluorescent polystyrene microspheres with a nominal diameter of 2 mm and a density equal to 1.05 g/ml (FluoSpheres, Invitrogen, Carlsbad, CA). The concentration of particles C was computed in order to achieve the desired number of particles ( 5-10 particles) in each interrogation volume (Li, 2005). A multi-pass interrogation scheme was used in the computation of the vector field and since a 50% overlapping was used the final interrogation area is pixels, leading to a spatial resolution of 0.1 mm. An acqueous sodium iodide solution was used to match the index of refraction of the reactor. The geometry investigated in this work is characterized by conical head and outlet, the inlet jet diameter is mm and the reactor chamber is mm. Inlet flow rates are kept between 24 and 118 ml/min, resulting in the following range of averaged inlet velocities: 0.5 – 2.5 m/s, corresponding to Reynolds numbers in the inlet jets ranging between 900 and 4500. As already reported experimental data are then compared with CFD predictions obtained with LES. This approach is based on the idea of solving directly the bigger scales of motions of the flow, whereas smaller flow structures are modelled. In fact, large eddies contain most part of the flow energy, while the smaller scales are geometry and mean flow independent and therefore easier to model. A mathematical filter is applied to the Navier-Stokes equations under the hypothesis of incompressible flows, and the resulting sub-grid scale stress tensor is then modelled with a sub-grid scale model. In Fig. 1 the instantaneous flow field at several time steps is reported for an inlet velocity of 1 m/s and a corresponding Reynolds number in the inlet jets of 1000, calculated with a grid size of 400,000 cells and by resorting to the Smagorinsky-Lilly model and sub-grid-scale model. As showed by the experimental data, the two jets impinge in the centre of the chamber, where most of their kinetic energy is transformed into turbulent kinetic energy, which is quickly dissipated. This region (of high and confined turbulent dissipation rate) is where most of the mixing occurs. It is also interesting to notice that due to the confinement of the jets there is no possibility of by-pass, and the inlet fluids must flow through this region. Moreover, the impinging plane being slightly unstable, results in oscillations of the jets in all possible directions and in circular periodical movements of the impinging plane, guaranteeing also very good macro-scale mixing and no back mixing effects.

Feng, H., Olsen, M.G., Liu, Y., Fox, R.O., Hill, J.C. (2005). Investigation of turbulent mixing in a confined planar-jet reactor. AIChE Journal, 51, 2649 - 2664 Gavi, E., Marchisio, D.L., Barresi, A.A. (2007). CFD modelling and scale up of Confined Impinging Jet Reactors. Chemical Engineering Science, 62, 2228-2241 Li, H., Ewoldt, R., Olsen, M.G., Turbulent and transitional velocity measurements in a rectangular microchannel using microscopic particle image velocimetry, Experimental Thermal and Fluid Science, 29, 435 - 446 Liu, Y., Fox, R.O. (2006). CFD predictions for chemical processing in a Confined Impinging jets Reactor. AIChE Journal, 52, 731 - 744 Sheng, J., Meng, H., Fox, R.O. (2000). A large eddy PIV method for turbulence dissipation rate estimation. Chemical Engineering Science, 55, 4423 – 4434

Extended Abstract Status: Not Uploaded