Importance of preferential segregation by aerodynamics in dust rig tests

TL;DR

This study investigates how aerodynamic effects influence dust particle trajectories and impact velocities in a conveying device, with implications for erosion in gas-turbine engines, comparing LES and RANS simulation accuracy.

Contribution

It demonstrates the significance of preferential segregation by aerodynamics in dust impact behavior and evaluates the predictive capabilities of RANS versus LES simulations.

Findings

Large particles impact with velocities close to exit velocity.

Small particles follow flow streamlines and slow down significantly.

RANS predicts mean impact statistics but underestimates dispersion.

Abstract

This work studies the design of a device conveying dust and sand in order to understand how the particles impinge, erode and rebound from a target plate. The motivation behind this study is understanding dust ingestion and erosion in aviation gas-turbine engines. The conveying system consists of a constant area duct in which particles are injected in the upstream direction using a particle seeder. The particles exit the duct through a converging nozzle and are directed towards a target plate. Particles of varying sizes are tested at two different gas speeds. The particle velocity at the conveyor duct exit follows a trend inversely proportional to diameter. After exiting, particles respond differently to changes in the flow field based on their diameter. The largest particles move ballistically, so they impact the target with nearly the same velocity they had at the duct exit. However, small particles follow the flow streamlines around the target. This causes them to both significantly slow down and to disperse in all directions. The combination of the exit velocity and the near-target trajectory behaviors leads to a non-monotonic trend of particle impact velocity as a function of particle diameter. The importance of these effects depends strongly on the relative angle between the conveyor duct and the target plate. LES are compared to RANS, and it is demonstrated that RANS are capable of accurately predicting mean particle impact statistics. However, RANS results display narrower statistical variation than LES, which suggests that particle dispersion is underpredicted in RANS.