Sunday, 28 January 2024
Hall E (The Baltimore Convention Center)
Handout (1.5 MB)
Introduction
Every year, air pollution kills millions of people, while others face serious short and long term health complications due to exposure. With the majority of people living in urban areas, study of air quality in urban environments is therefore of growing importance. Smoke from wildfires has especially become a concern, with recent waves of wildfires resulting in the transport of smoke to major urban areas such as New York, Toronto, Montreal, and Vancouver among others. Air quality is seriously affected by wildfire smoke and has the potential to significantly impact the health of those exposed. Lesser effects include eye and respiratory irritation, with more serious effects including the exacerbation of preexisting lung diseases and even premature death (Liu et al., 2015). As winds carry wildfire smoke from their sources, the smoke is often blown directly over densely populated, urban environments, affecting millions of people. As such, it is necessary to broaden our understanding of how urban air quality is affected by atmospheric flow. Computational fluid dynamics (CFD) provides one way of investigating atmospheric dispersion processes and issues related to air quality and health in urban environments and which can have many advantages. In particular, CFD has been shown to be useful in predicting pedestrian-level wind conditions and safety, near and far field pollutant dispersion, and the in-depth study of other dynamical processes. CFD makes use of the Navier-Stokes equations for conservation of momentum, mass, energy, and pollutant transport to solve the governing equations of fluid flow. However, direct numerical simulations solving these equations are far too computationally expensive for the highly complex scenarios found in urban environments, and thus various approximations to the Navier-Stokes equations are applied. While Large Eddy Simulations (LES) can be highly accurate due to their explicit modeling of turbulent flow, these too require large amounts of computational resources and suffer from a lack of best practice guidelines. For these reasons, the dominant method used in CFD has become the process of solving for the steady-state, mean flow, known as Reynolds-Averaged Navier-Stokes (RANS) simulations. RANS simulations maintain a high level of accuracy when applied correctly, and due to their modeling of turbulence properties, manage to do so at significantly decreased computational cost. There are multiple approaches taken to modeling the effects of turbulence, with one of the most commonly implemented types being eddy-viscosity models such as k-𝜀 models and k-𝜔 models. One of the largest challenges in CFD comes in the appropriate selection of a turbulence model, as each has its own strengths and weaknesses, and it can often be hard to know beforehand which model will provide the most accurate results for a given scenario prior to testing. However, the standard k-𝜀 model and Shear Stress Transport (SST) k-𝜔 model have both shown moderate to high success in urban pollutant dispersion simulations. In addition, the parameter known as the turbulent Schmidt number, which relates the eddy viscosity with the turbulent mass diffusivity, has been found to greatly affect the accuracy of RANS based pollutant dispersion simulations. The correct value for this parameter is often also not known before starting simulations, and as such it is necessary to test multiple values.
Methodology
This study uses a RANS based approach to simulate wind flow within Montreal’s city center for the date of July 17th, 2023, which featured moderately strong southwest winds carrying significant amounts of wildfire smoke through the city. Inlet wind profiles are derived using the methodology developed by Richards and Hoxey (1993), with reference velocity measurements from a weather station at the Montreal International Airport. An air quality sampling station located at the airport is used to estimate the concentration of wildfire smoke being transported into the city, and its subsequent dispersion is simulated. The computational domain is approximately 9×7.5×1.7 km in size, with downtown building geometries explicitly modeled (Fig. 1), and is created following the COST Action 732 best practice guidelines devised by Franke et al. (2007). A mesh sensitivity analysis is performed to ensure the results are grid independent. Multiple turbulence models, including the standard k-𝜀 model and the SST k-𝜔 model, and turbulent Schmidt number values are tested to determine the most appropriate selection for this scenario.
Results
The wind fields found in the simulations are validated using long-range doppler LiDAR wind profiler (Halo Photonics Vertical Pro) measurements collected near the center of the domain, shown in Fig. 2, and the predicted concentrations of smoke are compared against other air quality sampling stations within the domain. Areas found to have locally high wind speeds, poor natural ventilation, or increased amounts of wildfire smoke are investigated. A CFD approach has advantages over the other primary methods used to assess the effects of wildfire smoke on air quality such as satellites or air quality monitoring stations, as CFD simulations can provide details about the amount of smoke at every point within the simulated region of interest. The topographic effects of Mount Royal are also analyzed. By simulating the dispersion of wildfire smoke over urban environments such as Montreal, areas which pose risks to the health and safety of people can be identified with the goal of better informing and protecting citizens.
References
Franke, J., Hellsten, A., Schlünzen, H., & Carissimo, B., 2007: Best Practice Guideline for the CFD simulation of flows in the urban environment, COST Office.
Liu, J. C., Pereira, G., Uhl, S. A., Bravo, M. A., & Bell, M. L., 2015: A systematic review of the physical health impacts from non-occupational exposure to wildfire smoke, Environ. Research., 136, 120-132, https://doi.org/10.1016/j.envres.2014.10.015.
Richards, P. J., & Hoxey, R. P., 1993: Appropriate boundary conditions for computational wind engineering models using the k-ϵ turbulence model, J. Wind Eng. Ind. Aerodyn., 46-47, 145-153, https://doi.org/10.1016/0167-6105(93)90124-7.
