A modeling technique could help buildings and cities breathe easier
For many people, the price of modern life is being sealed off from fresh air. Even in fairly blissful climates, we spend a good part of our lives in airtight buildings that are kept comfortable by recycling air through lumbering forced-air ventilation systems.
What’s more, residential and commercial buildings account for 40 percent of U.S. energy consumption, and a significant share of that is tied to heating and cooling.
Catherine Gorlé, founder of the Wind Engineering Lab at Stanford Engineering, thinks buildings of the future could harness natural ventilation to stay cool or warm up at just the right times, saving energy and money while improving indoor air quality and comfort at the same time.
Over the past three years she has been working with a variety of collaborators to, quite literally, turn the modeling techniques of computational fluid dynamics inside out. Historically, engineers developed computational fluid dynamics to simulate airflows on the outside of airplane wings. Gorlé wants to use these techniques to simulate how to regulate the internal temperatures of buildings by pulling in cool air and pushing out hot air at night. By lowering the temperature of the entire building at night, the structure can serve as a heat sink during the day, avoiding uncomfortable increases in air temperature.
And while this inside-out shift may seem conceptually simple, most engineers are reluctant to trust the simulation models unless they’ve been validated in wind-tunnel experiments. Even then, Gorlé says, these validated predictive models are fraught with uncertainty.
“A wind tunnel is also a simplification of reality,” she says. “Even if your model can predict what happens in the wind tunnel, it doesn’t mean you’ve captured reality.”
So it is with a caution born of experience that Gorlé and her colleagues are seeking to apply these modeling techniques to predicting how a building might “breathe” on a cool versus hot day—independent of wind conditions, what people are doing inside the building, or how its design and construction materials affect internal air flows.
As a result, most designers consider natural ventilation too risky and expensive. For one thing, natural ventilation usually requires open-air atriums that claim large amounts of usable space. But even when buildings feature big atriums, along with automated windows and vents to shuffle air in and out at the right moments, building owners generally prefer mechanical systems they can control with certainty.
To test their strategy for predicting natural ventilation indoors, the researchers ran detailed simulations on Stanford’s Yang and Yamasaki Environment and Energy building, essentially turning it into their equivalent of a wind tunnel to validate their software models.
Y2E2, as the building is known on campus, was designed with what are currently considered state-of-the-art natural ventilation techniques. It features a four-story atrium fitted with computer-controlled windows on each floor, which open and close in response to the combination of outdoor and indoor temperatures. This design enables the building to breathe by doing what are called “night flushes,” pulling cool air in while pushing out hot air out after sunset.
The question Gorlé posed was whether their computational models could accurately predict the building’s temperature changes over the course of the night. As a starting point, they modeled four separate nights with different indoor and outdoor conditions.
They actually ran two separate computer simulations of building temperatures. The first was a fairly simple program, which crunched data on a handful of major variables and took only about a second to run on a laptop. The second simulation, a complex simulation using computational fluid dynamics, modeled the intricate interaction between scores of factors, presumably making it far more accurate, but also more costly and difficult to run because it required several hours on very high powered hardware.
What Gorlé and her team found was that a smart combination of the two approaches could be both accurate and affordable. The simple version predicted temperatures with a confidence level of about 5 degrees Fahrenheit—good, but not good enough. The researchers then ran far more complex simulations with the second model, and used the results from that to update the simpler program with more refinements. That updated simple model, the researchers found, had a tighter confidence level of 3 degrees Fahrenheit. The predicted temperatures also compare well to temperature measurements in the building.
With a little more work, Gorlé believes it should be possible to create computerized models that are both affordable and accurate enough to design robust natural ventilation systems in the real world.
“This has important practical implications,” Gorlé says. “People who build buildings want them to be comfortable and functional. That is a tall order given the variety of operating conditions a building might experience over its lifetime. By quantifying the effect of this variability during the design, we can be confident that the building will perform as intended.”
In addition to their natural ventilation work, Wind Engineering Lab researchers are also trying to model the airflow for entire cities with the ultimate goal of affecting urban design in a way that would make metropolitan areas healthier, more comfortable, more resilient and energy-efficient. Along these lines, they are studying whether data from wind sensors could be used to improve their models. Another thrust of research delves into air conditions in the developing world, trying to model the health benefits of better ventilation of homes in the slums of Bangladesh, where children living in small single-room units with their families are highly susceptible to infectious diseases.
“All this is about trying to represent reality better,” Gorlé says. “Weather conditions change. The way buildings are used changes. The sources of pollution change. We cannot predict these things exactly, but we can get a much better understanding of their impacts.”
