ITE Journal - May 2020 - 47

Based on the FHWA guidebook, a signalized RCI can handle
up to 25,000 VPD on the minor street.24
Two-lane minor streets should be signalized in RCIs at
demands ranging from 3,000 to 11,000 VPD based on
NCDOT research.25
Because RCIs have superior signal progression and are not as
vulnerable to driver confusion, MUTs and CFIs only become
feasible at minor street demands above 25,000 VPD.
Agencies often make exceptions to these rules, but they should
serve well to start.
One other technique needed to construct the SaFID tables is
the ability to chain CMFs. If we have a CMF for the conversion of
condition a to b, and a CMF for the conversion of b to c, multiplying
the CMF for a to b by the CMF for b to c should provide the CMF
for a to c without losing much accuracy. For example, for all crashes
the conversion of an intersection from a signal to a two-way stop
has an average CMF of 1/0.81 = 1.23 and from a two-way stop to
all-way stop has an average CMF of 0.32, so the CMF for converting
from a signal to an all-way stop should be 1.23 * 0.32 = 0.4.

SaFID Tables
Table 2 shows the SaFID table based on all crashes, while Table 3
shows the SaFID table based on injury crashes. Both tables show the
SaFID for any combination of major street and minor street number
of through lanes and vehicle demand. The demands are in terms of
average annual daily traffic, or AADT, in VPD. The CMFs displayed
are for the conversion of a conventional signalized intersection to
the design named in the cell.
Tables 2 and 3 are dominated by four design types, including
AWSC, roundabouts, RCIs, and MUTs. A TWSC is never the
generally safest choice. A conventional signal only shows up in
a small sliver of each table, at the highest demand levels handled
with two-lane major and minor streets where a roundabout is not
feasible. Though the partial CFI did not appear in Table 2, it did not
miss by much. The CMF for all crashes for a MUT is 0.85,20 while
the CMF for a partial CFI is 0.88.21 Especially at high demands with
six-lane or eight-lane roads, a partial CFI may be a worthy choice
for capacity without losing too much safety benefit.

Using the Tables
In view of the importance of safety, the author urges highway
agencies to adopt the SaFID as the default choice during intersection improvement projects. Conventional TWSC and signal
intersections are not generally the safest feasible options and should
therefore not be the default designs during projects. There are many
reasons why an agency may not be able to build the SaFID in any
particular project, including cost, impacts, delays, and effects on
non-motorized travelers. However, in all cases agencies ought to
be prepared to say why they did not end up building the SaFID.

Entering the project development process with the SaFID as the
default should shift the burden of proof to advocates of generally
less-safe designs, where the burden should lie.
Project teams can use Table 2 or Table 3 or both in their
processes. In some cells the tables have the same entry, indicating
that a design is generally safest using either all crashes or injury
crashes. However, in some cells the two tables have different entries,
so agencies and project teams will have to choose which type of
crash is more important.
One of the reasons not to choose the SaFID during a project is
that the research justifying the design as the safest does not apply
to the case in question. Indeed, there are many places where the
existing research reflected in Tables 2 and 3 is out of scope. Tables
2 and 3 apply to four-legged intersections, for example, and may
not apply to a project improving a three-legged junction. Those
claiming that their project site is out of scope of the research
underlying Tables 2 and 3 ought to be careful not to stretch that
argument too far. Just because the research has not been done on
three-legged RCIs, for example, does not mean that those designs
are less safe than conventional designs.
Project teams should use Tables 2 and 3 early in the process
in conjunction with a couple other tools that support quick
judgments on different designs. For capacity, CAP-X helps analysts
quickly judge which alternatives promise greater efficiency.22 For
the quality of pedestrian and bicyclist service at intersection alternatives, the guidebook from NCHRP project 7-25, to be published
in 2020, should provide a quick but helpful look at any intersection
design. Together, the SaFID tables, CAP-X, and the forthcoming
NCHRP 7-25 guidebook provide a powerful suite of early intersection design filters.

Follow-Up Work
Tables 2 and 3 will hopefully prove helpful to intersection design
teams, but they could be improved with several types of additional
research. First, more research on some designs already in the tables
would be welcome. Second, we could use research on designs not
in these tables, including offset and quadrant intersections, that
likely will be considered more in the next few years. Third, we need
research on the validity of combining CMFs, so that project teams
can evaluate the safety of hybrid designs. Fourth, we have no CMFs
and almost no safety research on grade-separated intersections
(the junction of two non-freeways using a bridge), while many
of these relatively high cost solutions are being proposed. Fifth,
researchers should begin to derive CMFs for interchanges so an
interchange SaFID table may be assembled. Finally, work should
continue on adequate crash surrogates to help designers estimate
the crash potential of designs that have not been built yet. A recent
paper based on research sponsored by the NCDOT (26) provided a
start in this direction. itej
w w w .i t e.or g

May 2020

http://www.ite.org

ITE Journal - May 2020

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