Experimentally validated CFD–URANS study of portable air purifier performance in a classroom

Abstract

This study presents a validated Computational Fluid Dynamics (CFD) investigation of PM2.5 removal in a realistically modelled school classroom equipped with air purifiers. Using an unsteady Reynolds–averaged Navier–Stokes (URANS) approach with the elliptic-relaxation ζ–f turbulence model, the simulations reproduce the main flow features and pollutant decay well, showing strong agreement with field measurements (individual R2 from 0.8805 to 0.9975; average R2=0.9768). The complex realistic classroom geometry is faithfully captured via a robust porosity approach. This method represents furniture as high-resistance zones, ensuring geometric fidelity while avoiding computationally expensive mesh refinement. The real-world air purifier performance is evaluated against nominal specifications, revealing significant differences: measured Clean Air Delivery Rates (CADR) ranged from 213 to 289m3h−1 for a nominal 360m3h−1 setting, while the 600m3h−1 maximum setting yielded an actual CADR of 373.9m3h−1. Investigations into the number, flow rates, and positioning of purifiers demonstrate that spatial distribution is as important as total flow capacity. Results indicate that distributed designs achieve approximately 13% faster particle removal compared to clustered configurations with the same total flow rate, whereas centrally positioned units outperform corner placements by roughly 15%. Overall, multiple spatially distributed units significantly enhance room-scale mixing, reducing PM2.5 from hazardous levels (250μgm−3) to safe thresholds (<10μgm−3) in times ranging from 80 minutes with three purifiers at medium flow to 4.7 hours with a single unit. Simulation results were compared to measurements obtained with a calibrated system of sensors.