Development and Validation of Advanced Physical Models for Urban Wind Microclimate Assessment

ABSTRACT The quality of the pedestrian outdoor environment is increasingly recognized as a public health and planning priority. Wind-induced discomfort, outdoor thermal stress, and urban heat island effects are the subject of a growing body of regulatory guidance that requires developers and urban planners to demonstrate compliance with quantitative comfort criteria at pedestrian level. Meeting these requirements within the timescales of architectural practice demands simulation tools that are physically rigorous, geometrically faithful, and computationally efficient. This thesis presents a series of validated contributions to ArchiWind, a CFD-based platform for pedestrian wind comfort assessment built on the open-source solver OpenFOAM and developed at NablaFlow. The work addresses three main areas. First, the platform is validated against the Architectural Institute of Japan (AIJ) Benchmark Case E — a canonical dense urban environment in the Niigata city centre — demonstrating good agreement with wind tunnel measurements across 80 pedestrian-level probe points and 16 inflow directions (RMSE = 0.174, Pearson r = 0.769 for the after-construction scenario using the Realizable k-ε turbulence model on the Detailed mesh). The platform correctly resolves the spatial pattern of the construction effect ΔU/U_ref, confirming its suitability for before/after planning impact assessments. Sensitivity studies show that the Draft mesh quality setting yields essentially the same global accuracy as the Detailed setting (ΔRMSE = −0.002), and that the k-ω SST turbulence model performs comparably to Realizable k-ε. Second, two methodological enhancements to the CFD workflow are introduced. A physics-based porous media model for tree canopy representation replaces the previous static default height assignment with a Voxel-Based Canopy Profiling (VCP) algorithm that extracts canopy height directly from the input STL geometry and introduces a geometry-scaled numerical stability limiter (K_stability), eliminating the floating-point exceptions and divergence episodes previously associated with high-density vegetation zones. A two-stage physics-driven convergence criterion is developed and validated across eight real-world test cases, replacing traditional global residual monitoring with a sequential algorithm that monitors the stagnation of pedestrian-level physical fields at 1.5 m AGL; runtime reductions of 27–62 % are achieved with comfort map differences below 1.76 % and probe-level errors below 0.1 %. Third, the physical scope of the platform is extended. Thermal buoyancy is implemented via the Boussinesq approximation through a transition from OpenFOAM's simpleFoam to buoyantBoussinesqSimpleFoam; solver equivalence with the isothermal baseline is confirmed with RMSE = 0.00024 m s⁻¹ and Pearson r = 1.000, and active buoyancy produces physically consistent tilted plumes, surface-driven updraughts, and horseshoe vortex patterns. A complete outdoor thermal comfort assessment module is developed, computing Mean Radiant Temperature (MRT) and the Universal Thermal Climate Index (UTCI) from first principles using the CFD wind atlas, hourly Meteoblue climate records, and three-dimensional ray-casting for solar shading and sky view factor computation. Applied to a representative London urban site, the model produces seasonal compliance maps in accordance with the five-tier GLA London Plan classification, with mean summer MRT values of 18–25 °C and winter values of 2–5.5 °C. Taken together, the contributions advance ArchiWind from a wind-only assessment tool to a fully integrated urban microclimate platform capable of meeting current and emerging regulatory requirements for pedestrian comfort analysis in architectural and urban planning practice.