Urban densification increases aerodynamic resistance to incoming rural winds and suppresses channeling flow within street networks. The resulting weakening of advective ventilation raises concerns about pollutant removal across the roof level between the urban canopy layer and the overlying atmosphere. This study investigates how urban morphology influences total aerodynamic resistance and the mechanisms governing roof-level air exchange through windtunnel experiments and computational fluid dynamics (CFD) simulations. Idealized models were adopted to enable generalizable insights.

An idealized urban neighborhood with discretely arranged building blocks was first examined. Based on total drag force (Fd), the highest aerodynamic resistance occurred at moderate plan area density (λp ~0.25), while spanwise repeated arrays exhibited decreasing Fd with increasing λp over 0.11–0.56. At fixed λp, layouts, allowing more direct flow impingement on windward façades produced larger Fd. Poorly ventilated hotspots were mainly associated with recirculation in building near wakes, typical of wake-interference and isolated roughness regimes. Increased geometric contrast between adjacent buildings enhanced vertical transport across the canopy–roof interface, although local ventilation remained spatially heterogeneous.

A worst-case scenario without channeling flow was analyzed using quasi two-dimensional street canyons and isolated semi-enclosed cavities, with emphasis on fluctuating components of ventilation. Under skimming flow, air exchange was dominated by a mixing layer across the opening, yielding nondimensional ventilation rates (Q*) of 0.02–0.03. Vertically stacked multi-vortex structures inhibited internal mixing and significantly reduced ventilation efficiency, whereas geometry-induced flow perturbations generally enhanced roof-level exchange.

Three rooftop flow regimes were identified: recirculation zone, flow attachment, and conical vortex. When recirculation dominates, ventilation resembles shear-driven lateral exchange. Whether direct mean-flow advection occurs depends on the opening’s position relative to the flow reattachment point. Under oblique inflow, conical vortices over a rectangular prism markedly increased ventilation through strong suction and elevated turbulence. Turbulence intensity in the approaching flow further modified separation behavior and ventilation efficiency.

Overall, the study demonstrates that building-induced turbulence plays a central role in roof-level ventilation and that idealized models provide generalizable insights into how urban morphology governs air exchange in dense urban environments.

Urban förtätning ökar det aerodynamiska motståndet mot inkommande vindar och dämpar kanaliserat flöde i gatunätet. Den försvagade advektiva ventilationen försvagar borttransport av föroreningar vid taknivå, mellan den urbana canopy-nivån och den överliggande atmosfären. Denna studie undersöker hur urban utformning påverkar det totala aerodynamiska motståndet och de mekanismer som styr luftutbytet vid taknivå genom vindtunnelexperiment och simuleringar med Computational Fluid Dynamics (CFD). Idealiserade modeller användes för att möjliggöra generaliserbara insikter.

Ett idealiserat urbant kvarter med diskret arrangerade byggnadsblock analyserades först. Baserat på den totala dragkraften (Fd) uppnåddes högst aerodynamiskt motstånd vid en måttlig bebyggelsegrad (λp ~0.25), medan spannvisu pprepade byggnadsmatriser uppvisade avtagande Fd med ökande λp inom intervallet 0,11–0,56. Vid fast λp gav utformningar som tillät mer direkt flödesimpingementmot lovartsfasader större Fd. Otillräckligt ventilerade zoner var främst kopplade till recirkulation i byggnaders närliggande läområden, typiska för wake-interference- och isolated-roughness-regimer. Ökad geometrisk kontrast mellan intilliggande byggnader förstärkte den vertikala transporten över gränsytan vid taknivå mellan bebyggelsen och den överliggande atmosfären, även om ventilationen var rumsligt heterogen.

Ett värstafallsscenario utan kanaliserat flöde analyserades med kvasi-tvådimensionella gatukanjoner och isolerade semi-inneslutna kaviteter, med fokuspå de fluktuerande komponenternas roll i ventilationen. Under skimming flow dominerades luftutbytet av ett blandningsskikt över öppningen, med dimensionslösa ventilationsgrader (Q*) på 0,02–0,03. Vertikalt staplade multivirvelstrukturer hämmade den interna blandningen och reducerade ventilationseffektiviteten avsevärt, medan geometriskt inducerade flödesstörningar generellt förbättrade luftutbytet vid taknivå.

Tre dominerande takflödesregimer identifierades: recirkulationszon, flödesanliggning och konisk virvel. När recirkulation dominerar liknar ventilationen skjuvinducerad lateral luftväxling. Om direkt advektion med medelflödet uppstår beror på öppningens position i förhållande till återanliggningspunkten.Vid sned inflödesriktning ökade koniska virvlar över en rektangulär prismamodell ventilationen markant genom starkt sug och förhöjd turbulens. Turbulensnivån i det inkommande flödet påverkade dessutom flödesseparationen och därmed ventilationseffektiviteten.

Sammanfattningsvis visar studien att byggnadsinducerad turbulens spelar en central roll för ventilation vid taknivå och att idealiserade modeller kan ge generaliserbara insikter om hur urban morfologi styr luftutbytet i täta urbana miljöer.

Previous studies have rarely addressed single-sided ventilation driven by the external flow over the roof, which exhibits considerable potential owing to its highly turbulent nature and strong suction associated with leading-edge flow separation. In this study, wind tunnel experiments on single-sided ventilation through a roof opening were conducted using two isolated generic models: a cylinder and a rectangular prism, each with a set of replaceable openings. Both models were tested either flush- or floor-mounted. Two inflow conditions, each with three free-stream velocities, were considered. For both models mounted beneath the floor, the nondimensional ventilation rates (Q*) are comparable to values reported in the literature; for the prism, a slight increase in Q* with orientation suggests the development of a mixing layer along the streamwise extent of the floor-level opening. In the floor-mounted configuration, body-induced flow disturbances tend to enhance ventilation. Three primary governing rooftop flow regimes are identified—recirculation zone, flow reattachment, and conical vortex—whose relative dominance over the opening depends on inflow turbulence, wind incidence angle, and model configuration. When the opening lies entirely within the recirculation zone, Q* is proportional to the normalized local fluctuation intensity, with a coefficient of about 0.16. For certain yaw angles, the marked increase in Q* strongly correlates with the presence of a conical vortex over the prism model roof, which features strong suction and intense fluctuations. Direct advection through the opening could occur with a favorable opening size and location, allowing deep penetration of the reattaching shear layer.

As a preliminary investigation of the wind-driven purging process for densely built environments through the canopy layer, the ventilation efficiency of standalone semi-enclosed models incorporating a horizontal opening in the roof façade was investigated in the wind tunnel. For comparison, two models with different geometries were constructed, and each model was tested individually. Both models were equipped with replaceable roof covers, enabling the adjustment to the opening size. The ventilation efficiency was evaluated by continuous releasing and sampling of the tracer gas, from which the normalized purging velocity (PFVn) was derived. Additionally, the flow condition over the opening was monitored using the Laser Doppler Anemometer. It was found that separation flows from the frontal edge(s) of the model could introduce secondary circulations across large openings, resulting in dramatic increases in PFVn. Both the rectangular prism model and cylinder model possessed higher PFVn compared to prior studies on single-sided ventilation, while close values were observed with cylinder model mounted under the wind tunnel floor. Besides, the vertical distribution of integral length scales of streamwise velocity indicated the stratification feature of separation flows under low-turbulent incoming flow conditions. Measurement results provide validation data for further simulation studies including more detailed structures.

This study adapted the mean age of air, a time scale widely utilized in evaluating indoor ventilation, to assess the impact of building layouts on urban ventilation capacity. To distinguish it from its applications in enclosed indoor environments, the adapted index was termed the effective mean age of air (T_E). Based on an experimentally validated method, computational fluid dynamic (CFD) simulations were performed for parametric studies on four generic parameters that describe urban morphologies, including building height, building density, and variations in the heights or frontal areas of adjacent buildings. At the breathing level (z = 1.7 m), the results indicated three distinct distribution patterns of insufficiently ventilated areas: within recirculation zones behind buildings, in the downstream sections of the main road, or within recirculation zones near lateral facades. The spatial heterogeneity of ventilation capacity was emphasized through the statistical distributions of T_E. In most cases, convective transport dominates the purging process for the whole canopy zone, while turbulent transport prevails for the pedestrian zone. Additionally, comparisons with a reference case simulating an open area highlighted the dual effects of buildings on urban ventilation, notably through the enhanced dilution promoted by the helical flows between buildings. This study also serves as a preliminary CFD practice utilizing T_E with the homogenous emission method, and demonstrates its capability for assessing urban ventilation potential in urban planning.

A tracer gas measurement method was developed to investigate the single-side removal process of pollutants through the roof level, conducted in a wind tunnel utilizing a deep box model with various openings at its top. A homogeneous emission was modelled by filling the box with well-mixed nitrous oxide released from a line source inside. The removal efficiency from the openings with box at different orientations was evaluated by normalized purging velocity (PV) and spatial-mean age of air (). The PIV system was adopted to visualize the flow field above the opening. Varying interactions between the perturbations from the openings and the deflected flow over them resulted in diverse PVs at different orientations. There are also clues showing that the Reynolds-number independent conditions for cavity flows, such as inside a street canyon, are also related to the scale of the open area at the roof height.

