(162d) Large Eddy Simulations (LES) to Capture Turbulence Modulation in Inertial Particle-Laden Turbulent Channel Flows

Particle-laden flows find many applications in nature and industrial processes such as clouds formation, rain drops, fluidized bed reactors, combustion, etc. Due to the presence of particles, capturing the turbulence modulation accurately requires fundamental analysis to understand the dynamics of both the fluid and particle phases in these complex phenomena. The extent of modulation depends on many parameters such as Stokes number, mass loading, Reynolds number, and the ratio of channel width to particle diameter (L/dp) in particle-laden turbulent flows. Direct numerical simulations to predict turbulence modulation are computationally expensive for complex geometries and high Reynolds numbers. It is observed that the turbulent intensity of the gas phase decreases with an increase in particle loading, and at a critical volume loading (CPVL), there is a sudden collapse in turbulence due to disruption in turbulence production for turbulent channel flows [Muramulla et al., JFM, 2020]. In the present work, spectral LES simulations with Smagorinsky and dynamic Smagorinsky sub-grid models are performed to explore their capability in predicting turbulence modulation accurately. The presence of particles leads to turbulence modulation in particle-laden turbulent flows. The turbulence attenuation is observed with Smagorinsky and dynamic Smagorinsky models for two Reynolds numbers of 3300 and 5600 based on average gas velocity and channel width. At low volume fractions, LES models predict the turbulence attenuation with high accuracy (> 80%) but fail to predict the volume loading at which turbulence collapse happens. It is found that an inaccurate prediction of the turbulent energy production is the source of error in CPVL prediction which is caused by the “modeling error” in LES models.