Géza Pattantyús-Ábrahám Doctoral School of Mechanical Engineering 

BME, Faculty of Mechanical Engineering, Department of Fluid Mechanics

Supervisor: Dr. Kristóf Gergely

Investigation of urban ventilation and pollutant dispersion using scale-resolving turbulence models

Introducing the research area

According to the World Health Organization (WHO), air pollution claims seven million lives every year. This means that one in ten deaths is caused by poor air quality and its consequences, such as lung cancer, stroke, heart disease, and respiratory diseases. Another WHO research reports that 99% of the world’s population lives in places where the WHO air quality guidelines levels are not met. The need to improve urban air quality and ventilation is increasingly being cited as a justification for the rehabilitation of the cities. If we can promote the rapid removal of pollutants by designing appropriately shaped buildings or placing landmarks (e.g., vegetation, parking spaces for cars) suitably, we can help to make our neighborhood more livable.

Brief introduction of the research place

The infrastructure for my PhD research is provided at the Department of Fluid Mechanics of BME-GPK. I work in the Atmospheric Flows Research Group, under the supervision of Dr. Gergely Kristóf. During my research, I am mainly working with Computational Fluid Dynamics (CFD) simulations; however, I have also carried out wind tunnel measurements in the large wind tunnel of the Theodore von Kármán Wind Tunnel Laboratory. Moreover, I have also had the opportunity to work with colleagues from foreign research institutions such as Sun Yat-sen University (SYSU SESE) and the University of Karlsruhe (KIT IfH).

History and context of the research

My research can be positioned at the border of meteorology and fluid dynamics. The most important characteristics of urban flow and pollutant dispersion processes are (a) the three-dimensional nature of the flow structures, (b) the unsteadiness of the flow processes, and (c) the anisotropy of the turbulent scalar fluxes [1]. However, CFD solvers based on steady-state Reynolds Averaged (RANS) turbulence models, which are widely used in industrial practice and have been applied in most of the previous publications, are not able to accurately model the properties (b) and (c). The underestimation of the lateral diffusion of pollutants is a common problem in such simulations, as well as their inability to predict peak pollutant concentrations that can vary significantly in time [2].

Several studies, e.g. [3,4] favor the application of scale-resolving turbulence models, such as Large Eddy Simulation (LES) for capturing the flow structures, as they can resolve the turbulence both in space and in time. These turbulence models provide much more accurate velocity and concentration results, with a sufficiently fine spatial and temporal resolution, but their computational costs are significantly higher than those of RANS models (LES: weeks or months; RANS: hours or days). Therefore, extensive parameter studies and thus the detailed optimization of building geometry using scale-resolving turbulence models have not been possible in the past, or only on a very few occasions [5,6].

The research goals, open questions

The primary objective of my research is the development and application of numerical methods based on Large Eddy Simulation for modelling urban flow and pollutant dispersion processes. In my research, I answer the following questions:

 

1. How can urban morphology be described using sufficiently small number of geometric parameters?
2. How can appropriate boundary conditions be prescribed to model the natural wind?
3. What characteristic flow structures dominate the pollutant dispersion processes within the urban canopy?
4. How can the ventilation efficiency of a building configuration be quantified?
5. Which building arrangements can be considered favorable for air quality?
6. How can the optimization process of building shapes and arrangements be accelerated?
7. What atmospheric phenomena cannot be taken into account either in wind tunnel experiments or in conventional CFD models?
8. What conclusions can be drawn from time-averaged velocity and concentration results, and what characteristics cannot be assessed merely based on these?

Methods

Modelling the geometry of the cities

As can be seen in Figure 1, the buildings in cities around the world are usually placed in some kind of repeating pattern. Modelling the city as a periodic building pattern allows the geometry to be described with a sufficiently small number of geometric parameters, and in addition, by assuming periodicity, the size of the computational domain required for numerical simulations – and thus the simulation runtime while maintaining the same resolution – can be significantly reduced.

Figure 1. Periodic building patterns in cities around the world (source). From left to right: Casablanca, New York, Rome, Barcelona, Mexico City, Paris, La Plata.

Modelling natural wind

The characteristics of natural wind can be modeled by the so-called atmospheric boundary layer (ABL). In the numerical wind tunnel model constructed in the present research, the inlet geometry was fine-tuned for the ABL parameters to resemble those of the No. 4 boundary layer wind tunnel of the KIT IfH, as shown in Figure 2 [S1,S8].

Figure 2. From left to right: Layout and dimensions of the numerical wind tunnel. Vertical profiles of the streamwise mean velocity and the turbulence intensity. Velocity power spectra.

