Safety Evaluation of Corner Clearance at Signalized Intersections

CHAPTER 10. DISCUSSION

The CMFs for limited corner clearance on the receiving corners were consistent with expectation, indicating statistically significant increases in total, fatal and injury, rear-end, sideswipe, right-angle, and nighttime crashes. For limited corner clearance on the approach corners, the CMFs were counterintuitive, indicating statistically significant decreases in total, fatal and injury, and rear-end crashes. Intuition and past research suggest that limiting corner clearance (i.e., allowing driveways) on all corners would negatively affect safety due to complex and conflicting turning movements from the traffic into and, particularly, out of driveways in proximity to the functional area of the intersection. However, these particular CMFs in question (i.e., decreases in total, fatal and injury, and rear-end crashes for limited corner clearance on the approach corners) are among the most statistically significant results derived from this evaluation. The research team proposes a number of possible explanations for these results that are counter to the general hypothesis of the study.

As shown in table 3, rear-end crashes constitute more than half of all crashes, while angle crashes account for approximately one-quarter of all crashes. The reduction in rear-end crashes likely outweighs the increase in angle crashes and leads to the overall reduction in total crashes and fatal and injury crashes for this situation. Therefore, this discussion focuses on rear-end and angle crashes. The research team proposed the following potential hypotheses:

• The reduction in rear-end crashes on the approach corners may be associated with localized congestion from vehicles turning into and out of the driveways near the approach corners of an intersection. The vehicles turning into and out of driveways may lead to an increase in driveway-related angle crashes, as the CMFs indicate, although not with statistical significance. However, this reduction in operating speeds results in fewer rear-end crashes and likely fewer angle crashes within the adjacent signalized intersection, which tend to be more severe than driveway-related crashes. With a much higher proportion in overall crashes, the decrease in rear-end crashes is likely to be larger than any increase in angle collisions. This results in an overall reduction in total and fatal and injury crashes. The statistically significant driveway density coefficient in the model for rear-end crashes shown in table 6 seems to support this hypothesis.
  
  
• After passing through the signalized intersection, vehicles may accelerate. The interactions and conflicts from the turning vehicles (into and out of the driveways) on the receiving corners are likely to result in more crashes for all crash types. The turning vehicles from the cross streets also add to the overall traffic and likelihood of conflicts on the receiving corners. The mainline AADT on the receiving corners may not reflect this added traffic from the cross street and therefore is not captured in the model.
  
  
• The overall context of the sites with limited corner clearance is responsible for the difference rather than the specific effects of corner clearance. The limited corner clearance could be a surrogate for another factor that affects safety performance that is not captured in the models. That is, those intersections with more driveways on the approaches may have more traffic and are more likely to be congested than those without driveways on the approach simply by the nature of the roadway, not because of the presence of the driveway (e.g., stores and gas stations are there to serve the heavier traffic). The context of the intersection within the corridor is difficult to control for in a cross-sectional evaluation. In this study, the research team collected and analyzed corridor characteristic data elements, including driveway density (number of driveways per mile) and type of land use (residential, commercial, or mixed-use). The model estimation results suggest limited or no statistically significant effects of these elements on crashes. The evaluation set out to investigate the safety effects at intersections rather than the entire corridor and, as such, could not collect and include more corridor-related characteristics in the models or examine the effects on crashes along the related corridors.

Future research could explore the hypotheses proposed and discussed in this study. Crash prediction models that include operations-related factors—such as mean operating speeds, a speed profile for intersections along the mainline, or level of service—would greatly improve the results in determining the safety effects of corner clearance. Controlling for these types of factors may better explain the effects of corner clearance on rear-end and angle crashes and, therefore, on total and fatal and injury crashes. Future research could also verify the results using data from other States. The results presented in this study are based on data from the State of California and the City of Charlotte, North Carolina.

Readers may be able to test the hypotheses anecdotally as well. If a comparison of intersections in a jurisdiction shows that intersections with limited corner clearance are located along more congested corridors and have similar crash type distributions to the sample intersections in this study, then the reduction in rear-end crashes due to limited corner clearance on the approach is probably a result of the area type rather than the corner clearance. Therefore, improving corner clearance on mainline approaches may be less likely to increase rear-end crashes as a result. If the area type and crash type distribution do not follow with this hypothesis and the sample data, the results of this evaluation may not be as accurate when applied to those sites.

Additionally, the sample intersections used in this evaluation were not selected as a result of safety concerns due to angle crashes. In practice, potential projects are more likely to address corner clearance at intersections with a higher proportion of angle or turning crashes than represented in this study. Consequently, projects addressing approach corners may have a higher chance of reducing total crashes and yielding a higher net benefit when improving corner clearance than implied in the results of this evaluation.