Predictive modelling of pedestrian‑level outdoor ventilation in urban residential areas with varied morphological parameters at comparable building densities

• Morphology drivers of ventilation identified from 107 real residential blocks. • CFD cases reveal airflow patterns with varied morphologies at comparable densities. • A building-physics index λ BV that couples geometry and spacing effects was proposed. • Aligned-layout empirical models reach R²>0.80; enclosed ones achieve acceptable fits. • Validated on irregular blocks, the proposed models accurately predict outdoor MVR. Effective outdoor ventilation is crucial for pedestrian thermal comfort by enhancing heat dissipation. While many previous studies examine ventilation across a wide range of building densities, they overlook the reality that regulations tightly constrain density, leaving only slight variation. Moreover, density alone is a limited diagnostic predictor of ventilation performance. Assessments of ventilation under comparable(near‑uniform) building densities, but with varying morphological parameters within urban residential areas, remain scarce. To address this, this study proposes a framework for pedestrian-level ventilation in such contexts by coupling a large ensemble of computational fluid dynamics (CFD) simulations with morphology-based empirical models. We identified representative morphological parameters from 107 residential areas in Guangzhou, a hot–humid city, to reflect real-world configurations, including building length ( L ), height ( H ), spacing ( W ), and layout types (aligned, staggered, enclosed, and semi-enclosed). CFD results show that increasing L and H enhances the mean velocity ratio (MVR), whereas larger W and staggered layouts reduce it; semi-enclosed forms do not outperform enclosed layouts. Guided by building-physics reasoning, we proposed empirical models that predict pedestrian-level outdoor MVR through a new composite index (λ BV ) that integrates the key morphological parameters identified above. For aligned layouts under perpendicular and southeast approach flows, the models achieve strong fits ( R ² = 0.80 and 0.81, respectively). For the more complex enclosed layouts, the corresponding R ² values are 0.60 and 0.76—acceptable given the added flow complexity. All models were validated against irregular configurations within the studied morphological parameter. The study provides a dependable tool for estimating pedestrian-level outdoor ventilation in aligned and enclosed residential areas with varied morphological parameters under comparable densities in hot–humid climates, while applications beyond the studied climatic, stability, or morphological ranges require further recalibration and validation.