Investigating unintended effects of I-35W safety improvements
In 2009, the Minnesota Department of Transportation (MnDOT) began major safety improvements on I-35W as part of the Federal Highway Administration’s Urban Partnership Agreement (UPA). The aim was to reduce congestion on the Twin Cities’ highly traveled I-35W corridor between the Minnesota River and I-94. Among the improvements was the addition of a priced dynamic shoulder lane (PDSL) on parts of the 17-mile stretch of highway. However, following the opening of these improvements, the frequency of rear-end crashes increased in certain sections—especially in the PDSL regions. To determine the underlying causes of this increase, MnDOT enlisted the help of University of Minnesota researchers in the Department of Civil, Environmental, and Geo- Engineering.
“Our primary objective was to determine if these increases were direct effects of the improvements or if they were due to changes in the traffic conditions,” says Professor Gary Davis, the principal investigator of the study and an RSI researcher. “MnDOT was interested in extending some or all of these improvements to other corridors but needed to know what the safety impacts were to aid its decision making.”
Besides Davis, the research team included Jingru Gao, a master’s candidate advised by Davis, and John Hourdos, director of the U of M’s Minnesota Traffic Observatory.
The researchers began with a preliminary analysis to determine the study region and singled out two specific sections that were experiencing more rear-end crashes after the improvements. Next, these two sections were further divided, and data on crashes, traffic flows, weather conditions, and PDSL activation were compiled for each section. Researchers then analyzed the data to estimate the change in rear-end crash risk following the UPA project while controlling for those other factors.
The analysis resulted in several significant findings. First, researchers concluded that the operation of the PDSL had no direct effect on the likelihood of rear-end crashes. In addition, they found evidence that crashes were most likely to occur when lane occupancies were around 20 to 30 percent; crash likelihood tended to decrease for lane occupancies below and above this range. Finally, researchers concluded that the PDSL region experienced significantly more congestion following the UPA improvements because of the removal of a bottleneck in the old I-35W/Crosstown Commons, causing the bottleneck to move northward to the I-35W and I-94 junction. This resulted in an increase in higher-risk traffic conditions in the PDSL region. When controlling for this change in traffic conditions, they found no significant increase in rear-end crash risk attributable to the PDSL.
“The PDSL sections showed substantial increases in lane occupancy following the project,” Davis says. “The observed increases in rear-end crash frequency can be explained by increases in higher-risk traffic conditions—and the increase in higher-risk traffic conditions was most likely caused by the removal of the old Crosstown traffic bottleneck.”
Brian Kary, MnDOT freeway operations engineer, concurs. “The research study showed us that the design of the PDSL did not increase crashes on this stretch of 35W,” he says. “Had we not done the PDSL, congestion and ultimately crash rates would have been worse.”
In the future, the methodology demonstrated by this study could be used to evaluate how other freeway-related projects affect safety, Davis says, because the study worked out a way to estimate changes in hourly crash risk while controlling for variations in traffic conditions. In addition, the major finding—that the current implementation of a PDSL did not have an adverse effect on safety—can be used to provide guidance to MnDOT as it considers using PDSLs at other highway locations.