A Sensor-Driven State-Space Framework for PM2.5-Informed Natural Ventilation Opportunities Under Coupled Pollution–Thermal–Humidity Constraints

Abstract

Abstract Natural ventilation can dilute indoor-generated pollutants, but opening windows can also increase coupling to outdoor fine particulate matter (PM 2.5 ). We developed a sensor-driven screening framework that combines sustained outdoor PM 2.5 with a pragmatic temperature–relative-humidity (T–RH) envelope to identify candidate natural-ventilation opportunity hours. Using harmonised AirGradient observations accessed through OpenAQ, we compared four residential communities—KL_TTDI (Kuala Lumpur, Malaysia), ID_BogorSelatan (Bogor, Indonesia), SG_Midwood (Singapore), and AU_Palmerston (Gray/Palmerston, Australia)—from 1 January to 1 February 2026 (local time), with 724–744 hourly bins per site. Sustained PM 2.5 was represented by a trailing 24-h mean and evaluated at $$\theta_{\text{pm}} = 15\;\upmu {\text{g}}\;{\text{m}}^{ - 3}$$ , matching the 2021 World Health Organization 24-h air-quality guideline value. The operational meteorological screen was $$24 \le T \le 30\;^\circ {\text{C}}$$ and $$RH \le 65\%$$ . Primary clean-hour fractions were 14.5% in KL_TTDI, 34.3% in ID_BogorSelatan, 30.8% in SG_Midwood, and 97.4% in AU_Palmerston, whereas standalone T–RH feasibility was 97.0%, 71.5%, 98.5%, and 20.6%, respectively. The resulting monthly regimes were PM 2.5 -constrained in KL_TTDI and SG_Midwood, mixed in ID_BogorSelatan, and predominantly clean but T–RH-limited in AU_Palmerston. The framework identifies outdoor screening opportunities; paired indoor–outdoor monitoring is needed before translation to occupant guidance or automated control.