Urban areas have their own signature infrastructure elements ranging from the non-descript to the well known, which include buildings or towers and various forms of vegetation cover. While the infrastructure elements meet various societal needs and may serve as historic symbols and add aesthetic appeal, they also modify airflow, which influences the urban microclimate - defined by temperature, moisture, and wind perturbations (turbulence). Urban microclimate impacts water use, energy use, and pollutant transport, as well as the overall comfort and well being of the inhabitants of urban areas.
When wind hits a tall building, the building creates frictional and pressure drag forces that result in turbulence on the other side. In this process, the air goes from an area of high pressure - the building's windward side - to an area of low pressure - the building's leeward side. In addition, the spacing between buildings and other features of the urban landscape affect airflow eddying patterns and wind speed. And if you've ever been tossed about by the wind while walking on a city sidewalk, you have experienced how an urban canyon composed of rows of buildings can create wind channeling.
While wind tunnel studies and field experiments are always helpful for urban flow studies, revealing the airflow patterns in an urban area is a difficult task since scientists don't have the continuous datasets revealing the airflow topology features they need for studies. Collecting the data would require covering a whole urban area with sensors for the continuous measurement of flow, which is infeasible. In addition, much of the data that does exist comes from field experiments performed in terrains outside the urban areas because garnering air characteristics inside them is expensive, time-consuming, and tedious.
A research team from the Mechanical Engineering department at University of Utah is using high-performance computing resources from XSEDE to instantaneously and accurately simulate how infrastructure elements, such as parks, buildings, and parking lots, as well as their specifications and variations, affect air characteristics and quality in urban settings. Professors Eric Pardyjak and Rob Stoll, and Research Professor Todd Harman are leading the project, with Ph.D. student Arash Nemati Hayati serving as principal investigator. The work is being done as part of the Green Environmental Urban Simulation for Sustainability (GEnUSiS) project supported by the National Science Foundation.
Through scientific visualizations, the team is able to delve into the complex physics of airflow and define locations of strong rotation called "vortex cores", which determine how pollutants may concentrate in different regions of urban areas. The researchers also consider other topological flow features, such as "saddle points" and "stagnation points", all of which indicate how the airflow pattern changes at different locations between buildings and other elements in cities. These features are relevant to understanding pollutant concentration and energy transport within urban areas.
The eventual goal is for civil engineers, urban planners, and other professionals to use these simulation tools to influence the development of growing cities, improve vegetation cover in places that have dense infrastructure, and more efficiently plan suburbs. Climatologists and atmospheric scientists also could apply these tools toward improving their understanding of how airflow attributes vary among urban areas of low and high infrastructure density. And for urban inhabitants, advancements could mean less air pollution and lower energy costs.
To perform the simulations, the team is using Stampede at the Texas Advanced Computing Center (TACC). Stampede, one of the most powerful and significant current supercomputers in the U.S. for open science research, is a resource of the Extreme Science and Engineering Discovery Environment (XSEDE), a single virtual system funded by the NSF that scientists use to interactively share computing resources, data, and expertise. The team also is employing computational resources at the National Center for Atmospheric Research (NCAR).
The investigation involves using VisIt software for the visualization of large data sets. With the parallel server-client capabilities of VisIt, the researchers visualized the airflow in downtown Salt Lake City. "Without XSEDE, we could never accomplish this, since the data of the city simulations is so large that the serial mode of the software can never handle it", stated Arash Nemati Hayati.
The team has used VisIt in parallel mode employing up to 512 cores on Stampede. A special benefit of XSEDE to the project, Arash Nemati Hayati said, is that it simplifies visualizations for high-resolution urban flow studies. The powerful Stampede resource also has allowed them to confirm the viability of their computational tool called Uintah:MPMICE against empirical data for urban flow studies in typical street canyons.
"Fortunately, we had a valuable empirical wind tunnel dataset for the street canyons that helped us to examine and understand the challenges in accurate modeling of complex topology features in urban areas", Arash Nemati Hayati explained.
Arah Nemati Hayati is very complimentary concerning TACC's visualization portal. "It makes it very simple to do data visualization directly from the web without any software required other than a web browser", he stated.
The team has discovered that boundary conditions outside of the urban areas are one of the most significant issues in validating their numerical results. Inflow conditions, including the wind flow profile and its fluctuations upstream of urban areas, they found, are a potential source for remarkable errors in determining airflow characteristics and how the flow pattern changes in urban areas.
While the researchers have made good progress in defining the appropriate inflow conditions, they intend to address remaining challenges in modeling what takes place in response to the conditions at building and ground surfaces, and move on to larger simulations - specifically, city scales of 5 km x 5 km x 0.3 km - to further investigate the interaction between urban elements and airflow characteristics.
Arah Nemati Hayati said that in the long term Uintah:MPMICE, which has the capability of fully coupled fluid-structure interaction modeling, could be used for more complex flow studies such as storms and hurricanes. In the near term, the team wants to continue to refine the tool for simulating urban flow to make it as accurate as possible.
Another activity on the horizon for the team is a showing of their visualizations at the Leonardo museum, Salt Lake City, in the spring of 2015. The team won the honour of displaying their work at the museum by entering a contest at the University of Utah in November 2014 titled "Research as Art".
The team has presented its work at three conferences and is preparing a paper for the journal of Environmental Fluid Mechanics, which will be submitted in the near future.
The conference proceeding papers include: