BTS Navigation Bar

NTL Menu


Technical Assistance Report: Evaluation of the Use of Strobe Lights in the Red Lens of Traffic Signals





Click HERE for graphic.





                    TECHNICAL ASSISTANCE REPORT

      EVALUATION OF THE USE OF STROBE LIGHTS IN THE RED LENS
                        OF TRAFFIC SIGNALS

                        B. H. Cottrell, Jr.
                     Senior Research Scientist

             Virginia Transportation Research Council
   (A Cooperative Organization Sponsored Jointly by the Virginia
   Department of Transportation and the University of Virginia)

     In Cooperation with the U.S. Department of Transportation
                  Federal Highway Administration

                     Charlottesville, Virginia

                           November 1994
                           VTRC 95-TAR-5





                TRAFFIC RESEARCH ADVISORY COMMITTEE

    L. C. TAYLOR, Chairman, Salem District Traffic Engineer, VDOT
    B. H. COTTRELL, JR., Executive Secretary, Senior Research
          Scientist, VTRC
    M. G. ALDERMAN, Traffic Services Manager, VDOT - Traffic
          Engineering
    J. R. BROWN, Bowling Green Resident Engineer, VDOT
    J. L. BUTNER, Traffic Engineering Division Administrator, VDOT
    J.    CHU, Northern Virginia District Freeway Traffic
          Operations Engineer, VDOT TMS Center NOVA
    C. A. CLAYTON, Transportation Engineer Program Supervisor,
          VDOT-Traffic Engineering
    D. E. COLE, Bristol District Traffic Engineer, VDOT
    J. C. DUFRESNE, Traffic Engineering Division, VDOT
    Q. D. ELLIOTT, Williamsburg Resident Engineer, VDOT
    C. F. GEE, State Construction Engineer, VDOT
    S. D. HANSHAW, Suffolk District Freeway Traffic Operations
          Engineer, VDOT
    J. T. HARRIS, Transportation Engineer Program Supervisor, VDOT-
          Location and Design
    K. J. JENNINGS, Truck Weight Program Manager, VDOT-Maintenance
          Division
    T. A. JENNINGS, Transportation Systems Management Engineer,
          Federal Highway Administration
    T. W. NEAL, JR., Chief Chemist, VDOT - Materials Division
    R. L. SAUVAGER, Assistant Urban Division Administrator, VDOT
    D. H. WELLS, Principal Transportation Engineer, Transportation
          Planning Division
    K. W. WESTER, Northern Virginia District Operations Engineer,
          VDOT
    W. W. WHITE, Suffolk District Operations Engineer, VDOT

                                ii





                             ABSTRACT

     The objective of this study was to evaluate the effectiveness
of using strobe lights in the red lens of traffic signals and, if
appropriate, to recommend guidelines for their use. Strobe lights
are used as a supplement to the red lens to draw the attention of
drivers to a traffic signal. VDOT has 22 intersections with strobe
lights in six of its nine districts; this is up from 3
intersections in 1987. Nineteen of these are in the western part of
the state, and most have the strobe light in the red signal over
the left through lane. The Barlo strobe light, a horizontal bar
positioned across the middle of the red lens with about 60 flashes
of white light per minute, is used at all 22 intersections. The
cost of a red signal head with a Barlo strobe light is about $765.

     Strobe lights are used by VDOT primarily for (1) areas with a
high truck volume and high speed; (2) areas with a high accident
rate; (3) areas with road geometries, especially grades
(downgrade), horizontal curves, and other features, that result in
limited sight distance; and (4) isolated intersections where a
signal is unexpected.

     Based on the analyses, there was no evidence indicating that
strobe lights are consistently effective in reducing accidents. The
limitations of the analyses are identified in the study. There is
no basis for recommending the use of strobe lights unless there are
other bona fide measures of effectiveness that can be used to
justify installing them.

                                111








      EVALUATION OF THE USE OF STROBE LIGHTS IN THE RED LENS
                        OF TRAFFIC SIGNALS

                        B. H. Cottrell, Jr.
                     Senior Research Scientist

                           INTRODUCTION

     Strobe lights have been used as a supplement to the red lens
to draw the attention of drivers to a traffic signal. They have
been used in situations where a signal is unexpected, a signal may
be difficult to see, or there is a high accident rate or a
potentially hazardous intersection. Specific applications of strobe
lights include the following:

     1.   isolated, high-speed, rural intersections

     2.   the first signalized intersection into an urbanized area
          after an extended road section without a signal

     3.   the first signalized intersection after a transition from
          a grade-separated or limited access highway to an at-
          grade highway with intersections

     4.   locations where background lighting and signs (visual
          noise) are a problem.

     Since the applications of strobe lights in the red indication
are limited in the United States, few studies have evaluated their
effectiveness. The study results that are available are
inconsistent and inconclusive In many cases, especially in North
Carolina, strobe lights are included among multiple safety
improvements at intersections, making it impossible to determine
their effectiveness. A similar situation exists at new signal
installations.

     In June 1987, 3 intersections in two Virginia Department of
Transportation (VDOT) districts had strobe lights. In April 1994,
22 intersections in six VDOT districts had them. Apparently, the
interest and popularity of strobe lights have increased in
Virginia. The Barlo strobe light, a horizontal bar positioned
across the middle of the red lens with about 60 flashes of white
light per minute, is used at all of the intersections. A red signal
head with a Barlo strobe light costs about $765.

     According to the Federal Highway Administration (FHWA),2
there is insufficient evidence to support the inclusion of strobe
lights in the Manual of Uniform Traffic Control Devices (MUTCD).3
Other concerns regarding strobe light usage expressed by traffic
engineers include whether they distract the driver's attention from
other traffic control devices and other vehicles and whether their
attention-getting value diminishes with time. Thus, VDOT's Traffic





Engineering Division requested a scientific evaluation of the use
of strobe lights in the red lens of traffic signals.


                       OBJECTIVES AND SCOPE

     The objectives of this study were to evaluate the
effectiveness of the use of strobe lights in the red lens of
traffic signals and, if appropriate, to recommend guidelines for
their application. Only applications of the Barlo strobe light were
studied.


                              METHODS

     Two activities were conducted to accomplish the study
objectives:

     1.   Data collection. A questionnaire survey was sent to the
          nine VDOT district traffic engineers (DTEs) to obtain
          information on strobe light use. The questionnaire sought
          an inventory of strobe light installations, maintenance
          experiences, and reasons for installing the strobe.
          Accident data were collected for selected sites.

     2.   Data analysis and evaluation. The questionnaire
          information was summarized and analyzed to assess the use
          and performance of strobe lights, and accident analyses
          were conducted. Of the 22 intersections that had strobe
          lights, 6 of them had 3 years of associated accident data
          before and after the strobe lights were installed. The
          accident analysis was performed in three parts: a review
          of accident trends at the six sites; a before-and-after
          analysis; and a before-and-after analysis with a
          comparison group and a check for comparability for three
          of the six sites for which a comparison site was
          identified. Rear-end, angle, and total accidents that
          involved at least one vehicle on the strobe light
          approaches were examined.


                              RESULTS

    Results from the Survey of VDOT District Traffic Engineers

     The results of the survey are shown in Table 1. Nineteen of
the 22 strobe lights are in the western part of the state, with 8
in Salem, 7 in Bristol, and 4 in Staunton.

                                 2






Click HERE for graphic.


                                 3





Click HERE for graphic.


                                 4





Click HERE for graphic.


                                 5





Location of Strobe Light

     At 17 intersections, the strobe light is in the red signal
indicator over the left through lane. This location was selected
based on the notion that the strobe light would be detected at a
greater distance in the left through lane signal because horizontal
curves and possibly overhanging foliage on the right shoulder might
block the view of the right through lane signal.

     At the 4 intersections in the Staunton District, strobe lights
are in the red signal indicator over both through lanes. It is
suspected that two strobe lights were used to enhance the
visibility of these devices.

     In Suffolk, the strobe light is in a separate red signal head
next to the traffic signal over the right lane, a configuration
that Suffolk District staff had observed in North Carolina and the
City of Virginia Beach.

     At 17 intersections, strobe lights were installed on both
directions of the major roadway; at 4 intersections, the devices
were needed in only one direction. Strobe lights were installed on
all approaches at only 1 intersection.

     At 9 intersections, strobe lights were installed with a new
traffic signal; at 7, they were installed less than 12 months after
a signal installation. Of the remaining 6 intersections, 5 had a
traffic signal in place at least 3 years before the strobe light
was installed.


Reasons for Installation

     Reasons for installing the strobe lights included one or more
of the following: (1) high truck volumes and high speed; (2) a high
accident rate; (3) road geometries, especially grades (downgrade),
horizontal curves, and other features, that resulted in limited
sight distance; and (4) an isolated intersection where a signal was
unexpected.


Maintenance

     Although maintenance of the strobe lights has generally not
been a problem, certain districts reported that they had problems
initially. In one district, the strobe light exploded because it
failed to release stored energy; this was solved by using a strobe
from a different manufacturer. In another district, there were a
number of failures with the control circuits, which decreased after
several discussions with the supplier and manufacturer. In a third
district, a transformer exploded when the side-street strobe lights
were flashing; these strobes were removed. One district noted that
any malfunction was usually due to the power pack, which is
relatively expensive (about $110).

                                 6





     The annual preventive maintenance routine typically includes
cleaning the explosion guards, lenses, and reflector; replacing the
incandescent lamp; and inspecting the seals for leaks that could
result in water damage. Extreme care must be exercised because of
the very high voltage in and around the strobe apparatus.


Indications for Use of Strobe Lights

     When asked where strobe lights should be used, the DTE's
responses were (1) at isolated intersections, (2) at locations with
limited sight distance, (3) at high-speed locations, and (4) at
locations with poor alignment, curves, and/or grades. A potential
concern was that many requests would be made for strobe lights.
However, the district offices have received very few. During the
last 3 years, there have been only three requests statewide.

     Respondents noted that potential uses of strobe lights include
(1) with a hazardous indication beacon for a warning sign, (2) in
school flashing lights, and (3) with emergency vehicles. In the
Salem District, a strobe light in an amber lens was installed above
a sign warning of "trucks crossing highway 800 ft" near a track
stop on Route 220 in Franklin County. There is limited sight
distance (600 ft) southbound because of an upgrade.


Other Expressed Concerns

     The Bristol District requested guidelines on when to remove
strobe lights. There was some concern about liability in the event
of an accident following removal.

     Based on comments from some DTE staff, in some areas,
motorists and VDOT personnel perceive strobe lights as effective in
improving safety at an intersection. For example, DTE staff from
one district commented that the strobe lights were especially
useful at dawn and dusk and whenever visibility is poor.


                 Accident Analysis and Evaluation

Review of Accident Trends

     Traffic and geometric data are shown in Table 2. All of the
sites have a four-lane divided highway as the main approach.

     Before-and-after accident data are shown in Table 3. For rear-
end accidents, there was no change at four sites and an increase of
100 percent or more at two sites. At three sites, angle accidents
decreased between 38 and 75 percent; at two sites, they increased
25 and 400 percent, respectively; and there was no change at one
site. For total accidents, two sites each had no

                                 7








Click HERE for graphic.


                                 8





Click HERE for graphic.


                                 9





change, an increase of at least 80 percent, and a decrease of at
least 25 percent, respectively. The accident experience at the
Route 7 site was identical for both periods, and the number of
accidents was the lowest of all the sites. Based on a review of the
data trends for all six sites, there was no evidence to indicate
that the strobe light had an impact on the incidence of accidents.

     Three sites have one strobe light per approach, and three have
two. Both sites where rear-end and angle accidents increased had
one strobe light per approach. For total accidents, accidents
increased more than 80 percent at two sites with one strobe light
and decreased at least 25 percent at two sites with two lights per
approach.

     Although the sample size was too small to allow definitive
conclusions, the use of two strobe lights per approach appears to
be more effective than the use of one per approach. It seems
logical that if two strobe lights reduced accidents then one would
also reduce accidents, but possibly to a lesser degree. Because
accidents tended to increase at sites with one strobe light,
factors other than the strobe lights may have contributed to the
increase. Thus, the study sites were further examined to determine
what factors other than the strobe lights might have contributed to
the accident experience. The following factors were considered:

     -    The intersection of Routes 460/220 ALT is in an
          industrial area with a high volume of trucks. The side
          street, which is a primary arterial, is being widened
          from two to four lanes. Subsequently, the intersection
          will be rebuilt to include dual left-turn lanes from the
          mainline in one direction and a sweeping right-turn lane
          in the opposing direction. The additional capacity should
          help reduce the number of accidents.

     -    The intersection of Routes 522/739 S was once the
          location of the first signal inbound near a town. Around
          the time the strobe light was installed, a new signal was
          installed about 670 m (2,200 ft) in advance of Site 4;
          therefore, it is no longer the first signal encountered.
          A right-rum lane was also added to the mainline.

