Académique Documents
Professionnel Documents
Culture Documents
Please refer to this document as follows: Hesjevoll, I.S. (2017), 2+1 roads, European Road Safety
Decision Support System, developed by the H2020 project SafetyCube. Retrieved from
www.roadsafety-dss.eu on DD MM YYYY
Please note: The studies included in this synopsis were selected from those identified by a
systematic literature search of specific databases (see supporting document). The main criterion for
inclusion of studies in this synopsis and the DSS was that each study provides a quantitative effect
estimate, preferably on the number or severity of crashes or otherwise on road user behaviour that is
known to be related to the occurrence or severity of a crash. Therefore, key studies providing
qualitative information might not be included in this synopsis.
2+1 roads
1 Summary
Hesjevoll, I.S., April 2017.
1.2 KEYWORDS
2+1 roads; median barrier; rural road
1.3 ABSTRACT
This concerns implementing a 2+1 road design on previously two-lane roads, i.e. a road design with
three lanes, where the middle lane alternates as a passing lane for the two opposing directions. 2+1
roads with median barrier (cable) are found to reduce the rate of severe and fatal injuries by about
51-63 %, depending on the road type and speed limit. However, the effect on less severe injury rates
is smaller and in some cases not statistically significant. This finding occurs when the measure is
implemented at previously wide two-lane roads (13 m) and when narrow roads (9m) are widened at
certain sections to allow for the alternating passing lanes. Implementation of 2+1 roads without a
median barrier appears to reduce the number of crashes, but this might vary between road types.
The reviewed studies are limited in number and most do not control for relevant confounding
factors, so the results should be interpreted with care.
1.4 BACKGROUND
1.4.1 What are 2+1 roads?
A 2+1 road consists of three driving lanes, two in one direction and one in the opposite direction. The
middle lane alternates between one direction and the other (typically every 1-2.5 km), allowing for
alternate opportunities for overtaking (see Figure 1). The opposing traffic flows are often separated
by a median barrier, but in some cases, 2+1 roads are divided by means of road markings only. The
design is typically implemented on wide two-lane roads (13 m, mostly rural motorways), rendering
narrower shoulders and/or driving lanes. Alternatively, 2+1 roads have been implemented on narrow
roads (9m) where the road has been widened at certain sections to allow for the alternating passing
lanes.
2
2+1 roads
3
2+1 roads
4
2+1 roads
2 Scientific overview
In the Polish context, 2+1 roads with no median barrier were implemented primarily to improve
traffic flow.
On the regular/wide 2+1 roads, the share of passing lanes is approximately 40 %, as opposed to
15-30 %, on narrow 2+1 roads.
A 2+1 design without a median barrier (i.e. where the opposing directions are separated by means
of road markings only) was used in the Polish study, and is common in several countries, including
Germany and Denmark. This design is often primarily thought of as a traffic flow measure.
Some studies that were not coded for the DSS, either due to methodological issues or because the
results were not codable, indicate that this design has positive safety effects, at least under certain
conditions. For instance, a German study (Weber & Löhe, 2003) evaluated the safety effects of a 2+1
road design implemented at different roads, and found that for roads with 10.000-16.500 vehicles
per 24h, implementation of 2+1 roads did not have negative safety effects even when bicyclists and
pedestrians were allowed on the roads. A Danish evaluation study (Buch, 2015) found that the effect
of 2+1 design on accident rates and injury rates varied between road types. For instance, 2+1
expressways had higher accident rates than expressways without the 2+1 design, but beneficial
effects of the 2+1 design were found on conventional rural roads with no or grade separated
junctions.
5
2+1 roads
In the evaluation of narrow 2+1 roads (Vadeby et al., 2015), it was found that rut depth growth was
consistently higher than for conventional roads. Both the Annual average daily traffic (AADT) and
the share of HGVs were important factors in explaining rut depth growth. For AADTs under 8000
vehicles, rut depth growth was 10-15 % higher than for conventional roads, and for AADTs of 8000
and above, the difference amounted to around 25 %. The authors note that the share of HGVs is
important in considering the construction of a narrow 2+1 road. They also recommend that the
share of HGVs is considered when determining the length of overtaking lanes (Vadeby et al., 2015).
The study of Cafiso et al. (2015) did a before-after study, with 4 years of traffic and crash data for
each period (2005-2008 and 2010-2013), comparing road sections where a 2+1 design were
implemented with comparable rural roads without such a design. The counts of injury and fatal
crashes (combined) for the treatment section was compared with the expected crash count, which
was estimated from the control sections in an Empirical Bayes analysis. The analysis controlled for
several confounding factors such as traffic flow, time trend, number of access points, and length of
the road sections.
In the two Swedish evaluation studies, the safety impact of implementing 2+1 roads was estimated
based on injury rates (number of injuries per million axle pair kilometres) and FSI rates (number of
fatal or severe injuries per million axle pair kilometres). In both studies, the control condition to
which the 2+1 converted road sections are compared, is a national “normal quota” of accident rates
6
2+1 roads
for similar road types that were not converted to 2+1 roads. In the study of Carlsson (2009), these
normal quotas are based on 13m roads from 1993-2005, and are used to predict the expected
accident rates had the reconstruction not been implemented. It is not clear whether the investigated
road segments were also included in the estimation of the normal quota before their conversion to
2+1 roads. More generally, limited detail is provided about the evaluation in this study, which
confounds the interpretation of the results.
