The University of Texas at Austin- What Starts Here Changes the World
Services Navigation


Gridlocked: Transportation engineers seek ways to curb traffic woes


How is traffic congestion affected when a highway on-ramp is closed? If grocery stores and post offices are built within walking distance from workplaces, will they help reduce traffic on clogged roadways? How much of a reduction in traffic will result if a commuter rail is built?

Answers to questions like these are the target of research done by two University of Texas at Austin transportation engineers. Their work aims to one day make your drive home easier.

Travis Waller on campus at The University of Texas at Austin

Dr. Travis Waller’s research focuses on complex transportation systems, such as roadways, railways or airline systems, that can be modeled as networks. His models allow travelers in the network to receive information about costs or travel times throughout the system as they travel, so they can continually re-evaluate their travel decisions.

Photo: Charlie Fonville

“A lot of my work involves uncertainty,” says Dr. Travis Waller. “If we’re planning for 20 years in the future, we don’t know now what the travel demand will be. We don’t know how many people will even live here exactly. But we have to build a system and try to manage that system now.”

Waller’s research deals broadly with transportation networks—the highways, roads, stoplights and traffic signs that shape the way we travel on land. He was cited as a top young researcher by the Massachusetts Institute of Technology’s magazine, Technology Review, last year for his work creating software models of dynamic traffic networks.

Traditionally, transportation models account for steady-state traffic conditions—those that do not vary with time. They represent, for example, the number of vehicles on a roadway over a 24-hour period. Dynamic models, which can provide the number of vehicles per minute, or even second, on a roadway, are much more useful in evaluating traffic networks.

These models use complex mathematical equations, formulated by Waller, to attempt to simulate traffic conditions at any given time under any given circumstance. There are many different models that describe a wide range of transportation problems/systems. For example, one model uses equations from algebra that formulate the travel time of a particular road segment using the amount of traffic that will use that segment as the variable. Then, the equations for each segment are strung together, providing a description of the network.

Waller adapted a principle used in finance, where analysts attempt to make decisions that are good, on average, for all possible outcomes.

“We can apply these exact same principles to transportation planning, but no one has. But we should,” he says. “A lot of times it takes time for ideas that evolve in different disciplines to meet.”

Waller's models give other researchers the ability to determine the instantaneous effects of their tool for one driver or 100, how it affects each intersection, each highway's traffic count, each traffic light's timing.He used this principle of risk to overturn the assumption of many traffic engineers that demand levels at certain places—like roadways, or even the number or lanes on a roadway—are known. His work introduced models that treat the future demand levels as uncertain, providing a much better, more realistic estimate of the performance of the traffic system.

“We need to be able to account for driver reactions to car accidents, intelligent transportation systems installed in their vehicles and many other factors,” he says. “Right now there’s a lack of tools to inform drivers of traffic conditions, but a number of researchers are working to correct that.”

Radio traffic updates are a “primitive” example of driver tools, but Waller sees the future as having much more sophisticated means of communication, such as in-car navigation using the global positioning system.

If they’re so important, why don’t we have tools like these now? To create them, researchers need to anticipate how large groups of people simultaneously react to information. That’s where Waller’s highly sophisticated computer models help.

His models give other researchers the ability to determine the instantaneous effects of their tool for one driver or 100, how it affects each intersection, each highway’s traffic count, each traffic light’s timing.

While researchers’ tools in the lab may be years away from the driver, transportation engineers continue to design long-lasting roads, intersections, bridges and other systems.

Chandra Bhat in a classroom at The University of Texas at Austin

Dr. Chandra Bhat displays one of the formulas he developed to help create models of commuter travel patterns.

Waller keeps in mind that once a road is built, it is designed to last more than 50 years.

“Taking into account high-tech tools that drivers may use in the future doesn’t appear to be a pressing need,” he says, “but the decisions we make to build infrastructures now will affect people for decades to come.”

Dr. Chandra Bhat, another transportation engineering professor and the Abou-Ayyash Professorship Fellow, also understands the need for tools to help traffic engineers build for the future. His extensive experience in the area has shown him the need for it.

