ITE Journal - May 2020 - 46

However, the library of safety research results is rarely used during
transportation improvement program (TIP) projects to improve
intersections. During those projects, teams typically use site crash
data to justify the project and then use physics-based nominal
design standards. Only in a few states with comprehensive intersection control evaluation (ICE) policies do project teams typically use
the available CMFs.
To help intersection project teams use the available CMFs more
often and effectively, the author assembled tables showing the safest
feasible intersection design (SaFID) for each combination of size
and demand on the major and minor streets. The tables should
be easy to use. Project teams should start their investigations of
alternatives with the design that the research shows to be the safest,
and then examine other factors that are meaningful in a design
decision. If project teams end up choosing an alternative that is not
the safest according to the research, they should have to document
why. Starting with consideration of the SaFID should mean that
agencies end up building safer intersections. The objectives of this
paper are to show the SaFID tables, provide background on how
they were developed, and discuss how they should be used.

Sources
With two notable exceptions, the CMFs in the SaFID tables are
from the Clearinghouse.1 The author used only CMFs with three
stars or better. The documentation in the Clearinghouse had to be
clear on the before condition, the after condition, and the context
in which the crash data were collected. This effort used CMFs
for a generic four-legged intersection. In some cases, the author
averaged CMFs to create an overall CMF. For example, if a study
had separate CMFs for urban roads and for rural roads, the author
averaged those to get an overall CMF for that study, and if two
studies provided CMFs for the same design the author averaged
those to get an overall CMF for that design.
Table 1 shows the references from the Clearinghouse used
to assemble the SaFID tables and the corresponding average
CMF values. Note that a reduced conflict intersection (RCI) is

also known as a restricted crossing U-turn (RCUT) intersection,
superstreet, or J-turn.
The author considered two sources of CMFs not in the
Clearinghouse in constructing the SaFID tables. First, the FHWA
median U-turn (MUT) guidebook (20) contains a review of the
safety research on that design and implies average CMFs for the
conversion from a conventional signal to a MUT of 0.85 for all
crashes and 0.7 for injury crashes. Second, a 2015 research report
for the Utah DOT (21) showed a CMF of 0.88 for all crashes
for the conversion of a conventional intersection to a partial
continuous flow intersection (CFI). The report did not provide
a result for injury crashes. The analysis looks to be of relatively
good quality. The partial CFIs examined in Utah had two left turn
crossovers at each site.
The available set of CMFs described above captures most
four-legged intersection designs used in the US as of 2020. In the
FHWA CAP-X tool, the only other four-legged intersection designs
listed are full CFI (four left turn crossovers), quadrant, bowtie,
and split.22 None of these is common. The only other common
intersection types that the author could think of are jughandle and
offset intersections. While jughandles are common in a few states, in
North Carolina they are not considered to be a competitive design as
they require more right-of-way than a partial CFI while delivering
only a fraction of the delay-saving benefits. Meanwhile, on offset
intersections a recent literature review conducted by the NCDOT
did not provide any studies with trustworthy CMF values, and the
Clearinghouse does not mention them. Overall, with the possible
exception of offset intersections, it looks like we have a pretty full set
of CMFs for common and feasible intersection designs.
To construct the SaFID tables, the author also considered the
feasibility of the various designs with the following rules:
All-way stop control (AWSC) is viable on two-lane roads with
demands less than 7,500 vehicles per day (VPD) on each road.
Based on the latest national guide, a single-lane roundabout
can handle up to 25,000 VPD total and a two-lane
roundabout can handle up to 45,000 VPD total.23

Table 1. CMF values and references.
Changing from...
Two-way stop control

Conventional signal

May 2020

Changing to...

All crashes

Injury crashes

Average CMF

References

Average CMF

References

All-way stop control

0.32

0.28

2, 3

Conventional signal

0.81

4-8

0.74

6-9

One-lane roundabout

0.51

10-13

0.16

Unsignalized RCI

0.58

14-16

0.42

14, 16

One-lane roundabout

0.74

0.45

Two-lane roundabout

0.89

12 and 17

0.54

17, 18

Signalized RCI

0.85

0.78

i te j o urnal

ITE Journal - May 2020

Table of Contents for the Digital Edition of ITE Journal - May 2020