Carlsson, June 1998-July 58 rural road 13-14 m wide, Shoulders and/or lanes
2009, 2006 (Dec 2007) segments, total speed limits of narrowed to allow for 2+1
Sweden length about 90-110 km/h. design with median
1750 km, barrier. Additional
approximately measures implemented to
70 % of Swedish varying degrees. Speed
2+1 roads as of limits mostly unchanged.
2007.
Cafiso et 2005-2013, 16 rural road 12,5 m wide, Shoulders and lanes are
al., 2015, excluding segments, speed limit 90 narrowed and the road is
Poland implementation where 2+1 km/h for widened (to 13,5 m) where
period (1 year) design was passenger the 2+1 design is
implemented on cars, 70 km/h implemented. Speed limit
13 km (excluding for heavy appears to be unchanged,
1+1 lane sections vehicles no median barrier. No
between passing additional measures are
lanes) mentioned.
Carlsson (2009) notes that upon reconstructing existing roads into 2+1 roads, many other road
improvements were made to different, but in many cases, extensive degrees. This included
improving and increasing (oncoming) ramp length, improvements on the side of the road,
improvements in paving and road making, and improving the operating standard especially in
relation to winter maintenance. Additionally, for the 2+1 road type with junctions, junction layout
was improved, as were the access points, and the number of access points was reduced. The issue of
additional road reconstruction activities implemented in conjunction with 2+1 roads is not
mentioned by Vadeby et al. (2015), rendering it unclear to what extent this might be a confounding
factor.
7
2+1 roads
The study of Vadeby and colleagues (2015) is a before-after study, with the same control condition
(normal quotas) as the study by Carlsson (2009). In the evaluation by Vadeby and colleagues (2015),
general trends in road safety are accounted for. Regression to the mean is accounted for in cases
when accident quotas deviated from the “normal quota”.
2.2.2 How well has the effect of 2+1 roads been studied?
Two studies (both from Sweden) were reviewed for 2+1 roads with median barriers. In at least one
of the studies, implementation of the 2+1 road design was accompanied by various other road
reconstruction measures. Hence, the results should be interpreted as the result of several measures
combined. As noted in one of the reports (Vadeby et al., 2015), further investigations of the measure
are warranted. There is insufficient evidence to conclude whether 2+1 roads with median barriers
mainly reduce the severity of accidents or whether the total accident numbers are affected. There is
limited information on how the measure affects different types of road users (e.g. HGVs, PTWs).
2+1 roads without median barriers appear to have beneficial effects on safety in terms of crash
counts, but the evidence is somewhat limited. There are, however, other reports on the safety
effects of 2+1 roads (Buch, 2015; Lee et al., 2010; Weber & Löhe, 2003). While these studies have
methodological issues that imply their results should be interpreted with care (most commonly a
lack of statistical analysis), they all appear to indicate that 2+1 road design with or without median
barriers does not influence road safety in a negative manner, and several studies find positive safety
effects.
Most estimates are for injury rates (number of injured road users per million axle pair kilometres),
but one of the estimates for 90 km/h speed limits is based on crash counts.
The results, displayed in table 2, show that the implementation of 2+1 roads is associated with
lower, although in some cases not statistically significant changes in injury rates. The results for
severe and fatal injury rates are more consistent in terms of direction and statistical significance
than are the results for less severe injuries. Generally, 2+1 roads with median barriers are found to
severe and fatal accidents by 40-60 %, while the effect on rates of less severe injuries range from an
8 % increase to a 29 % decrease. As is also seen from table 2, 2+1 roads are not always found to
reduce injury rates at roads with speed limits of 110 km/h.
8
2+1 roads
Table 2. Effects of 2+1 roads on road safety by accident severity and speed limit.
Estimates Results (n estimates) Results (% of estimates)
n - ↘ - ↘
Speed limit
110 2 1 1 50 50
100 2 2 100
90 3 3 100
Severities
Person 3 1 2 33 66
injuries
FSI rate 3 0 3 - 100
Note: * The summarized estimates are for the full road sections evaluated in each study, not
distinguishing intersections/junctions from the remaining road, in accordance with what was coded for
the repository.
A note on effects on severity versus frequency of crashes. The results of the reviewed studies
mainly focused on severe and fatal accidents. Effects on damage only crashes were considered in
the Swedish studies, but due to the degree of uncertainty associated with these data (more
underreporting, change in reporting procedure during study period), they were not considered in the
vote-count analysis. However, findings of Carlsson (2009) indicate that the rate of less severe
accidents may have decreased a smaller amount, or even increased, after implementation of 2+1
road design.