Bhat works with metropolitan planning organizations in several cities across the United States, helping develop models that act as a guide in mapping the locations of highways, shopping centers and other areas relevant to drivers.

Situated in Austin, Texas, a city that was cited in the 2004 Urban Mobility Report as having the highest traffic delay of metropolitan areas its size, Bhat has a fascinating—and gridlocked—laboratory for his studies right underfoot.

In the Austin metropolitan area, a city of about 800,000 drivers, commuters spend about 31 percent more time traveling from point A to point B during peak travel times than during non-peak times, and with predictions of the area population doubling by 2025, a true transportation crisis looms in the near future.

With a toll system in the works and a commuter rail proposal on the November ballot, Austin resembles many other cities across the nation that are grappling with traffic woes and needing more information about traffic flow and commuter options.

Bhat's computer modeling of human behavior has fundamentally changed the ability to build realistic models of behavior. ... Shortly after he developed it, Daniel McFadden, the 2000 Economics Nobel laureate, cited it in his Nobel Prize article.In a fascinating study last winter of Austin commuters Bhat and his colleagues gained insight into how commuters navigate through their day and how individual travel plans would be affected by the implementation of a toll road or commuter rail system. The goal was the collection of enough data to determine a formula that would lure commuters out of their personal vehicles and onto some form of congestion-relieving public transportation.

Bhat found non-work stops, such as child care drop-off, grocery shopping or exercising at a gym, have significant implications on peak traffic levels and overall air quality.

“Like many things in the 50 years since our road networks were designed, household structure has undergone substantial changes,” said Bhat. “An increasing number of two-adult, two-worker families and working single parent and/or working single adult families use these roads. Because of schedule and time constraints, individuals in these households are chaining non-work activities with their commute to be more efficient, making them more reliant on cars.”

Bhat’s survey results revealed behavioral tendencies that likely are reflected in dense urban areas from coast to coast—because of non-work stops, people prefer their cars to public transportation. Encouragingly, survey respondents did indicate that if a commuter rail system were implemented extensively around their travel area, they would use it, primarily because of its reliability and availability.

Traffic on Interstate 35 in Austin, Texas

Traffic moves north and south on Interstate 35 in Austin, Texas.

Photo: Marsha Miller

While Waller predicts traffic levels based on driver reaction to the immediate traffic snarls, Bhat predicts traffic levels based on drivers’ overall lifestyle needs. Bhat’s computer modeling of human behavior has fundamentally changed the ability to build realistic models of behavior in a broad number of disciplines. Shortly after he developed it, Daniel McFadden, the 2000 Economics Nobel laureate, cited it in his Nobel Prize article. Bhat’s methods of predicting human decisions are now applied routinely to problems in economics and marketing.

To create his models, Bhat collects data on the choices drivers make, such as the number of trips taken by a commuter, at what time of day, how many people are in the vehicle and what route the driver chooses. He then develops complicated mathematical formulas associating these travel choices with information such as demographics of individuals and households, the level of service offered by transportation systems available (such as travel time, cost and reliability) and the “land-use configuration” of an area, which describes whether or not there are offices and shopping areas relatively close to homes.

One formula, for example, calculates how a person’s probability of choosing a certain travel mode (such as car, bus or carpool) varies depending on his/her demographics and the availability and cost of each mode.

Whether approaching the traffic conundrum from Waller’s perspective or Bhat’s angle, the answer to the country’s traffic woes is not a simple one, and as Bhat points out, “there’s no silver bullet that’s going to immediately solve the problem.” For the meantime, as researchers puzzle out realistic ways to alter deeply engrained commuter behaviors, the nation’s 200 million drivers will continue to have a slow ride.

Stephanie Logerot

Related Stories:

Related Sites:

Office of Public Affairs
P.O. Box Z
Austin, Texas
78713-7509

512-471-3151
Fax 512-471-5812


  Updated 2014 October 13
  Comments to utopa@www.utexas.edu