The NMC's web-based visualization tool displays simulations from DTA models. The left image shows the model's estimation of the time it takes to reach different sections of downtown during the morning peak. The right image shows the model's estimation of bus passengers boarding and disembarking at each stop on a specific route. Image courtesy TACC.
Ah, the wonders of modern urban life: There's the culture and cuisine, higher ed and health care, and the hours in bumper-to-bumper traffic - wait, that last one's not high on your list?
It wasn't high on the city of Austin's list either, which is why they've enlisted researchers within the Center for Transportation Research (CTR) at The University of Texas at Austin, US, to tame their traffic snarls. In partnership with the Texas Advanced Computing Center (TACC), Austin is now harnessing supercomputers to relieve traffic congestion once and for all.
To corral Austin's traffic, researchers adapted existing dynamic traffic assignment (DTA) models to discern how urban expansion affects traffic flow. They subdivided Texas's capital city and analyzed zone data provided by the Capital Area Metropolitan Planning Organization.
But with a population pushing two million residents in the greater Austin metropolitan area and a footprint covering an area over 271 square miles (704 square kilometers), it didn't take long before the DTA models generated more data than could be managed by in-house resources.
"Sometimes our models took days to run, and if you're a decision maker that's an eternity," says Natalia Ruiz-Juri, assistant director for the CTR's Network Modeling Center (NMC).
Natalia Ruiz-Juri, assistant director for the Center for Transportation Research's Network Modeling Center.Photo courtesy TACC.
To bypass this data traffic jam, Ruiz-Juri and the NMC looked to the engineers at TACC. Greg Abram, a research scientist in TACC's Scalable Visualization Technologies Group, Weija Xu, TACC' s manager of Data Mining & Statistics Group, and Amit Gupta, TACC research associate, turned their attention to decreasing model simulation time.
Together, the team developed an interactive, web-based visualization tool to offer Austin ways to view traffic simulation data with a focus on the evolution of spatial information. By consulting TACC, Ruiz was also able to leverage time on the Extreme Science and Engineering Discovery Environment (XSEDE) to gain access to TACC's Stampede. That enabled Xu's team to test and develop algorithms that improved the DTA models.
"By working with the TACC team, we have seen some of our computations run five to 10 times faster, allowing us to input proposed scenarios and estimate the potential consequences so Austin city planners can make quicker more well-informed decisions," Ruiz-Juri notes.
As Austin grows, the tools provided by TACC will enable city planners to anticipate traffic troubles - heading 'em off at the pass, so to speak.
--Lance Farrell