2.4 CONCLUSION
Three primary studies were reviewed and summarized. The main finding is that implementing 2+1
roads, with or without median barriers, appears to improve road safety, especially by reducing the
rates of fatal and severe injuries. While the evaluations studies find the effects of the measure to be
very large, these estimates may be somewhat inflated by methodological challenges. Little is known
about whether this measure differentially affects the safety levels of different types of road users,
and evaluation studies that could not be coded for the DSS indicate that the effect of 2+1 roads with
no median barrier depends on factors such as the road type and design of intersections.
9
2+1 roads
3 Supporting document
3.1 METHODOLOGY
3.1.1 Literature search strategy
The databases Science Direct, TRID and Taylor & Francis, Web of Science, and Springer Link were
used to identify relevant studies for 2+1 roads. Since the searches returned few relevant results, an
additional search was performed in Google Scholar. Additionally, the Handbook of Road Safety
Measures was screened for relevant studies. The Handbook of Road Safety Measures includes a
chapter of 2+1 roads that was updated in 2011. This chapter is based on an extensive literature
search for publications relating the measure to road safety. Therefore, the systematic literature
search was limited to 2010 and later in the following. The chapter on 2+1 roads in the Handbook of
Road Safety Measures did not contain a meta-analysis, and among the studies included in this
chapter, only studies that met the criteria below were reviewed for this synopsis.
Publications in English, Swedish, Danish and Norwegian were considered for coding. Studies were
screened for relevance based on title and abstract, and publication lists of relevant studies were
screened for additional relevant studies. Due to a lack of peer-reviewed journal articles, the search
also included conference papers and research reports.
Attempts at searching in the TRID database proved to be unhelpful, as “+” could not be included in a
search term, and no relevant studies were identified using “plus”. Searches conducted excluding
“plus” and “+” returned unmanageable numbers of irrelevant hits.
3.1.2 Principles
Limitations/exclusions for search in all databases:
• Title-ABSTR-KEY
• English, Norwegian, Danish or Swedish language
10
2+1 roads
#1 2+1 roads 47
("2+1 road") and (safety OR accident OR risk OR injur* OR crash) 20
Limitations: English language, published 2011 or later, in the field of engineering.
Google scholar
search search terms / operators / combined queries hits
no.
#1 ("2+1 road*" OR "two plus one road*") AND (safety OR experience* OR risk 325
OR accident* OR injur* OR crash*) 2010 and later.*
*search also conducted with the same terms in Norwegian, Swedish, and Danish
Finally, 1 publication could not be accessed. The 3 remaining publications were coded.
Buch, T. S. (2015). Safety evaluation of 2+1 roads – Accident rates and injury rates. Trafitec Report.
[Danish, executive summary in English]
Lee, S. K., Kim, Y. R., Moon, J. P., sung Choi, J., & Kim, T. H. (2010, June). Operational Analysis of 2+
1 Roadway and its Use in Developing Geometric Design Standards in South Korea. In 4th
International Symposium on Highway Geometric Design.
11
2+1 roads
Weber, R., & Löhne, H. (2003). Verkehrssicherheit und Verkersablauf auf b2+1-Strecken mit
allgemeinem Verkehr. Berichte der Bundesanstanlt für Strassenwesen, Heft V 109. [German,
executive summary in English]
Sandle, I., Aspinall, B., Hasen, D., & Smart, J. (2005). Wide single 2+ 1 Carriageways in the UK.
Presented at 3rd International Symposium on Highway Geometric Design. June 29-July 1, 2005,
Chicago, Illinois
Duplicate data:
Berg, T., Carlsson, A., & Moberg, J. (2005). 2+1 Roads with Cable Barriers--A Swedish Success Story
from 3rd International Symposium on Highway Geometric Design, June 29-July 1, 2005, Chicago,
Illinois: Compendium of Papers CD-ROM
No consideration of safety:
Munehiro, K., Takemoto, A., Takahashi, N., Watanabe, M., & Asano, M. (2012). Performance
evaluation for rural two-plus-one-lane highway in a cold, snowy region. Transportation Research
Record: Journal of the Transportation Research Board, (2272), 161-172.
Choi, J., Kim, Y., Bergh, T., Kim, S., & Kim, S. (2016). Sustainable Design of Rural Roads with 2+1
Road Design: Levels of Service and Traffic Flow Performance. KSCE Journal of Civil Engineering, 20,
1032-1039.
Coded studies
Cafiso, S. (2015). Investigating the influence on safety and traffic performance of 2+1 road sections
in Poland. 5th International Symposium on Highway Geometric Design, June 22-24, 2015,
Vancouver, Canada
Vadeby, A., Anund, A., Ekström, C., Gustafsson, S., Lundberg, T., Olstam, J., & Tapani, A. (2015).
Säker framkomlighet [Safe Accessibility]. VTI report 898.
Note: English, less detailed versions of the Swedish publications may be found in the following:
Vadeby, A. (2016). Traffic safety effects of narrow 2+ 1 roads with median barrier in Sweden. In 17th
International Conference Road Safety on Five Continents (RS5C 2016), Rio de Janeiro, Brazil, 17-19
May 2016. Statens väg-och transportforskningsinstitut.
Carlsson, A. (2009). Evaluation of 2+1 roads with cable barrier. Final report. VTI report 636A.
12