The effect of turbulent structures originating from the higher atmosphere with a size exceeding that of the computational domain, can be considered by the so-called Transient Wind Forcing (TWF) model. It was demonstrated that the rapid changes in wind direction and magnitude can cause a significant difference in the resulting dispersion field compared to that of conventional, constant-wind CFD simulations, and even that of wind tunnel measurements [S4].

Modelling pollutant dispersion

The emission of traffic-related air pollutants can be modeled by placing sources near the traffic lanes. In the numerical wind tunnel, dispersion is modeled using the Eulerian approach and the analogy between heat and mass transport processes; and in the TWF model, Lagrangian particle tracking (Discrete Phase Model, DPM) is applied. Both methods were validated with experimental results [S1,S4,S5,S8].

Results

GPU-based Large Eddy Simulation (“numerical wind tunnel”)

The numerical wind tunnel was utilized to compare the ventilation efficiency of 28 different building patterns of equal total volume, in the case of perpendicular wind direction (Figure 3) [S2,S10]. Based on the configuration containing five parallel streets and two cross-streets (shown in Figure 2), the following conclusions can be drawn. Note that the reference case is the H/W = 1 height-to-width aspect ratio street canyons, which is widely used in the literature.

●     Asymmetric canyons are advantageous only if the building height ratio is high (max. 33% concentration decrease at pedestrian head height).

●     Towers in matrix (aligned) arrangement are unable to mitigate pedestrian exposure to traffic-induced air pollutants, regardless of the tower width or tower height.

●     Towers in staggered arrangement, however, can decrease pedestrian exposure with increasing building height heterogeneity (max. by 46%).

●     High-rise buildings can effectively mitigate pedestrian exposure in matrix, intermediate, and staggered arrangements (max. by 55%).

●     Vertically elevated uniform buildings can cause a substantial pollutant reduction at pedestrian head height, proportional to the ground clearance (max. by 60%, short-term effect).

Based on the results of the screening via the GPU-based Large Eddy Simulation, it can be concluded that buildings with 1-2 stories of ground clearance should be placed at the edges of cities and next to larger openings within, such as squares. In the middle of the cities, the construction of buildings with varying roof height, which are able to generate enough turbulence to promote mixing at ground level, is advised.

Figure 3. Comparison of the ventilation efficiency of different periodic building patterns.

Based on the preliminary studies described above, in addition to the reference case, street canyons of H/W = 0.5…1.5 aspect ratio (towers on top of lower continuous buildings either in matrix/aligned and staggered arrangement) were chosen for further, detailed analysis. In addition to the constant total volume, the overall plan area of these building configurations is also constant, which means that they can accommodate the same number of apartments and offices and the same number of street-level shops.

Wind tunnel experiments, measurement-driven Large Eddy Simulation

The ventilation of the three chosen building layout was tested in the large Goettingen-type wind tunnel of the Department of Fluid Mechanics. The velocity field was mapped by Laser Doppler Anemometry (LDA), and the concentration field was examined by Fast Flame Ionization Detection (FFID). The measured velocity time series were also used for the propulsion of the TWF model (Figure 4) [S4]. The experimental and simulation results consistently proved the beneficial effects of varying roof height on pedestrian exposure: 92% improvement was found in the matrix (aligned) tower arrangement, and 237% improvement was found in the staggered tower arrangement compared to the series of uniform street canyons [S5].

Figure 4. Wind tunnel measurements, the periodic time-instantaneous velocity field, as well as the aperiodic mean dispersion field (based on particles emitted at street level) produced by the measurement-driven LES model.

Concentration fluctuations

Although most studies in the literature (and the part of my research presented above) use the mean concentration distribution only as a basis for characterizing urban ventilation; when dealing with unpleasant odors, infectious, flammable, or explosive substances, the assessment of the peak concentrations cannot be ignored. These concentration peaks can exceed the mean concentration by orders of magnitude, as the pointwise probability distribution of the concentration can be described by the Gamma distribution [2]. My research has shown that although varying roof height results in lower mean concentrations at ground level, the representative concentration maxima (the 99th percentile of the concentration distribution) for towers with matrix or staggered arrangements can significantly exceed the extreme values measured within the uniform street canyons [S9].

Expected impact and further research

Based on the results obtained so far, it is confirmed that the GPU-based LES can be used to quickly select potentially effective building layouts at an early stage of urban planning. Using the GPU, the simulation run time can be decreased by two orders of magnitude, while keeping the hardware costs at the same level [S8]. As a real-world engineering application, the ventilation of the Újgyőr market area in Miskolc was studied for the present and four planned building arrangements [S6, video in Hungarian]. In addition, the further development of the TWF model is underway: our main goals are (1) to validate the thermal model [S7] capable of simulating flows near heated walls and (2) to develope the particle-based dispersion model to be capable of predicting concentration fluctuations.