Every year, air pollution kills millions of people, while others face serious short and long term health complications due to exposure. With the majority of people living in urban areas, study of air quality in urban environments is therefore of growing importance. Smoke from wildfires has especially become a concern, with recent waves of wildfires resulting in the transport of smoke to major urban areas such as New York, Toronto, Montreal, and Vancouver among others. Air quality is seriously affected by wildfire smoke and has the potential to significantly impact the health of those exposed. Lesser effects include eye and respiratory irritation, with more serious effects including the exacerbation of preexisting lung diseases and even premature death (Liu et al., 2015). As winds carry wildfire smoke from their sources, the smoke is often blown directly over densely populated, urban environments, affecting millions of people. As such, it is necessary to broaden our understanding of how urban air quality is affected by atmospheric flow. Computational fluid dynamics (CFD) provides one way of investigating atmospheric dispersion processes and issues related to air quality and health in urban environments and which can have many advantages. In particular, CFD has been shown to be useful in predicting pedestrian-level wind conditions and safety, near and far field pollutant dispersion, and the in-depth study of other dynamical processes. CFD makes use of the Navier-Stokes equations for conservation of momentum, mass, energy, and pollutant transport to solve the governing equations of fluid flow. However, direct numerical simulations solving these equations are far too computationally expensive for the highly complex scenarios found in urban environments, and thus various approximations to the Navier-Stokes equations are applied. While Large Eddy Simulations (LES) can be highly accurate due to their explicit modeling of turbulent flow, these too require large amounts of computational resources and suffer from a lack of best practice guidelines. For these reasons, the dominant method used in CFD has become the process of solving for the steady-state, mean flow, known as Reynolds-Averaged Navier-Stokes (RANS) simulations. RANS simulations maintain a high level of accuracy when applied correctly, and due to their modeling of turbulence properties, manage to do so at significantly decreased computational cost. There are multiple approaches taken to modeling the effects of turbulence, with one of the most commonly implemented types being eddy-viscosity models such as k-𝜀 models and k-𝜔 models. One of the largest challenges in CFD comes in the appropriate selection of a turbulence model, as each has its own strengths and weaknesses, and it can often be hard to know beforehand which model will provide the most accurate results for a given scenario prior to testing. However, the standard k-𝜀 model and Shear Stress Transport (SST) k-𝜔 model have both shown moderate to high success in urban pollutant dispersion simulations. In addition, the parameter known as the turbulent Schmidt number, which relates the eddy viscosity with the turbulent mass diffusivity, has been found to greatly affect the accuracy of RANS based pollutant dispersion simulations. The correct value for this parameter is often also not known before starting simulations, and as such it is necessary to test multiple values.
Methodology
This study uses a RANS based approach to simulate wind flow within Montreal’s city center for the date of July 17th, 2023, which featured moderately strong southwest winds carrying significant amounts of wildfire smoke through the city. Inlet wind profiles are derived using the methodology developed by Richards and Hoxey (1993), with reference velocity measurements from a weather station at the Montreal International Airport. An air quality sampling station located at the airport is used to estimate the concentration of wildfire smoke being transported into the city, and its subsequent dispersion is simulated. The computational domain is approximately 9×7.5×1.7 km in size, with downtown building geometries explicitly modeled (Fig. 1), and is created following the COST Action 732 best practice guidelines devised by Franke et al. (2007). A mesh sensitivity analysis is performed to ensure the results are grid independent. Multiple turbulence models, including the standard k-𝜀 model and the SST k-𝜔 model, and turbulent Schmidt number values are tested to determine the most appropriate selection for this scenario.
Results
The wind fields found in the simulations are validated using long-range doppler LiDAR wind profiler (Halo Photonics Vertical Pro) measurements collected near the center of the domain, shown in Fig. 2, and the predicted concentrations of smoke are compared against other air quality sampling stations within the domain. Areas found to have locally high wind speeds, poor natural ventilation, or increased amounts of wildfire smoke are investigated. A CFD approach has advantages over the other primary methods used to assess the effects of wildfire smoke on air quality such as satellites or air quality monitoring stations, as CFD simulations can provide details about the amount of smoke at every point within the simulated region of interest. The topographic effects of Mount Royal are also analyzed. By simulating the dispersion of wildfire smoke over urban environments such as Montreal, areas which pose risks to the health and safety of people can be identified with the goal of better informing and protecting citizens.
References
Franke, J., Hellsten, A., Schlünzen, H., & Carissimo, B., 2007: Best Practice Guideline for the CFD simulation of flows in the urban environment, COST Office.
Liu, J. C., Pereira, G., Uhl, S. A., Bravo, M. A., & Bell, M. L., 2015: A systematic review of the physical health impacts from non-occupational exposure to wildfire smoke, Environ. Research., 136, 120-132, https://doi.org/10.1016/j.envres.2014.10.015.
Richards, P. J., & Hoxey, R. P., 1993: Appropriate boundary conditions for computational wind engineering models using the k-ϵ turbulence model, J. Wind Eng. Ind. Aerodyn., 46-47, 145-153, https://doi.org/10.1016/0167-6105(93)90124-7.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner