Content uploaded by Ronald K Faller
Author content
All content in this area was uploaded by Ronald K Faller on Nov 05, 2014
Content may be subject to copyright.
This article was downloaded by: [University of Nebraska, Lincoln]
On: 03 November 2014, At: 13:56
Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered
office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK
Journal of Transportation Safety &
Security
Publication details, including instructions for authors and
subscription information:
http://www.tandfonline.com/loi/utss20
Cost-Benefit Analysis of Crash Cushion
Systems
Kevin D. Schruma, Francisco D. B. De Albuquerquea, Dean L.
Sickinga, Karla A. Lechtenberga, Ronald K. Fallera & John D. Reida
a Midwest Roadside Safety Facility, Whittier Research Center,
University of Nebraska-Lincoln, Lincoln, Nebraska, USA
Accepted author version posted online: 10 Mar 2014.Published
online: 28 Oct 2014.
To cite this article: Kevin D. Schrum, Francisco D. B. De Albuquerque, Dean L. Sicking, Karla A.
Lechtenberg, Ronald K. Faller & John D. Reid (2015) Cost-Benefit Analysis of Crash Cushion Systems,
Journal of Transportation Safety & Security, 7:1, 1-19, DOI: 10.1080/19439962.2013.846448
To link to this article: http://dx.doi.org/10.1080/19439962.2013.846448
PLEASE SCROLL DOWN FOR ARTICLE
Taylor & Francis makes every effort to ensure the accuracy of all the information (the
“Content”) contained in the publications on our platform. However, Taylor & Francis,
our agents, and our licensors make no representations or warranties whatsoever as to
the accuracy, completeness, or suitability for any purpose of the Content. Any opinions
and views expressed in this publication are the opinions and views of the authors,
and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content
should not be relied upon and should be independently verified with primary sources
of information. Taylor and Francis shall not be liable for any losses, actions, claims,
proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or
howsoever caused arising directly or indirectly in connection with, in relation to or arising
out of the use of the Content.
This article may be used for research, teaching, and private study purposes. Any
substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,
systematic supply, or distribution in any form to anyone is expressly forbidden. Terms &
Conditions of access and use can be found at http://www.tandfonline.com/page/terms-
and-conditions
Journal of Transportation Safety & Security, 7:1–19, 2015
Copyright © Taylor & Francis Group, LLC and The University of Tennessee
ISSN: 1943-9962 print / 1943-9970 online
DOI: 10.1080/19439962.2013.846448
Cost-Benefit Analysis of Crash Cushion Systems
KEVIN D. SCHRUM, FRANCISCO D. B. DE ALBUQUERQUE,
DEAN L. SICKING, KARLA A. LECHTENBERG,
RONALD K. FALLER, AND JOHN D. REID
Midwest Roadside Safety Facility, Whittier Research Center, University of
Nebraska-Lincoln, Lincoln, Nebraska, USA
Crash cushions vary in geometry and cost. In this study, crash cushions were categorized
in three different categories: redirecting with repair costs greater than $1,000 (RGM),
redirecting with repair costs less than $1,000 (RLM), and nonredirecting sacrificial
(NRS). Typically, RGM systems are less expensive initially, but life-cycle costs are
high. RLM systems typically reciprocate this trend. NRS crash cushions (e.g., sand
barrels) are generally less expensive but require total replacement after a crash has
occurred, which may be impractical at high-traffic volume locations. Due to limited
funding, there is often a need to identify the most cost-effective crash cushion category
for highway scenarios with different roadway, traffic, and roadside characteristics. This
study was commissioned to determine benefit-cost ratios for each crash cushion category
in a wide range of roadway and roadside characteristics using the probability-based
encroachment tool, Roadside Safety Analysis Program. Only RGM and RLM systems
were cost-effective for freeways and divided rural arterials, but all three categories
competed against the unprotected condition on undivided rural arterials and local
roads.
Keywords roadside, crash cushions, benefit-cost, RSAP
1. Introduction
1.1. Background
Crash cushions are used to reduce the severity of an impact with a fixed, narrow object.
This is usually accomplished using energy absorption to reduce the vehicle’s kinetic energy
and, ultimately, its speed at a safe deceleration rate. Crash cushions are ideal for fixed
objects that cannot be removed, relocated, or shielded by longitudinal barriers (American
Association of State Highway and Transportation Officials [AASHTO], 2011b). The need
for a crash cushion is partially dependent on the clear zone distance, which is the minimum
distance at which a fixed object may be placed and still leave enough recovery area for the
driver to avoid that fixed object.
