Turbulent Fluctuations Do Most of the Work in Pumping Ventilation Pollutant Removal

Velocity fluctuations, not the steady mean flow, do almost all of the work of flushing pollutants from rooms ventilated by pumping ventilation, according to a new large-eddy simulation study of an isolated three-story building. The researchers found that the unsteady mass flux through the openings accounted for more than 90% of the total pollutant transport, a quantitative confirmation that turbulence dominates this ventilation mode. The work, published in Architectural Intelligence, also validated the simulation approach against earlier wind tunnel measurements, with velocity statistics predicted to within a mean deviation of 5%.

Pumping ventilation is a special case of single-sided ventilation, in which a room has two openings on the same leeward facade rather than on opposite walls. It was first identified in wind tunnel experiments by Daish and colleagues in 2016, and it is driven by periodic vortex shedding in the building’s wake. As vortices peel off the leeward face, they create oscillating pressure differences between the two openings, pumping air in and out of the room far more effectively than ordinary single-sided ventilation at the same total opening area. This has made pumping ventilation an attractive candidate for improving air exchange in modern urban buildings, where deep floor plans and privacy concerns often rule out cross ventilation.

Although previous studies had established that unsteady pressure fluctuations dominate pumping ventilation, few had quantified exactly how much of the ventilation rate or pollutant removal could be attributed to turbulence rather than to the mean flow. The new study set out to close that gap by simulating airflow and tracer gas concentration fields in rooms of a three-story cubic building under pumping ventilation, testing three different opening separations defined by the normalized distance between the centers of the two openings. The team used large-eddy simulation with the WALE sub-grid scale model, chosen for its ability to capture near-wall turbulence under the complex, high-Reynolds-number conditions characteristic of vortex-shedding flows.

The simulations were validated against wind tunnel experiments performed in the boundary-layer wind tunnel at the Institute of Industrial Science, the University of Tokyo, using a 1/30-scale model of the building. In those experiments, a constant-temperature hot-wire anemometer with a split-fiber probe measured instantaneous stream-wise velocity at the opening center and in the wake, while the effective ventilation rate was determined with a tracer gas method. Ethylene at 1.21% volume fraction was injected continuously at 2 liters per minute through eight uniformly distributed vertical rods, and the purging flow rate was calculated from the injection rate divided by the time-averaged indoor concentration.

The agreement between simulation and experiment was close. The normalized oscillating frequency at the opening predicted by the simulation, about 0.15, was within roughly 6% of the experimental value of 0.16. Time-averaged velocity profiles in the building wake matched the measurements with deviations below 3% at two of the three vertical measurement lines within the building height range, and the overall deviation for mean and fluctuating velocity statistics stayed below 5%. The simulated air change rates based on the purging flow rate, when computed from the center-point concentration in the same way as the experiments, all fell within a 10% error margin, with the closest agreement, about 1.9%, at the narrowest opening separation tested.

Concentration predictions were somewhat less accurate but still reliable. The simulation underestimated both the mean indoor concentration and its fluctuation, with relative errors generally below 20% and the lowest deviation, 12.6%, at the narrowest opening separation. The authors attribute part of this systematic underestimation to a difference in how the pollutant source was represented: the wind tunnel used eight discrete injection rods, while the simulation treated the entire room as a volumetric source at an equivalent total release rate. The physical presence of sampling tubes and injection rods in the experiment also occupied indoor space, compressing the volume available for pollutant dispersion and raising the measured concentrations above what an unobstructed room would show.

The central result came from decomposing the total mass flux through the opening into a convective component, based on the time-averaged velocity and concentration, and an unsteady component, based on the correlation between fluctuating velocity and fluctuating concentration. For all three opening separations, the unsteady mass flux accounted for over 90% of the total flux through the opening plane and was an order of magnitude larger than the convective flux. Part of this dominance is geometric: the convective flux was positive, carrying pollutants out, only in a limited region near the upper left corner of the opening, while it was negative, returning pollutants to the room, over much of the middle and lower right of the opening. These opposing contributions largely cancel when averaged over the opening area, leaving turbulence as the effective removal mechanism.

To characterize the instantaneous transport events, the researchers recorded time histories of concentration and stream-wise velocity at the opening center over 30,000 time steps and performed a quadrant analysis, classifying each sample by the signs of the velocity and concentration fluctuations. Ejections, in which indoor aged air leaves the room, and sweeps, in which outdoor fresh air is entrained, together accounted for more than 80% of the samples, while inward and outward interactions, which produce negative unsteady mass flux and suppress removal, made up less than 20%. For the two narrower opening separations, sweeps occurred slightly more frequently than ejections, meaning fresh-air entrainment was marginally more common; at the largest separation, ejections became the more frequent event.

Opening separation emerged as the key design lever. As the normalized separation increased from 0.25 to 0.75, the inflow penetrated more deeply into the room, velocity and velocity fluctuations grew in the room’s depth, and the unsteady mass flux was enhanced. At the largest separation, the mean wind speed in the depth of the room became nearly identical to that at the openings, suggesting that pumping ventilation can achieve a large ventilation depth when the opening layout is well designed. Larger separations also increased the frequency of ejection events at the opening and made the distribution of unsteady mass flux across the opening plane more uniform. Frequency distributions showed that concentration fluctuations were strongly positively skewed, with frequent low concentrations and rare extreme peaks, while velocity fluctuations were more symmetric and closer to a normal distribution, acting as the driving force behind the concentration behavior.

The authors caution that the findings come from a reduced-scale model and cannot be directly extended to full-scale buildings without scale conversion, and that the sub-grid scale model’s inherent limitations mean turbulent vortices smaller than the filtering scale may not be fully reproduced. Nevertheless, they argue that the dominant role of unsteady mass transport reflects universal characteristics of wind-driven natural ventilation, offering a theoretical basis for optimizing opening layouts, natural ventilation design, and indoor pollutant control in actual residential buildings. Future work, they write, will extend the analysis to pumping ventilation in real urban settings and to the temporal and spatial transport mechanisms of turbulent coherent structures.

Subject of Research: Simulation and wind tunnel study of unsteady mass transport in pumping ventilation pollutant removal in a three-story residential building

Article Title: Unsteady mass transport mechanism in the pollutant removal of pumping ventilation in a multi-story residential building

Article References: Gui, J.-N., Lin, C., Wang, X.-J., Zhong, H.-Y., Sun, J., Qian, F.-P., Zhao, F.-Y., & Kikumoto, H. (2026). Unsteady mass transport mechanism in the pollutant removal of pumping ventilation in a multi-story residential building. Architectural Intelligence, 5(1), Article 17. https://doi.org/10.1007/s44223-026-00128-4

Image Credits: AI Generated

DOI: 10.1007/s44223-026-00128-4

Keywords: pumping ventilation, large-eddy simulation, wind tunnel experiment, natural ventilation, pollutant removal, turbulent fluctuations, mass flux, quadrant analysis, opening separation, multi-story building, indoor air quality, Unsteady