Here, it is presented a possible methodology and experimental model for benchmarking of air quality in cities. The concept behind the methodology is that a city’s inherent structure affects the potential for contaminant removal due to the resistance it poses to inflow. The approach is based on homogenous emission across the street surface network, representing a worst-case situation. Different levels of complexity can be used for benchmarking, making it valuable for evaluating different layouts. Additionally, an urban ventilation index suitable for these kinds of experimental studies has been suggested. 

The drag force generated by aligned arrays of cubes of different packing density and exposed to different wind directions in a wind tunnel is discussed. Results allowed to build a drag force rose which shows that the drag force increases with increasing packing density till λ_p = 0.25 for any wind direction. It is also shown that, independent of the packing density, the drag force increases with increases deviation of WD from the perpendicularity.

The Reynolds number independence (Re independence) criterion of Re > 11,000 is widely adopted to fulfill the dynamic similarity between the urban flow modeling and the down-scale measurements. However, for 2D street canyons with H/W ≥ 1.5, experiments, numerical simulations, and in-situ observations have reported different vortex-flow regimes with similar building configurations but at different scales. This study uses both wind tunnel experiments and numerical simulations to revisit the Re-independent flow regimes and Re independence criteria with an extensive Re range for idealized 2D street canyons with various aspect ratios (H/W = 1.1, 2.4, 3, 4, and 5). We introduced an optimized ratio of relative changes (RRCs) to evaluate the flow regimes’ similarity. The wind tunnel experiment confirms that the cavity flow with H/W = 1.1 meets the Re independence when reference building Re (Reref) exceeds 11,000. Simulations validated by the experiment results are conducted to investigate detailed flow regimes and the critical Re (Rec) range for each aspect ratio. The canyons with H/W = 2.4, 3, and 4 are dominated by a single asymmetric vortex when the Re independence is satisfied, while there are two vertically-stacked counter-rotating vortices in the canyon with H/W = 5. The value range of Rec increases with aspect ratio from 1.9 × 104–2.6 × 104 (H/W = 2.4) to 1.3 × 105–2.1 × 105 (H/W = 3), and 2.1 × 106–6.4 × 106 (H/W = 4 and 5). Our results indicate that the fully Re-independent flow regimes in deep canyons have fewer vortices than the literature value with down-scale experiments and simulations. The variant Rec with different aspect ratios suggests the requirement to conduct the Re-independence test for different model configurations.

Validated by wind tunnel data, this study numerically investigates the integrated impacts of wind and thermal buoyancy on urban turbulence, ventilation and pollutant dispersion in full-scale 2D deep street canyons (aspect ratio AR = H/W = 3 and 5, W = 24 m). Isothermal urban airflows for such deep street canyons can be Reynolds-number-independent when reference Reynolds number (Re) exceeds the critical Re (Rec~106,107 when AR = 3, 5), i.e. AR = 5 experiences two main vortices and one-order smaller NEV* (~10−3, the normalized net escape velocity) than AR = 3 with a single main vortex (NEV*~10−2). With sufficiently large Re (Re > Rec) and the same air-wall temperature difference (Ri = 2.62, 4.36 when AR = 3, 5), four uniform wall heating patterns were considered, including leeward-wall heating (L-H), windward-wall heating (W–H), ground heating (G-H), and all-wall heating (A-H). Various indicators were adopted to evaluate street ventilation and pollutant dilution capacity (e.g. age of air (τ,s), NEV*, pollutant transport rates (PTR)). Full-scale wall heating produces a strong upward near-wall buoyancy force, which significantly influences flow patterns and improves street ventilation for most cases. When AR = 3, L-H strengthens the single-vortex airflow. When AR = 5, L-H converts the isothermal double vortices into a single-clockwise vortex. For both ARs, W–H reverses the main clockwise vortex to an enhanced counterclockwise one, moreover G-H and A-H cause a more complicated multi-vortex pattern than isothermal cases. Overall, when AR = 3, L-H and W–H increase NEV* by 68% and 40% than the isothermal case. When AR = 5, four wall heating patterns all raise NEV* considerably (by 150%–556%). For both ARs, the L-H, W–H and A-H amplify the contribution of mean flows on removing pollutants but reduce that by turbulent diffusion compared with isothermal cases.