Address correspondence to Francisco D. B. De Albuquerque, Midwest Roadside Safety Facility,
Whittier Research Center, University of Nebraska-Lincoln, 2200 Vine Street, Lincoln, NE 68583-
0853, USA. E-mail: danielbenicio@hotmail.com
1
Downloaded by [University of Nebraska, Lincoln] at 13:56 03 November 2014
2 K.D.Schrumetal.
Crash cushions were defined according to their average repair costs. A distinction
between systems in the redirecting family was made by selecting $1,000 as the descriptive
cost, which provided consistent grouping of the systems with respect to common practices
in the industry, such as those in the Roadside Design Guide (RDG) (AASHTO, 2011b).
For simplicity, the Roman numeral for 1,000 (M) was used in the designations. The re-
sulting categories were Redirecting with repair costs less than or equal to $1,000 (RLM),
Redirecting with repair costs Greater than $1,000 (RGM), and Non-Redirecting Sacrificial
(NRS).
The repair costs of the RLM category were relatively low because of concept of
restorability and, given a design impact, the cost of the parts needed to repair the system are
inexpensive. However, there is a trade-off for these low repair costs. They require a higher
up-front investment in installation.
In contrast, the repair costs of the RGM category are higher per impact event because
these systems generally make use of permanent deformation or damage to dissipate energy.
As a result, the cost of the parts needed to repair the system can be expensive. However,
the trade-off is that these systems present lower installation costs. RLM and RGM systems
are able to redirect vehicles when hit on their side, which is a significant advantage over
NRS crash cushions.
Ultimately, NRS crash cushions primarily comprise sand barrels that may be placed in
different configurations depending on the size and shape of the fixed object. These crash
cushions use the concept of incremental momentum transfer to sand particles (i.e., the
kinetic energy of the vehicle is dissipated as the vehicle hits the barrels). The mass of each
barrel varies. In a design impact, the lighter barrels are hit first, and the heavier barriers are
struck as the vehicle continues through the crash cushion. The absorption of the vehicle’s
kinetic energy makes the vehicle slow down at a safe deceleration rate until it brings the
vehicle’s energy low enough that “bulldozing” through the sand will be enough to stop
the vehicle (i.e., a velocity less than 10 mph or 16.1 km/h) (AASHTO, 2011b). Because
any impacted barrel typically suffers significant permanent deformation, the repair costs
for these systems can approach the initial installation costs because they may have to be
completely replaced. Also, because these systems are nonredirecting crash cushions, they
may allow vehicles to gate through them, potentially inducing a more harmful event. On
the other hand, these systems typically had the lowest installation costs.
1.2. Problem Statement
Guidelines contained in the RDG list crash cushions as a safety treatment for fixed objects
that cannot be removed, relocated, or shielded by longitudinal barriers (AASHTO, 2011b).
However, the use of a crash cushion may not be economically justifiable under certain traffic
and roadside characteristics. For example, the installation of a high-cost crash cushion may
not be economically justifiable on a road with low-traffic volumes and large lateral offsets
because the crash frequency will tend to be very low. As a result, the use of different crash
cushions may depend on varying roadway, roadside, and traffic characteristics, making the
selection of a specific crash cushion type challenging for transportation safety engineers.
Therefore, there is a need to develop crash cushion selection guidelines that can be
used to assist engineers in selecting the crash cushion that results in the highest accident
cost reduction per unit of direct cost (i.e., installation and repair cost) associated with the
chosen crash cushion. However, to provide flexibility in design options available to the
Downloaded by [University of Nebraska, Lincoln] at 13:56 03 November 2014
Analysis of Crash Cushion Systems 3
engineer, crash cushion categories (e.g., RLM, RGM, or NRS) needed to be compared such
that selection guidelines pertained to broad categories rather than specific systems.
1.3. Objective
The objective of this research study was to develop crash cushion selection guidelines to
help highway engineers select the most cost-beneficial crash cushion to be used on various
highway scenarios considering a wide range of roadway, roadside, and traffic characteristics.
1.4. Scope
The objective of this research study was achieved through various tasks. First, crash cushion
systems were examined to understand dimensions and associated costs for each system via
manufacturer product sheets and surveys sent out to State Departments of Transportation
(DOTs) and manufacturers. Next, using the Roadside Safety Analysis Program (RSAP),
roadway parameters were chosen for the study based on their influence in determining ac-
cident cost. Then, by modifying these parameters, several highway scenarios were modeled
to evaluate the benefit-cost (BC) ratios of each crash cushion. Next, direct costs were de-
termined based on mobilization, labor, installation, maintenance, and repair costs. Societal
costs were determined based on the 2010 Federal Highway Administration (FHWA) com-
prehensive costs. Finally, BC analyses were conducted to determine whether the placement
of a type of crash cushion was economically justifiable. Example applications of the results
were included to assist engineers in the selection process.
2. Crash Cushion Systems
The QuadGuard is a proprietary crash cushion manufactured by Energy Absorption Sys-
tems, Inc., a subsidiary of Trinity Highway Products, LLC (Energy Absorption Systems,
Inc., 2013b). It utilizes crushable cartridges that need to be replaced after an impact event.
These cartridges are placed within a structure of quad beams that are designed to “fish-
scale” backward as a vehicle strikes the end. The length of the QuadGuard was 15 ft (4.6 m),
and the width was 2.5 ft (0.8 m).
The QUEST crash cushion is a proprietary crash cushion manufactured by Energy Ab-
sorption Systems, Inc., a subsidiary of Trinity Highway Products, LLC (Energy Absorption
Systems, Inc., 2013c). It telescopes backward to dissipate kinetic energy. The length of the
QUEST was 19 ft (5.8 m) and the width was 2.0 ft (0.6 m).
The Trinity Attenuating Crash Cushion (TRACC) is a proprietary crash cushion manu-
factured by Trinity Highway Products, LLC (Trinity Highway Products, 2013). It telescopes
backward while tearing through metal plates. The length of the TRACC crash cushion was
21.25 ft (6.5 m) and the width was 2.0 ft (0.6 m).
The TAU II is a proprietary crash cushion manufactured by Barrier Systems, Inc.
(Barrier Systems, 2013). It absorbs the kinetic energy of the vehicle using disposable
energy absorbing cartridges. The length of the TAU II was 23 ft (7.0 m) and the width was
4.0 ft (1.2 m).
The QuadGuard Elite is a proprietary crash cushion manufactured by Energy Absorp-
tion Systems, Inc., a subsidiary of Trinity Highway Products, LLC (Energy Absorption
Systems, Inc., 2013a). It utilizes self-restoring cylinders made from high-density polyethy-
lene (HDPE). The cylinders are placed within a structure of quad beams that are designed
Downloaded by [University of Nebraska, Lincoln] at 13:56 03 November 2014
4 K.D.Schrumetal.
to fish-scale backward as a vehicle strikes the end. The length of the QuadGuard Elite was
27 ft (8.2 m), and the width was 2.0 ft (0.6 m).
The Reusable Energy-Absorbing Crash Terminal (REACT 350) is a proprietary crash
cushion manufactured by Energy Absorption Systems, Inc. (Energy Absorption Systems,
Inc., 2013d), a subsidiary of Trinity Highway Products, LLC. HDPE cylinders are placed
in a single row and restrained by cables on either side. The length of the REACT 350 was
28.75 ft (8.8 m) and the width was 3.0 ft (0.9 m).
The Smart Cushion is a proprietary crash cushion manufactured by Smart Cushion
Innovations (SCI) Products, Inc. (SCI Products, 2013). The length of the SCI was 21.5 ft
(6.6 m), and the width was 2.0 ft (0.6 m).
NRS systems are typically represented by sand barrels that can be arrayed in numerous
designs. Sand barrels can be arrayed to shield almost any fixed object. Further, sand barrels
are inexpensive and easy to design and construct. However, repair costs can be high because
the system usually requires total replacement of the impacted barrels. Sand barrels cannot
redirect vehicles in the event of a side impact, do not guarantee that lighter barrels are struck
first, and perform poorly in coffin corner impacts. Higher-speed highways generally require
sand barrel configurations that contain more barrels. The masses of the barrels increase
as the system approaches the hazard. This provides a relatively safe deceleration rate for
the vehicle until it slows to a safe velocity, which was specified in the RDG to be 10 mph
(16.1 km/h) (AASHTO, 2011b).
3. Survey of Crash Cushion Costs
To estimate the cost of installation of all crash cushions used in this study, a survey
questionnaire was sent to the following Midwest States Pooled Fund States Departments
of Transportation (DOTs): Illinois, Iowa, Kansas, Minnesota, Missouri, Nebraska, Ohio,
South Dakota, Wisconsin, and Wyoming. The State DOTs were asked to provide infor-
mation pertaining to each crash cushion that they currently implement. This information
included the average installation cost, the average crash repair cost, and the average regular
maintenance cost per year. Additional information included inventory need and costs for
each crash cushion type, repair time needed once the system has been involved in a crash,
and information on the test level and speed limit of each particular crash cushion used.
Only a few States replied to the survey. Also, not all responders answered the questions
adequately, which decreased the number of survey responses even further. Responses from
Kansas, Minnesota, and Wisconsin were used in the study.
A summary of costs and dimensions of each crash cushion type evaluated in this study
is shown in Table 1. Dimensions were taken from manufacturer product sheets for typical
Test Level 3 (TL-3) designs. The cost of the sand barrels in this table was the average
of three online distributers for the same design configuration (Transportation Safety &
Equipment Co., 2011; Transportation Supply, 2011; Twin Discovery Systems, Inc., 2011).
Crash cushion size was directly associated with the safety performance of the crash cushion.
Costs were independent of the crash cushion size because the States did not provide detailed
cost information as a function of crash cushion size or safety performance level.
4. Repair Cost Estimation
Manufacturers of the systems described herein were solicited for repair cost estimations
for each of the NCHRP Report No. 350 crash tests conducted for the given system. Crash
test numbers 3–31, 3–33, and 3–37 were conducted for all of the redirecting systems in
Downloaded by [University of Nebraska, Lincoln] at 13:56 03 November 2014
Analysis of Crash Cushion Systems 5
Table 1
Costs and dimensions used in the benefit-cost analysis
Crash Cushion Installation Cost Length, ft (m) Width, ft (m)
QuadGuard $17,769 21 (6.40) 2.0 (0.61)
QUEST $11,510 19 (5.79) 2.0 (0.61)
TRACC $11,400 21.25 (6.48) 2.0 (0.61)
TAU I I $15,433 23.0 (7.01) 4.0 (1.22)
QuadGuard Elite $33,017 27.0 (8.23) 2.0 (0.61)
REACT 350 $36,067 28.75 (8.76) 3.0 (0.91)
SCI $19,371 21.5 (6.55) 2.0 (0.61)
Sand Barrels $2,540 16.5 (5.03) 6.0 (1.83)
this article. For each of these tests, the manufacturers provided the estimated cost for repair
parts and the estimated time to repair the system. Assuming a labor cost of $50 (USD) per
hour, the average repair costs for each system from the three mutual tests were determined
and are shown in Table 2.
The target velocity of each of the three mutual tests was specified to represent the
85th percentile speed in real-world accidents. Therefore, the average repair costs in Table 2
were adjusted for each of the three functional classes considered in this article according
to the average impact velocity of those functional classes. Previous research has shown
that the average impact velocity for freeways, arterials, and local highways were 45.3 mph
(73.0 km/h), 39.3 mph (63.2 km/h), and 34.9 mph (km/h), respectively (Albuquerque et al.,
2009). Impact severity (IS) is a function of the square of this velocity. By determining the
IS for the two different speeds, the IS for the real-world accident velocity for the given
functional class can be estimated according to Equation 1.
IS =v50
v85 2
(IS)(1)
where
IS =reduced impact severity
IS =impact severity of the test conditions
Table 2
Average repair costs in U.S. dollars for standard National Cooperative High-
way Research Program (NCHRP) Report No. 350 crash tests
System Avg Repair Cost For Mutual Tests
SCI $67.33
REACT 350 $66.67
QuadGuard Elite $638.33
TRACC $1,933.33
TAU I I $2,518.83
QuadGuard $3,909.67
QUEST $9,683.33
Downloaded by [University of Nebraska, Lincoln] at 13:56 03 November 2014
6 K.D.Schrumetal.
Table 3
Reduced average repair costs based on functional class
Freeway Avg Arterial Avg Local Avg
System Repair Cost Repair Cost Repair Cost
SCI $35.83 $26.97 $21.27
REACT 350 $35.47 $26.70 $21.06
QG Elite $339.67 $255.65 $201.61
TRACC $1,028.77 $774.30 $610.62
TAU I I $1,340.33 $1,008.79 $795.55
QG $2,080.43 $1,565.82 $1,234.83
QUEST $5,152.74 $3,878.17 $3,058.39
v50 =velocity of an average impact.
v85 =target velocity of the crash test.
Applying the average impact velocities to Equation 2, the IS was reduced for freeways,
arterials, and local highways using ratios of 0.5253, 0.3954, and 0.3118, respectively.
Because it was assumed that repair cost was directly related to IS, the average repair costs
for mutual tests were multiplied by these same ratios. Therefore, the costs associated with
the reduced velocity approach are shown in Table 3.
5. New Categories Based on Repair Costs
As aforementioned, three new categories were developed for the purpose of conducting a
comparative study between similar groups of systems. These categories were based entirely
on the repair costs data supplied by manufacturers and in no way are meant to classify a
system according to performance, ease of installation, or any other subjective method of
description. Based on the $1,000 threshold, the crash cushions were categorized according
to Table 4.
Table 4
Crash cushion categories
Category RDG Designation Study Definition System
RLM Low Maintenance Repair Cost ≤$1,000 SCI
REACT 350
QuadGuard Elite
RGM Reusable Repair Cost > $1,000 TAU II
QUEST
TRACC
QuadGuard
NRS Sacrificial NA Sand Barrels
Downloaded by [University of Nebraska, Lincoln] at 13:56 03 November 2014
Analysis of Crash Cushion Systems 7
Figure 1. Crash cushion placement on (a) divided and (b) undivided highways.
6. Highway Scenario Modeling
6.1. Arbitrary Unprotected Roadside Condition
Hundreds of highway scenarios were modeled using RSAP, which is a probability-based
encroachment tool used to estimate the cost-effectiveness of roadside safety treatment
alternatives (Mak & Sicking, 2003). Different crash cushions were used on each modeled
scenario to determine the B/C ratio of each system relative to the do-nothing alternative as
well as relative to the other systems.
Different highway scenarios were created by varying values of traffic, roadway, and
roadside parameters used to characterize a specific scenario. A hypothetical highway sce-
nario was modeled in RSAP. This scenario is shown in Figure 1 and shows 4 ×2-ft bridge
piers placed on the roadside and in the median of divided highways (as shown in Figure 1a)
and on the roadside of undivided highways (as shown in Figure 1b).
6.2. Sensitivity Analysis
A sensitivity analysis was conducted to determine highway and traffic characteristics that
significantly affect accident costs in RSAP. If a parameter had a significant influence on
accident cost change, then the parameter would be considered further in the study.
The parameters that were analyzed in the sensitivity analysis were crash cushion offset,
average daily traffic, horizontal curvature, number of traffic lanes, lane width, and shoulder
width. These parameters were then programmed into RSAP and their corresponding values
were chosen based on typical ranges (i.e., low, medium, and high values) observed on
freeways and local roads. In other words, values varied based on the functional roadway
class. The traffic volume ranges were determined with assistance from AASHTO (2011a)
Geometric Design of Highways and Streets. Curvature was chosen based on a summary
of State standards given in NCHRP Report No. 638 (Sicking et al., 2009). Offsets were
set out as far as 35 ft (10.7 m). According to the RDG, clear zones of 30 ft can allow as
much as 80% of the vehicles enough room to recover (AASHTO, 2011b). By increasing
this distance, even more errant vehicles would be able to safely recover before impacting
the fixed object. However, identifying the exact critical offset was outside the scope of this
research.
This significance of a parameter was determined based on the assumption that fluctua-
tions of less than 20% were insignificant. The significant parameters included were (1) crash
Downloaded by [University of Nebraska, Lincoln] at 13:56 03 November 2014
8 K.D.Schrumetal.
Table 5
Sensitivity results for a freeway
Annual Accident Percent
Parameter Range Cost ($) Difference (%)
Crash Cushion Offset (ft) 6 2,840.88 27.5%
12 (baseline) 2,227.42 na
18 1,729.37 22.4%
Average Daily Traffic
(veh/day)
5,000 2,453.54 20.6%
10,000 (baseline) 3,091.41 na
20,000 5,200.27 68.2%
Horizontal Curvature
(degrees)
0 2,453.54 26.7%
2 (baseline) 1,937.24 na
4 3,534.91 82.5%
No. of Lanes 4 2,453.54 12.3%
6 (baseline) 2,798.80 na
8 3,309.96 18.3%
Lane Width (ft) 10 2,453.54 6.7%
12 (baseline) 2,299.99 na
14 2,114.62 8.1%
Shoulder Width (ft) 8 2,453.54 0.0%
10 (baseline) 2,453.54 na
12 2,453.54 0.0%
cushion offset, (2) average daily traffic, and (3) horizontal curvature. Because the sensitivity
results for freeways and local highways indicated the same dependencies, the analysis was
not required for arterial highways. The resulting sensitivity of the aforementioned variables
for freeways and local highways are given in Tables 5 and 6, respectively.
6.3. Parameter Values
Parameters with a sensitivity of more than 20% were selected for a detailed analysis in
RSAP. Three parameters met this requirement and are shown in Table 7.
Constant, but reasonable values were chosen for parameters deemed insensitive in this
analysis. The lane width was 12 ft (3.66 m) and the shoulder width was 8 ft (2.44 m). Two
lanes were used on local roads and undivided rural arterials. Four lanes (i.e., two in each
direction) were used on freeways and divided arterials.
7. Societal and Direct Cost Estimation
7.1. Societal Costs
According to the Federal Highway Administration (FHWA), the average cost of a human
life was $2.6 million dollars in 1994 (AASHTO, 1996). This accounted for the loss of
income over the remainder of the victim’s life and the willingness of society to pay for the
accident. That number has since increased through inflation. In 2010, the gross domestic
Downloaded by [University of Nebraska, Lincoln] at 13:56 03 November 2014
Analysis of Crash Cushion Systems 9
Table 6
Sensitivity results for a local highway
Annual Accident Percent
Parameter Range Cost ($) Difference (%)
Crash Cushion Offset (ft) 3 566.98 37.7%
8 (baseline) 411.81 na
13 276.02 33.0%
Average Daily Traffic (veh/day) 1,000 411.81 58.1%
3,000 (baseline) 982.57 na
5,000 1,170.41 19.1%
Horizontal Curvature (degrees) 0 411.81 38.9%
5 (baseline) 673.73 na
10 638.37 5.2%
No. of Lanes 2 411.81 25.3%
4 (baseline) 551.41 na
6 626.56 13.6%
Lane Width (ft) 8 411.81 6.8%
10 (baseline) 385.59 na
12 373.34 3.2%
Shoulder Width (ft) 4 411.81 0.0%
6 (baseline) 411.81 na
8 411.81 0.0%
product implicit price deflator was 111.141 (Bureau of Economic Analysis, 2011). Utilizing
this value, the costs of each injury level on the KABCO scale (with K being a fatality and
O being property damage only) scaled up for inflation according to Equation 2. Using this
approach, the KABCO costs were scaled to the values shown in Table 8.
AccCost =PGDP2010
GDP1994 (2)
where
GDP2010 =111.141
GDP1994 =80.507
P=the principal in 1994 dollars.
Table 7
Roadside Safety Analysis Program modeling parameter values
Parameter Freeways Rural Arterials Local Highways
ADT (1,000s) 5, 10, 25, 50, 75,
100
1, 5, 10, 20, 30 0.2, 0.5, 1, 3
Curvature (Deg) 0, 2, 4 0, 3, 6 0, 5, 10
Offset, ft (m) 5.0 (1.5), 15.0 (4.6)
25.0 (7.6), 35.0
(10.7)
5.0 (1.5), 10.0 (3.1),
15.0 (4.6), 20.0
(9.1), 35.0 (10.7)
5.0 (1.5), 10.0 (3.1),
15.0 (4.6), 20.0
(9.1), 35.0 (10.7)
Downloaded by [University of Nebraska, Lincoln] at 13:56 03 November 2014
10 K. D. Schrum et al.
Table 8
Societal costs for each injury level
Injury Level Cost (US$)
K$3,589,335
A$248,492
B$49,698
C$26,230
PDO $2,761
Using this scale and the predicted accident frequency, RSAP was able to determine an
accident cost for each crash cushion at each location. Simulated accident costs are contained
in Guidelines for Crash Cushion Selection (Schrum et al., 2013).
7.3. Direct Costs
7.3.1. Mobilization and Labor Costs. Mobilization costs were not included in this study
because mobilization is highly variable and dependent on the site location. Costs could
be high if the distance to the site was great or low if the distance was minimal. However,
because these costs to mobilize would be equal for the all systems being compared in the
analysis, they cancel out of the analysis.
Installation costs ascertained from the State of Wisconsin were used for RGM and
RLM crash cushions, as shown in Table 1. However, Wisconsin did not list inertial sand
barrels (NRS systems) in their survey response. As a result, price estimates were taken
from online transportation safety equipment dealers. Using a 1,400-lb (635-kg) barrel, the
average cost from three dealers was $2,540 (Transportation Safety & Equipment Co., 2011;
Transportation Supply, 2011; Twin Discovery Systems, Inc., 2011).
Labor and utility truck costs were assumed to be $50 and $125 per hour based on
correspondence with the State of Wisconsin. A difference was observed when comparing
each crash cushion type in the time required to make repairs. Labor costs included labor
for a two-man crew to make repairs. Labor cost estimates submitted by the Wisconsin
DOT assumed a setup and takedown time, including travel time, to be one hour each. This
time was considered separately for NRS crash cushions. According to survey response
submitted by the Minnesota DOT, the approximate time for repairs of Energite III (i.e.,
setup and takedown time) was an average of 4 h.
A two-man crew was used for setup, takedown, and repair of the crash cushion. For
each crash cushion, a fixed cost based on a setup and takedown time of the work zone was
assumed to be one hour for each phase, resulting in a total of four man hours and a labor
cost of $200. The truck was rented for one hour at $125. Summing each fixed cost resulted
in a total hourly fixed labor and utility truck cost of $325.
Because each crash cushion had a different repair time, each system also had a different
variable repair cost. Repair time and associated labor and utility truck costs for each crash
cushion system are summarized in Table 9 and were determined using Equation 3. Based on
the reported time to repair a system following a standard NCHRP Report No. 350 crash test
(Ross et al., 1993), the cost of labor and utility truck use was determined and is shown in
Table 9.
