(223e) Hybrid DNS-LES of Nanoparticle Nucleation: The Effects of Small-Scale Fluctuations On Metal Nucleation

Vapor-phase particle formation occurs in a wide variety of processes of engineering and scientific interest. In a number of these processes, gas-to-particle conversion via homogeneous nucleation is a dominant mechanism. Additionally, a number of these processes typically occur under turbulent flow conditions. In industrial processes, rapid mixing is often used to increase the saturation which drives the nucleation process. It is well known that homogeneous nucleation has a highly non-linear dependence on both temperature and concentration. As a result, turbulent fluctuations in temperature or concentration can result in orders of magnitude changes in the local nucleation rate. 

Methods used for simulation of fluids containing nanoparticles fall under three general categories: Direct numerical simulation (DNS), Reynolds-averaged Navier-Stokes simulations (RANS) and large-eddy simulation (LES). DNS solves explicitly for all scales of motion but is computationally prohibitive for all but the simplest flows. RANS simulations are the most computationally affordable and as a result have been used for analysis of many engineering flows. LES has the advantage of DNS in that it captures the unsteady evolution of large-scale flow features while using models similar to those found in RANS to simulate the small, more uniform, scales of motion. Both RANS and LES requires models for all non-linear interactions. These "turbulence models" have been widely developed and utilized for turbulent reacting flows. However, the effects of the small-scale, unresolved, fluctuations on homogeneous nucleation aren't entirely clear.

In this work we perform simulations of zinc nanoparticle nucleation in turbulent jets. The flows consist of a high-speed jet of zinc metal vapor issuing into a cooler stream. As the two streams mix, the metal vapor becomes super-saturated and nanoparticles form (and grow via condensation and coagulation). We utilize a Navier-Stokes + nodal general dynamic equation solver to perform a variety of DNS and hybrid DNS-LES studies to isolate and characterize the effects of mixing on homogeneous nucleation. In the DNS all length and time-scales are resolved. This serves as a benchmark simulation. In the hybrid DNS-LES approach, a combination of fully-resolved variables, and large-scale variables are used to obtain the particle field. More precisely, when computing the nucleation rate, and critical diameter, the temperature, pressure, and vapor concentrations are filtered to remove the small-scale features. This approach allows us identify the effects of the small, unresolved scales on nucleation, and illustrates the effects of insufficient resolution in the modeling of particle formation processes.