Khani: Research on Transit Networks

Alireza Khani often describes public transit as his “favorite mode of transportation.” While many people appreciate transit that is clean and timely, few regard it as more than a functional convenience. Khani, however, is drawn to examine transit closely, to know its complexities and challenges. Khani’s research is focused on problems related to transit networks. The network approach gives Khani a rather unique perspective; although a lot of transit research takes place across the country and around the world, few studies focus on a network model.

The Big Picture

By design, a transit network is dynamic and its services contain intentional interruptions (contrast this to a road system, which is designed to be stable and well-connected). A dynamic transit network changes according to time. Transit service contains gaps in connections due to location and time. So at some times, a rider could travel via transit to a destination quite easily, but at another time of day or on another day that same trip could be unavailable, could take much longer, could require multiple transfers or require walking a distance between transports. This interrupted and dynamic nature is part of what makes modeling transit networks complex—and part of what draws Alireza Khani to study and solve the problems that nature presents.

In addition to the global characteristics of a dynamic and interrupted network, each component of a transit network can be complex. A transit system usually involves several transportation modes—Light Rail Transit (LRT), Bus Rapid Transit (BRT), express busses, regular busses. Each mode has unique assets and costs. Transit is used by riders with unique destinations and unique priorities and preferences. Each tour (a complete transit experience, leaving to returning) involves multiple decision points for riders: choosing a departure time, a mode, a route, an access point, and a debarking destination. A rider might prefer LRT or express bus; another might prioritize time, saving money, or minimizing walking distance. In addition, individual preferences are also dynamic, possibly changing each trip. Individual choices introduce stochasticity (irregularity, a lack of predictable pattern) and heterogeneity (non-uniformity, diversity) into the model. Determining user behavior is complex because of the many options and personal choices.

“Riders make various choices, and their preferences vary on different days,” Khani summarizes. “In the Twin Cities, there are approximately 300,000* weekly transit riders. That adds a lot of complexity to a model!”

Metro Transit reported 266,916 average weekly ridership in 2016.

Khani is driven to understand rider behavior because planners need that information to predict ridership and adjust the system to provide service when and where it is needed. And new technology is helping Khani move that research forward.

Card Data

Past research often relied on self-reports from riders via on-board surveys or sample counts to understand rider behavior. Now, researchers have a new, more reliable tool for gathering ridership information—smart cards.

In the Minneapolis-St. Paul area, Metro Transit sells electronic payment cards (Go-To cards) that allow regular riders to just touch and go, ending the need for correct change and making passenger boarding faster. When riders pay with an electronic swipe card, data on the location, time, and route number are gathered automatically. Khani is working with Metro Transit and using card data to understand rider behavior.

“Metro Transit has information on all the uses or tags of Go-To card users. That is a huge database of information that is automatically collected,” says Khani. “It gives us greater ability to deal with the randomness in the system.”

Card data offers a lot of advantages for transit planning organizations. More data can be collected more quickly through electronic cards, and that data can be more easily parsed by rider or by day. More data aids in the creation of a more robust Origin-Destination (OD) demand matrix (which routes are most in demand). OD is one of the most important pieces of information used in decision making, transit modeling, and traffic analyses.

Metro Transit is among the pioneers using card data to study passenger patterns. Metro Transit reported about 55% of riders used electronic cards in 2016. The data improves planners’ ability to make good, system-level decisions.

For example, an on-board transit survey costs several million dollars to do. A consultant is hired to conduct interviews on the busses (asking about origins, destinations, routes, access, etc.). After one year, about 30,000 responses might be collected. Using card data only from UMN student riders, Khani has access to 37 million records.

University Riders

Khani and his team of researchers are using the Metro Transit database to examine transit patterns of UMN student riders. UMN students can purchase UPass which grants unlimited travel anywhere in the metro area for a semester (3-4 month time period). Metro Transit has been gathering rider data since the UPass was initiated in 2009.

“We will cluster riders into groups based on their usage and try to understand their behavior. Since we have only boarding information in most cases, we use chaining—a technique in transit research, to follow a rider in time to infer the complete tour, Khani explained. “We assume the first access point is somewhere near home and infer activities by the next time and location where the rider accessed the system. Sometimes the chain is incomplete—someone may take the bus in the morning and get a ride home in the afternoon. Our goal is to develop a more accurate OD matrix to enable better decisions about how to serve riders.”

Vehicle Data

Technology is also used to gather data on the transit vehicles themselves. Automatic Vehicle Location (AVL) devices placed on the transit vehicles track when they start, stop, move, or dwell at a stop. Khani is beginning a new project that will use AVL data to study the A Line, the Twin Cities’ first Bus Rapid Transit.

A Line

Khani’s project “After Study: Bus Rapid Transit 'A’ Line” is funded by Transitway Impact Research Program (TIRP). The goal is to find out how well the A Line BRT is working (it opened in June 2016) and how accurate projections were about its performance.

MnDOT, a TIRP stakeholder, is especially interested to learn how the A Line BRT impacts traffic flow on the roadways. The A Line is classified as an arterial BRT because it operates on an arterial road (Snelling Avenue) not a freeway. The A Line busses do not pull out of traffic when riders are getting on and off, so one lane of traffic is blocked when they stop, which could create a delay for other vehicles on the road. To reduce the time the bus is dwelling at a stop, riders are required to pay the fare before entering the bus.

Learn more about riding the A Line

AVL data has been collected on every bus, every day since the BRT began. Khani and his research team will analyze the large database and determine how much trip time is related to travel time, traffic delays, or dwell times (time at a BRT station).

Khani and his research team will also deploy video cameras at BRT stations to observe vehicles behind the BRT. They will be studying if cars line up or move to another lane and may study safety impacts.

Important for Agencies and for Residents

BRT and other higher-speed, higher-capacity transit opportunities require significant financial investment. Metro Transit wants to ensure that the money invested actually improves transit performance in the region. Khani’s study of the Twin Cities’ first BRT line will have importance for transit agencies, for riders, and for residents and businesses around the BRT. Khani’s co-investigator is Jason Cao from the Humphrey School of Public Affairs. Cao will be using data from a survey that Metro Transit conducted to examine user satisfaction with the BRT and perceptions of businesses and residents along the corridor. (Cao teaches urban planning. Many CEGE transit students take Cao’s class on transit planning.)

The Roads and Rails Ahead

There is a lot of speculation about what the future of transportation will look like. The rise of shared and autonomous vehicles promise change. Khani believes his research on transit networks will continue to be relevant however driving technologies change.

“Autonomous vehicles have advantages—they can be made shareable and affordable, drivers can use time more productively—but one problem autonomous vehicles do not solve is congestion. The ability to move large numbers of people is something that planners definitely want to include in the long term view. In dense urban areas, even with autonomous vehicles, there will still be a need for high-capacity transit services. We will still have transportation networks and will need to understand the impact of new technologies on those networks. Research on transit networks will be relevant long into our foreseeable future.”

Anna Rosin contributed to this story.