It has been long established that Reynolds number effects can lead to flow instabilities and/or transition from laminar to turbulent flow regimes. The nature of free shear jets is well understood and heavily covered in the fluid mechanics literature. On the other hand, the study of confined nozzles presents some challenges and is still a developing area of research. In this work, we focus on quasi-impinging jets, such as the ones feeding into a virtual impactor. Virtual impactors are popular, inexpensive aerosol collection devices capable of separating airborne solid particles. Recently they found increased application in areas that require concentration of dilute aerosols, such as biological-laden flows. In essence, this research is motivated by the need to fundamentally understand the fluid-particle interaction mechanisms entailed during virtual impaction. To this end, we rely on theoretical insight gained by numerical analysis of the classical equations within a one-way coupled Lagrangian framework.

In the first part of this investigation we perform a direct transient simulation of the two-dimensional incompressible Navier-Stokes equations for air as the carrier phase. The momentum and continuity equations are solved by FLUENT. The solutions of three separate computations with jet Reynolds numbers equal to 350, 2100, and 3500 are analyzed. The 2-D time-mean results established the nature of the jet potential core and clarifications about the role of the Reynolds number were proposed. Transient analysis deciphered the characteristics of the mirrored Kelvin-Helmholtz instability, along with particle-eddy interaction mechanisms.

In the second part we perform a large eddy simulation (LES) on a domain of a real-life sampler. The Lagrangian dynamic residual stress model is implemented and validated for two canonical turbulent flows. The newly contrived code is then applied to the study of a prototype device. A three-dimensional growth mechanism is proposed for the jet mixing layers. The Lagrangian dynamic model LES exhibited significant regions of high subgrid turbulent viscosity, compared to the dynamic Lilly-model simulation, and we were able to identify the origin, and learn the dynamics of five key coherent structures dominant during transition. Comparison with preliminary experimental data for the aerosol separation efficiency showed fairly good agreement.