     -    The intersection of Routes 660/719 has one leg of the
          side street for access to a shopping mall with heavy
          traffic, and the opposing side street approach has light
          traffic. In May 1994, two traffic signal changes were
          made to improve operations and safety at the
          intersection: the exclusive permissive left-turn signal
          phasing on the mainline was replaced with an exclusive
          left-turn phase, and a shared phase was replaced by split
          phases for the two side street approaches.

     -    The remaining sites are in rural areas with no
          distinguishing features.

It thus remains unclear whether there is a benefit for installing
two strobe lights per approach compared to one.

                                10





Click HERE for graphic.


                                11





     A futile attempt was made to determine if there were any
similar characteristics for sites with similar before and after
accident trends.


Before-and-After Accident Analysis Using the Z Test

     Comparison sites were identified for three of the six study
sites, and a before-and-after accident analysis using the Z test
was performed. The formula used for the Z test was:

                        Z = A - B/(A + B)

where A is the after accident frequency and B is the before
accident frequency. For a level of significance of 0.05,

                         Z0.05 =  -1.96.

     The results for rear-end, angle, and total accidents are shown
in Table 4. For rear-end and angle accidents, there was no
statistical evidence to indicate that the strobe light had an
effect. In fact, for total accidents at two of the six sites, there
was statistical evidence that the strobe light had a harmful
effect. It was unclear why accidents increased significantly, but
it is suspected that factors other than the strobe light were
responsible. This demonstrates one of the shortcomings of a before-
and-after accident analysis; it is assumed that the treatment was
the cause of any change in accidents.


Before-and-After Analysis with a Comparison Group and a Check for
Comparability

     This evaluation design was used because it is more rigorous
than a before-and-after design with a comparison group. The use of
multiple before-and-after readings allows some relief from the
regression toward the mean fallacy. The analysis procedure is
described in detail in a report by Griffin4 and in an evaluation
of wide edgelines by the author.5 It is described briefly here.

     Step 1. Check for comparability. If the rate of change in the
     frequency of accidents (expressed as natural logarithms) does
     not deviate by more than chance expectation during the before
     and after periods, then the comparison group is appropriate.
     If the treatment and comparison groups are not comparable,
     then further analysis is terminated.

     Step 2. Collapse the treatment and comparison groups across
     the before and after periods. If the rate of change in the
     frequency of accidents for the treatment and comparison groups
     does not deviate by more than chance expectation, then there
     is no evidence that the treatment affected the incidence of
     accidents. If the rate of change in the comparison group is
     more negative (or more positive), then the treatment was
     beneficial (or harmful).

                                12





These two steps are performed using the likelihood ratio chi-square
test. A2 x n (n is the total number of years of data) is
partitioned into two parts: chi-square comparability and chi-square
treatment.

The three treatment and comparison sites were as follows:

                 Treatment             Comparison

          1. Routes 301/206           Routes 301/218
          2. Routes 522/739 N         Routes 522/739 S
          3. Routes 17, 50/340        Routes 522/340


Each treatment and comparison site share a common mute and
therefore experience similar traffic on at least one of the
intersecting routes.

     In the analysis of rear-end accidents, the after periods of
the treatment and comparison sites were not comparable for site 1
and the before periods were not comparable for site 2. For site 3,
the sites were comparable, and there was no statistical evidence
that the strobe light was effective in reducing rear-end accidents.

     For angle accidents, sites 2 and 3 and their respective
comparison site were both comparable. Neither the before nor after
period was comparable at site I. For sites 2 and 3, there was no
statistical evidence that the treatment was effective in reducing
angle accidents.

     The results of the total accident analysis were similar to
those of the angle accident analysis; sites 2 and 3 and their
respective comparison site were comparable, but there was no
statistical evidence that the strobe light was effective in
reducing accidents. For site 1, the after periods were not
comparable.


Limitations of the Analysis

     Strobe lights flash only when the red signals in which they
are housed are on. Ideally, there should have been ways to ensure
that the accidents under review occurred when the strobe light was
flashing. Unfortunately, there is no reliable item on the standard
police accident report form that provides such information. The
item "driver action," which includes "disregard for the stop-go
signal/ran the red light," is potentially useful. However, the
majority of accidents under review had driver inattention, a catch-
all description with little value, as the driver action. Although
copies of the actual accident reports completed by the police would
have been helpful in determining driver action, the reports for
most of the accidents were more than 5 years old and not readily
available.