Publications, references, links

List of corresponding own publications:

[S1]      Kristóf, G., & Papp, B. (2018). Application of GPU-based Large Eddy Simulation in urban dispersion studies. Atmosphere, 9(11), 442. [DOI]

[S2]      Papp, B., Kristóf, G. (2019). Épületmintázatok optimalizálása a levegőminőség javításának érdekében GPU alapú nagyörvény szimulációval. XXVII. Nemzetközi Gépészeti Konferencia (OGÉT 2019) kiadványkötete (ISSN 2068-1267), 400-403.

[S3]      Papp, B., Kristóf, G., Gromke, C. (2020). Épületek szélterhelésének becslése GPU alapú nagyörvény szimulációval numerikus szélcsatornában. XXVIII. Nemzetközi Gépészeti Konferencia (OGÉT 2020) kiadványkötete (ISSN 2668-9685), 161-165.

[S4]      Kristóf, G., Papp, B., Wang, H., & Hang, J. (2020). Investigation of the flow and dispersion characteristics of repeated orographic structures by assuming transient wind forcing. Journal of Wind Engineering and Industrial Aerodynamics, 197, 104087. [DOI]

[S5]      Papp, B., Kristóf, G., Istók, B., Koren, M., Balczó, M., Balogh, M. (2021). Measurement-driven Large Eddy Simulation of dispersion in street canyons of variable building height. Journal of Wind Engineering and Industrial Aerodynamics, 211, 104495. [DOI]

[S6]      Szilágyi, M. Á., Papp, B. (2021). Miskolc átszellőzésének vizsgálata GPU alapú nagyörvény szimulációval. XXVIIII. Nemzetközi Gépészeti Konferencia (OGÉT 2021) kiadványkötete (ISSN 2668-9685), 80-83.

[S7]      Papp, B., Kristóf, G. (2021). The role of thermal convection in the dispersion of traffic-induced air pollutants in the urban environment. 20th International Conference on Harmonisation within Atmospheric Dispersion Modelling for Regulatory Purposes (HARMO20).

[S8]      Papp, B., Kristóf, G., Gromke, C. (2021). Application and assessment of a GPU-based LES method for predicting dynamic wind loads on buildings. Journal of Wind Engineering and Industrial Aerodynamics, 217, 104739. [DOI]

[S9]      Papp, B., Istók, B., Koren, M., Balczó, M., Kristóf, G. (2022).Statistical assessment of the ventilation of street canyons based on time-resolved wind tunnel experiments. PHYSMOD 2022 – International Workshop on Flow and Dispersion Phenomena, Book of extended abstracts (ISBN 978-80-87012-81-9), 143-155.

[S10]    Papp, B., Kristóf, G. (2022). Building Patterns Favorable for Air Quality: A Parameter Study Using LES. Proceedings of the Conference on Modelling Fluid Flow CMFF’22 (ISBN 978-963-421-881-4), 443-456.

Table of links:

• ”Tamás Roska” Scientific Lecture (Scientific Plenary Lecture given at the Engineering Sciences Section of the XXXV. National Scientific Students’ Association’s Conference (OTDK), in Hungarian.)

• MTMT profile

• GoogleScholar profile

• ResearchGate profile

List of references:

[1]        Tominaga, Y., & Stathopoulos, T. (2013). CFD simulation of near-field pollutant dispersion in the urban environment: A review of current modeling techniques. Atmospheric Environment, 79, 716-730. [DOI]

[2]        Cassiani, M., Bertagni, M. B., Marro, M., & Salizzoni, P. (2020). Concentration fluctuations from localized atmospheric releases. Boundary-Layer Meteorology, 177(2), 461-510. [DOI]

[3]        Tominaga, Y., & Stathopoulos, T. (2016). Ten questions concerning modeling of near-field pollutant dispersion in the built environment. Building and Environment, 105, 390-402. [DOI]

[4]        Blocken, B. (2018). LES over RANS in building simulation for outdoor and indoor applications: A foregone conclusion? Building Simulation, 11(5), 821-870). [DOI]

[5]        Li, Z., Ming, T., Liu, S., Peng, C., de Richter, R., Li, W., ... & Wen, C. Y. (2021). Review on pollutant dispersion in urban areas-part A: Effects of mechanical factors and urban morphology. Building and Environment, 190, 107534. [DOI]

[6]        Li, Z., Ming, T., Shi, T., Zhang, H., Wen, C. Y., Lu, X., ... & Peng, C. (2021). Review on pollutant dispersion in urban areas-part B: Local mitigation strategies, optimization framework, and evaluation theory. Building and Environment, 198, 107890. [DOI]