Lcost =Hourlycost RepairTimeavg(3)
Downloaded by [University of Nebraska, Lincoln] at 13:56 03 November 2014
Analysis of Crash Cushion Systems 11
Table 9
Summary of labor costs
System Man Hours for Repairs Labor Cost∗
SCI 1.33 $300.00
REACT 350 1.00 $225.00
QG Elite 1.00 $225.00
TRACC 2.33 $525.00
TAU I I 0.7 8 $174.75
QG 1.17 $262.50
QUEST 3.00 $675.00
∗Not including mobilization costs.
where
Lcost =Total labor and truck rental costs
Hourlycost =Hourly rate to repair the system ($225)
Repair Timeavg =Average time required to repair the system.
7.3.2. Regular Maintenance Costs. Responses from State DOTs indicated either a total
maintenance cost for all crash cushions (i.e., as opposed to average maintenance costs per
system) in the state or were a replication of the repair costs. Therefore, maintenance costs
were set to zero for this analysis, and this practice was confirmed in correspondence with
DOT officials who noted that these systems do not typically receive maintenance unless
they are struck, at which point the maintenance cost becomes a repair cost.
8. Benefit-Cost Analysis
Once all direct and societal costs have been estimated, they can be used in Equation 4 to
calculate the BC ratio for each safety alternative, including the “do-nothing” option.
BC2−1=(AC1−AC2)
(DC2−DC1)(4)
where
AC1=the accident cost of the baseline or “do-nothing” alternative design
AC2=the accident cost of the new alternative design
DC1=the direct cost of the baseline design
DC2=the direct cost of the new design or safety treatment used.
The accident costs used for each scenario and for each design alternative are tabulated
in Guidelines for Crash Cushion Selection (Schrum et al., 2013). The costs were annualized
using a design life of 25 years and a discount rate of 4%. This parameter represents the
difference between interest rates and the annual inflation rate and is commonly accepted as
the appropriate value for use in economic analyses for government-funded projects (Mak
& Sicking, 2003).
A ratio of 1.0 meant that at the end of the 25-year design life, the accident costs and
direct costs were offset. In general transportation investment practices, this would not be
Downloaded by [University of Nebraska, Lincoln] at 13:56 03 November 2014
12 K. D. Schrum et al.
Figure 2. Example of weighted average system. B/C =benefit/cost.
worth the effort. Instead, a minimum ratio of 2.0 is usually suggested, with a ratio of 4.0
being preferred.
Benefit-cost analyses were conducted in two ways: (1) an index method was developed
to compare categories of crash cushions to only the baseline option and (2) an incremental
method was incorporated to ascertain the optimal cost-effective option for each highway
scenario.
8.1. Index Method
One goal of this project was to determine cost-effective crash cushion categories for a given
highway scenario rather than a particular crash cushion. It may be possible to have a RLM
crash cushion as the best option, but there may be four RGM crash cushions that are also
cost-effective.
A system of weighted averages was used to determine if a category was cost-effective
for each highway scenario. This system accounted for the number of crash cushion types
above the BC threshold and the average BC ratios for each type. Effectively, if one system
within the category exceeded the BC threshold, the category as a whole was deemed
cost-effective, thus tending toward implementing a crash cushion. This system was best
explained through an example, as shown in Figure 2. The given BC ratios shown in Figure 2
were generated arbitrarily and do not reflect any of the tested scenarios.
In the hypothetical example illustrated by Figure 2, four RGM and three RLM crash
cushions were considered. A ratio of the number of beneficial crash cushions to the total
number of crash cushions for each category was calculated (rRGM and rRLM). The average
BC ratio of each type of crash cushion was determined (BCRGM and BCRLM), including the
ones that did not exceed the BC threshold. An index was used to rank the crash cushion
categories (IRGM and IRLM). This index was the product of the ratio, ri, and the average BC
ratio, BCi. Equations 5–10 show the calculations for the example shown in Figure 2.
rRGM =4
4=1.0(5)
rRLM =2
3=0.667 (6)
BCRGM =3.100 +3.050 +2.850 +2.150
4=2.788 (7)
BCRLM =3.200 +2.050 +1.850
3=2.367 (8)
Downloaded by [University of Nebraska, Lincoln] at 13:56 03 November 2014
Analysis of Crash Cushion Systems 13
IRGM =rRGMBCRGM =1.000 ×2.788 =2.788 (9)
IRLM =rRLMBCRLM =0.667 ×2.367 =1.579 (10)
Because IRGM and IRLM were greater than 0, both categories in this arbitrary example
were cost-effective. However, if a transportation agency adopts a minimum BC ratio of 2,
only RGM crash cushions would be recommended in this case.
