Spring 2019 Vol. 6, No. 1
Better prediction of train arrival times promises safety benefits
Photo: Shutterstock
A new model developed by Roadway Safety Institute researchers can more accurately predict train arrival times at railroad crossings—a finding with both safety and congestion-relieving benefits. Incidents at highway-rail grade crossings are a serious safety challenge: in 2015, there were nearly 3,000 collisions between vehicles and trains at grade crossings, resulting in 230 fatalities. In addition, lengthy train crossings can cause congestion on surface streets, leading to delays and blocked roads for emergency vehicles.
"Blocked crossings and the resulting surface street traffic delays are major concerns for emergency service vehicles as well as the general public,” says Daniel Work, an associate professor of civil and environmental engineering at Vanderbilt University and the study’s principal investigator. Effective management of emergency response resources on the road network requires knowledge of when trains will arrive at grade crossings. Trains blocking the grade crossings temporarily prevent emergency vehicles from accessing parts of the community they serve, Work says.
Will Barbour at a CSX locomotive shop in Jacksonville, Florida.
To combat this issue, Work, along with Vanderbilt graduate student Will Barbour, sought to generate accurate estimated times of arrival for trains at grade crossings over a long time horizon, which could then be used to proactively address surface transportation safety, manage emergency response, and support in-vehicle driver alerts on personal navigation devices.
“This is a challenging problem because the variability of travel times on the United States freight rail network is high due to large network demands relative to infrastructure capacity,” Work says. To address this challenge, he says the project became the first to look at the potential for high-fidelity freight rail data to be used to predict arrival times.
Dan Work
The advanced model researchers developed used a number of inputs, including train-positioning information, properties of the train, properties of the network, and the properties of potentially conflicting traffic on the network. The work was conducted in two phases. First, the researchers developed a historical algorithm to accurately model delays using train-positioning information. Next, they integrated real-time train position information into the forecasts and used data shared by passenger rail service provider Amtrak and Class I freight railroad CSX Transportation to test and validate the proposed algorithms.
Using the new model, researchers found that estimated time of arrivals at control points located close to grade crossings were dramatically improved—particularly for predictions made multiple hours from a crossing, which are valuable for proactive safety measures such as early warning applications and emergency response management systems. In the future, Work says the research team hopes to continue refining the model by incorporating additional inputs—such as weather, crew changes, route topography, and events in which trains meet or pass one another—to increase prediction accuracy.