CityDigitalTwin's microclimate solver is tested against published wind-tunnel experiments and full-scale field measurements: benchmark cases from the urban wind datasets of the Architectural Institute of Japan (AIJ), a Tokyo Polytechnic University (TPU) wind-tunnel experiment on air temperature around a building, and a wind-tunnel experiment on evaporation from open water. This page covers the wind, tree-canopy, temperature and humidity cases. Every result is compared point by point, at the exact locations where the data was taken.
The cases start with one building in a wind tunnel and build up to a real city district and trees measured in the field, then add heat and moisture: air temperature around a building in stable and unstable air, and evaporation from open water. Each quantity is normalised as in the original experiment.
The reference test for flow around a single 1:1:2 building: separation at the roof edge, the recirculating wake, and corner acceleration at pedestrian height.
Wind tunnel · 186 points
AIJ Urban Wind Environment benchmark, Case A; Tominaga et al. (2008), J. Wind Eng. Ind. Aerodyn. 96
A high-rise tower standing among city blocks, where a tall building brings strong wind down to the street. Wind speed is compared at pedestrian height.
Wind tunnel · 33 points
AIJ Urban Wind Environment benchmark, Case D
High-rise Shinjuku as it stood in 1978, for a north-north-westerly wind. The model is checked against both a wind-tunnel model and full-scale wind observations recorded near street level.
Wind tunnel · 33 points
Field observations · 13 points
AIJ Urban Wind Environment benchmark, Case F
Wind sheltering behind a 7 m belt of black pines on the Izumo plain in Japan, measured in the field. Trees are modelled as a porous canopy with aerodynamic drag, and the result does not depend on the simulation time step.
R is low here because the measured wind speed barely varies across the 28 points, so correlation says little; FAC2 is the meaningful measure for this case.
Field measurements · 28 points
AIJ Urban Wind Environment benchmark, Case G (field data, Kurotani et al. 2001)
A 1:1:2 building in a wind tunnel whose floor is either colder than the approaching air (stable: 17.7 °C floor, 49.4 °C air) or warmer (unstable: 45.3 °C floor, 11.3 °C air). Air temperature is compared at 48 points on four vertical lines around the building, in each case.
The imposed temperature difference is 32–34 K, so the typical error is about 4–5% of it.
Stable air · 48 points
Unstable air · 48 points
Tokyo Polytechnic University wind-tunnel experiment (Yoshie 2016); measurements digitized from Bazdidi-Tehrani et al. (2018), J. Build. Perform. Simul. 12
A wind tunnel whose whole floor is a 3 m water surface at 16 °C, under air at 20 °C flowing at 3 m/s. Over the water the air gains moisture and cools. Humidity and temperature are compared at three heights along the water, with every input taken from the experiment and nothing fitted.
Deviations are measured on the ratios q/qref and θ/θref, which change by up to about 20% and 10% over the water. For reference, the paper's own CFD model, built for this experiment, averages 1.4% on the same points.
Wind tunnel · 36 points
Kato, Nakane & Yamada (2009) experiment, as reported in Tominaga, Sato & Sadohara (2015), Sustainable Cities and Society 19; measurements and the paper's CFD digitized from its Fig. 5
Each result is scored with the statistics used in the urban wind and microclimate literature, so it can be compared directly with other models.
How closely the simulated pattern follows the measured one across all points. 1 means a perfect match.
Share of points where the simulated wind speed is within a factor of two of the measurement.
Wind speeds are divided by the approach wind at building height, so a scale model and a real city can be compared on equal terms.
(T − Tfloor) / |Tair − Tfloor|: 0 at the floor temperature and ±1 at the approaching air temperature, negative when the floor is the warmer of the two.
The typical size of the difference from the measurement, in physical units: kelvin for temperature.
The average difference from the measurement, as a percentage of the measured value.
Book a demo and we'll walk you through the benchmarks and how CityDigitalTwin performs on projects like yours.