                                13





     Strobe lights were installed at locations with potential
safety problems. These sites may have a propensity for higher than
normal accident rates. The fact that the strobe light sites were
selected for an atypical condition or their accident experience,
rather than randomly, may explain in part why the treatment and
comparison sites were not comparable in several instances.

     Although the statistical analyses were presented, the power of
the test is low because of the relatively small number of accidents
at the test sites. Consequently, the results of the trend analysis
were the basis for the conclusions.


                        Summary of Findings

     Based on these analyses, it is concluded that there is no
statistical evidence to suggest that the strobe lights were
effective in reducing accidents. Based on the annual accident
trends at the three treatment sites, there is no evidence that the
effectiveness of the strobe lights diminished over time. Given the
lack of evidence that strobe lights were effective, the absence of
a trend of diminishing effectiveness is not surprising.


                            DISCUSSION

              Driver Noncompliance and Strobe Lights

     At the October 1993 meeting of the Traffic Research Advisory
Committee, the top research priority identified by attendees was
noncompliance with traffic control devices. It was noted that
driver noncompliance, such as the running of red traffic signals,
is increasingly common. Perhaps this type of behavior may be one
factor that contributes to the lack of effectiveness of strobe
lights.


                   Alternatives to Strobe Lights

     Of the four reasons cited in the Introduction for installing
strobe lights, three involve conditions where a traffic signal may
not be expected. Such instances include an isolated rural
intersection and the first signal after an extended road section
without traffic signals. A method cited in the MUTCD to alert
motorists to a traffic signal ahead is the use of a signal ahead
warning sign (W3-3). Hazard identification beacons (flashing yellow
lights) should supplement the sign to increase its attention-
getting value. An alternative to the fourth application of strobe
lights, conditions where visual noise is a problem, is the use of a
back-plate to increase the signal target value.

                                14





     One of the four reasons for installing strobe lights cited by
the DTEs was road geometrics that limit sight distance. One
alternative, the use of "prepare to stop when flashing" warning
signs, is being tested in the Salem District.


                  CONCLUSIONS AND RECOMMENDATIONS

     1.   The results do not provide a basis for recommending the
          use of strobe lights. However, if other bona fide
          measures of effectiveness can be used to justify
          installing strobe lights, then they should be considered
          for use.

     2.   With regard to driver noncompliance, it is probable that
          willful, defiant behavior is not likely to be affected by
          the presence of a strobe light.

     3.   The conclusion that there is no evidence that strobe
          lights are consistently effective in reducing accidents
          can be used as justification for their removal.

     4.   VDOT's current focus on providing customer service and
          being responsive to the citizens of the Commonwealth may
          provide some impetus to continue to install strobe
          lights. One approach is to install devices requested by
          the citizenry if the devices do no harm. Engineering
          judgment is another factor that may cause the
          installation of additional strobe lights despite the
          findings of this study.


                            REFERENCES

     1.   West Virginia University, Department of Civil
          Engineering. Accident Countermeasures at High-Speed
          Signalized Intersections: Phase 1--Synthesis of Practice.
          Morgantown, 1984.

     2.   Federal Highway Administration. Manual on Uniform Traffic
          Control Devices. Washington, D.C., 1988.

     3.   Virginia Department of Highways and Transportation,
          Traffic Engineering Division. Correspondence to district
          engineers on strobe lights on traffic signals. Richmond,
          June 30, 1985.

     4.   Griffin, L.I. III. Three Procedures for Evaluating
          Highway Safety Improvement Programs. Texas Transportation
          Institute, College Station, 1982.

     5.   Cottrell, B.H., Jr. Evaluation of Wide Edgelines on Two-
          Lane Rural Roads. Virginia Transportation Research
          Council, Charlottesville, 1987.

                                15





                          ACKNOWLEDGMENTS

     The author thanks the following for their contributions toward
this study: Jan Kennedy for data analysis; the DTE staff that
responded to the survey or otherwise provided information; Mike
Perlater for reviewing the draft report; and Linda Evans for report
editing.

     This technical assistance effort was funded with state
planning research funds through the Federal Highway Administration.

                                16




(SLR.html)
Jump To Top