Real-time answers for traffic jams
A digital twin could make roads safer, shorten commute times, and reduce energy use.
Guest post by: Alisa Alering, Science Node
First published on Science Node on 21st January 2019
Imagine that spot on your commute where the traffic slows down. No matter what, you always find yourself behind a long line of cars, wishing the snarl would just disappear. All the while your patience is thinning, and your idling car is burning through fuel.
The thing about traffic, though, is that it’s wildly complex. You’re on your commute, but so are hundreds if not thousands of others. Then add cyclists, school buses, delivery trucks, and long-distance travelers. Not to mention time of day, road conditions, and the volume of traffic on alternate routes.
Replacing that slow intersection with a roundabout might help, or diverting large trucks to another route. Maybe the timing of traffic signals needs to be adjusted. But with so many factors to consider, it’s difficult for city officials to choose the best solution.
Juliette Ugirumurera wants to help. A computational scientist at the US Department of Energy’s National Renewable Energy Laboratory (NREL), Ugirumurera is working on using dynamic traffic simulation to investigate the causes of traffic congestion and come up with effective management strategies.
“We can look at what actually causes congestion, and at what causes accidents,” says Ugirumurera. “If a certain area is always congested, we want to find out why that is.”
It’s not just about you
Highway congestion wastes over 3 billion gallons of fuel each year and causes 7 billion hours of lost productivity. Research shows that congestion could be reduced by up to 30% through the use of near real-time traffic monitoring and adapted traffic control signals.
Which would be great, because if people spend less time in traffic they use less fuel and produce fewer CO2 emissions. But to make these fuel-saving real-time decisions, you need a computer that can transform complex data into useful information in a matter of seconds.
Ugirumurera’s team is working to optimize these complex, large-scale simulations by running them on high-performance machines, such as the Cori supercomputer at NERSC (National Energy Research Scientific Computing Center) and the Eagle supercomputer at NREL.
The Open Traffic Model (OTM) is a high-performance macroscopic simulator that uses more than 62,000 nodes in a neural network and is able to represent over 170,000 different links (e.g., roads).
Running the simulation on a single node initially took 15,000 seconds (about 4 hours). But improving the algorithms and dividing the computation over multiple nodes improved the speed and reduced the execution time to a few seconds.
“How do you create algorithms that can distribute your computation and simulation to different nodes, and then use their power to get faster results?” asks Ugirumurera. “That’s our big question.”
Next step: real life
To continue to leverage the power of high-performance computing for transportation research, NREL is collaborating with the Computational Urban Sciences Group at Oak Ridge National Laboratory (ORNL), the city of Chattanooga, Tennessee, and the Tennessee Department of Transportation in the development of a digital twin for the region.
“We’re creating a digital twin of the traffic in Chattanooga to get a better understanding of the situation,” says Ugirumurera. “This will inform their traffic management people and help them decide how to improve the use of the roads and reduce congestion and overall transportation energy use.”
This digital twin, a data-informed simulation, will capture real-time traffic conditions of the entire Chattanooga metropolitan region via sensors installed on roads. At this stage of traffic improvement, there’s still a lot of power in observation alone.
“Just having observability into what’s happening can help city planners a long way in knowing what to do,” says Ugirumurera.
What makes this model unique—and potentially uniquely helpful—is its large scale over a wide region and the capacity to consider connections between lots of different roads with a high variety of vehicles traveling over them and interacting.
The ultimate goal is to optimize traffic infrastructure and signals in order to achieve a 20% energy savings in the region.
“We hope the work can then be used in other cities and regions,” says Ugirumurera. “If policy makers and city planners make more informed decisions, that benefits the larger population.”
Ugirumurera previously worked with microgrid energy systems, which are also quite complex. This is when she first started looking at using multiple nodes to get results to complex problems faster.
“For me, it’s very important to see that the research can impact real life,” she says. “There are so many potential applications in computing—from cancer research to reducing poverty. It’s an opportunity to apply your skill and your drive to benefit real people in their day-to-day life.”
About the author: Alisa Alering, Managing Editor, Science Node
Originally trained as a librarian, Alisa loves tracking down the science behind her stories and learning something new about technology every day. With previous experience as a freelance writer and photo editor, she has held positions at Indiana University Press, PBS, and Google and earned degrees from Penn State and Indiana University. She particularly enjoys writing about women and diversity in technology, digital humanities, and the intersection of science and the natural world.
About Science Node
Science Node is an online magazine that connects the global research community, exploring how tech works and showing why it matters to our everyday lives. sciencenode.org