Adaptive filtering AHU: A novel approach for balancing the efficient filtering performance and energy conservation

The efficient retention of PM2.5 represents a key measure for enhancing indoor air quality. However, the dynamic nature of PM2.5 concentrations complicates the synergistic optimization of filtration efficiency and operational energy consumption. Current research primarily concentrates on two approaches: (1) developing intelligent dynamic control systems, and (2) enhancing filter material performance. Nevertheless, owing to limitations in system stability and adaptability, these methods exhibit constrained effectiveness in practical engineering applications. Based on these considerations, this study proposes an adaptive air filtration system incorporating dynamic dual air duct regulation. The proposed system employs a parallel architecture comprising a high-efficiency filtration duct and a bypass duct. Through dynamic adjustment of the inter-duct air flow ratio in response to real-time PM2.5 concentration, this configuration enables on-demand filtration while minimizing energy losses attributable to filtration resistance. Testing reveals that modulating the bypass ratios (20–80 %) achieve a 30 % reduction in fan power consumption. During heavy pollution episodes, the system demonstrates: (1) a 90.1 % improvement in PM2.5 standard attainment duration versus two-stage filtration, and (2) 26 % energy reduction compared to three-stage systems. Computational analyses across varying spatiotemporal conditions reveals that the adaptive system demonstrates superior performance under dynamic high-pollution scenarios, where its responsive control mechanism optimizes energy efficiency. These findings not only empirically validate the technical superiority of the dual air duct dynamic regulation system in resolving the inherent trade-off between filtration efficiency and energy conservation, but also present an innovative approach for advancing intelligent, health-conscious, and low-carbon indoor environmental control systems.