1. Impact of Intersection Angle on Safety
Daniel Carter
UNC Highway Safety Research Center
Presented at the Traffic Records Forum, August 6, 2017
New Orleans, LA

2. Definition
Intersection Angle

3. Current Design Guidance
Desire 90 degree intersections
Minimum critical angle – range of 60 to 75 degrees, depending on which policy/resource
AASHTO – 60
FHWA – 75
ITE – 60, 70, 75
Montana DOT – 60
Illinois DOT – 75
CalTrans - 75

4. What Do We Know? Very few crash-based studies
HSM recommendations from cross-sectional studies
Expert panel recommendation for further work
Visibility research suggest critical minimum angles greater than the current AASHTO policy
Older persons and visually-impaired pedestrians may be negatively impacted at skewed intersections

5. Research Objective
Derive quantitative relationships between intersection angle and safety
CMFs for intersection angle
Assessment of critical minimum angle
Revisions to design policies
Safety is measured in terms of crashes
The objective of this study is to derive quantitative relationships between intersection angle and safety. The relationships will be used to:
determine if there is a critical minimum angle at which safety is substantially diminished (75 degrees, 60 degrees, or other),
determine appropriate crash modification factors for reducing or eliminating the skew angle of an intersection, and
assess the need for revision of current geometric design policies and practices.

6. Methodology Goals Analysis approach
Before-after evaluation (not feasible)
Cross-sectional modeling
Large and diverse sample of intersections
Range of intersection angles
Urban and rural
3-leg and 4-leg

7. Data Acquisition Minnesota Supplemental data
Intersection (skew attribute)
Tee vs. Wye and Cross vs. Skew
Crash, roadway, traffic
Supplemental data
Verification of skew
Measured intersection angles
Other potential contributing data elements

8. Supplemental Data Collection
Angle measurements in GIS
123 57 55
125
Google Maps aerial & street view images – geometric/land use features

9. Minnesota Intersections
Angle Classification
Rural Intersections
Urban Intersections
Tot
No. %
Right Angle
1153
78.3
148
75.1
1301
Skewed
319
21.7 49
24.9
368
Total
1472
100.0
197
1669
4-leg
Angle Classification
Rural Intersections
Urban Intersections
Tot
No. %
Right Angle
710
80.0
136
61.5
846
Skewed
178
20.0 85
38.5
263
Total
888
100.0
221
1109
3-leg

10. Model Development Negative binomial models Modeling framework
State-of-the-Practice
Highway Safety Manual safety prediction models
Accounts for overdispersion (variance > mean)
Modeling framework
3-leg vs 4-leg intersections
Rural vs urban
Two-lane vs multilane
Total vs injury vs PDO

11. Model Development Model forms Model fit Base models
Flexible form models
Interaction model
All variables in final models are significant (p < 0.05)
Model fit
BIC selected as primary metric
AIC selected as secondary metric

12. CMF Development
CMFs derived from all models that included a significant intersection angle variable
Regression coefficients () applied to value of angle () for all significant angle terms
All CMF values normed to a nominal value of 1.0 for a 90 degree intersection

13. Results from MN Data were… counter to expectations
Results for skewed intersections < 40 degrees are based on a small sample of intersections

14. Expanded Study Wanted to verify results
Added data from another state - Ohio
Ohio DOT provided statewide intersection inventory, crash data, and traffic volume
Detailed intersection data included angles between approaches
Retained only stop-controlled intersections with 2, 4, or 6 lanes on the intersecting roads

15. Expanded Study Huge sample size available from Ohio
Number of Ohio Intersections in Study
Rural
Urban
Total
3 Leg
7,361
1,777
9,138
4 Leg
3,350
431
3,781

16. Results: Rural 4-Leg, Total
Parameter
Peak CMF Angle
Peak CMF Value
Cutoff Angle
Number of site-years MN 65
1.23 10
10157 OH
1.4 20
3335

17. Results: Rural 4-Leg, 2-Lane
Parameter
Peak CMF Angle
Peak CMF Value
Cutoff Angle
Number of site-years MN 60
1.28 10
9436 OH 55
1.13 20
3227

18. Results: Rural 4-Leg, Multilane
Parameter
Peak CMF Angle
Peak CMF Value
Cutoff Angle
Number of site-years MN 65
1.89 10
721 OH 70
1.34 50
103

19. Results: Rural 3-Leg, Total
Parameter
Peak CMF Angle
Peak CMF Value
Cutoff Angle
Number of site-years MN 65
1.09 10
6188 OH 55
1.27
6951

20. Results: Rural 3-Leg, 2-Lane
Parameter
Peak CMF Angle
Peak CMF Value
Cutoff Angle
Number of site-years MN - OH 40
1.21 10
6868

21. Results: Urban 3-Leg, 2-Lane
Parameter
Peak CMF Angle
Peak CMF Value
Cutoff Angle
Number of site-years MN - OH 30
1.4 10
1468

22. Results: Urban 4-Leg, 2-Lane
Parameter
Peak CMF Angle
Peak CMF Value
Cutoff Angle
Number of site-years MN - OH 50
1.23 30
373

23. Results: Other Multilane Models
Also developed models for:
Rural 3-Leg, Multilane
Urban 4-Leg, Multilane
Urban 3-Leg, Multilane
However, very little confidence in these models

24. Angle of Highest Priority
Results: Summary
Intersection Type
Angle of Highest Priority
Minnesota
Ohio
Rural 4-Leg Total 65
Rural 4-Leg, 2-Lane 60 55
Rural 4-Leg, Multilane 70
Rural 3-Leg Total
Rural 3-Leg, 2-Lane - 40
Urban 4-Leg, 2-Lane 50
Urban 3-Leg, 2-Lane 30
“Angle of Highest Priority” is where an agency will see the greatest benefit for realigning an intersection.

25. Where to Find These Results
Upcoming FHWA Final Report
HSIS Publication
After publication, Crash Modification Factors Clearinghouse (

26. Questions and Discussion