8.2. Incremental Method
It is possible that the option with the highest BC ratio (say option “A”) with respect to the
unprotected condition may not be the optimal option. Consider another option (say option
“B”) whose BC ratio is smaller with respect to the unprotected condition compared with
option “A.” The additional cost of option “A” may not be offset by its increased benefit
when compared to “B.” Therefore, even though the BC ratio of “A” with respect to the
unprotected condition is greater than “B’s,” the BC ratio of “B” with respect to “A” may be
larger than the threshold (e.g., BC =2).
Because of this possibility, an incremental BC analysis was conducted by categorizing
each system after individual simulations were carried out. This categorization was done
by averaging the simulated accident costs for each highway scenario within each category.
Similarly, the direct costs (i.e., annualized installation, repair, labor) were averaged for each
highway scenario. Then, Equation 4 could be applied to determine all possible BC ratios.
8.3. Understanding the Design Charts
Symbolic representations of the recommendations that follow are given in Table 10. The
alphabetic codes in Table 10 were also used in Figure 6. Figures 3–5 show design charts
that were created to assist engineers in selecting the most cost-beneficial option, based on
an incremental benefit-cost analysis, for a specific highway scenario. Figures 6–8 show not
only the most cost-beneficial crash cushion category, but also all other categories that were
cost beneficial based on a B/C ratio of at least 2. To use these charts, the engineer must
know the traffic volume (average daily traffic [ADT]), the degree of curvature of the road
(degrees), and the offset of the crash cushion from the roadway (ft). For Figures 3–5, blank
cells refer to RGM systems, “∗” cells refer to Do-Nothing option, and “∗∗” cells refer to
RLM systems. For Figures 6–8, “A” cells refer to all systems, “B” cells refer to RGM and
RLM systems, “E” cells refer to RGM systems only, and “N” cells refer to Do-Nothing
option.
For example, given the traffic and roadway characteristics described as follows, find
the most cost-effective crash cushion type to be used.
Table 10
Legend of graphical recommendations
Legend
∗Do Nothing A All Systems
RGM B RGM and RLM
∗∗ RLM E RGM only
N Do Nothing
Downloaded by [University of Nebraska, Lincoln] at 13:56 03 November 2014
14 K. D. Schrum et al.
Figure 3. Optimal recommendations for freeways. B/C =benefit/cost.
•Highway Class =Freeway
•AADT =75,000 vehicles per day (vpd)
•Offset =15 ft (4.6 m)
•Degree of Curvature =2 degrees
•Minimum BC ratio =2.0
Solution:
•Refer to Figure 3
Select a RLM Crash Cushion Other Cost-Effective Solutions:
•Refer to Figure 6
•RGM is also cost-effective
9. Conclusions and Recommendations
Based on the incremental BC analysis, it was found that RGM crash cushions were the
optimal cost-beneficial category of crash cushions, when a BC ratio of 2 was adopted, on
freeways and divided rural arterials with traffic volumes lower than 75,000 and 20,000 vpd,
respectively, as shown in Figures 3 and 4. However, RLM systems appeared to be the most
cost-beneficial category on freeway scenarios with traffic volumes of 75,000 and 100,000
vpd, as well as on divided arterial scenarios with traffic volumes of 20,000 and 30,000
vpd. RLM crash cushions were not found to be cost-effective on divided arterial highways
when a BC ratio of 4 was adopted as shown in Figure 4. The do-nothing alternative option
was not a cost-effective alternative on freeway scenarios as shown in Figure 3. The RGM
and do-nothing options competed on undivided arterials and local highways as shown in
Figures 4 and 5. In these cases, do-nothing alternative was preferable on scenarios with
larger offsets and/or low traffic volumes.
Downloaded by [University of Nebraska, Lincoln] at 13:56 03 November 2014
Analysis of Crash Cushion Systems 15
Figure 4. Optimal recommendations for rural arterials. B/C =benefit/cost.
Therefore, RLM systems would be cost-effective at locations that experience higher
crash frequencies, while RGM crash cushions would be a more feasible option at locations
with moderate or low crash frequencies. The do-nothing alternative would only be recom-
mended on locations where there is very large crash cushion offset and/or very low traffic
volume. This finding was attributed to the fact that scenarios with low traffic volumes and
large crash cushion offsets tend to present low impact frequencies. Thus, the do-nothing
Figure 5. Optimal recommendations for local highways. B/C =benefit/cost.
Downloaded by [University of Nebraska, Lincoln] at 13:56 03 November 2014
16 K. D. Schrum et al.
Figure 6. All cost-effective options for freeways. B/C =benefit/cost.
alternative was more attractive due to its zero-installation cost. These findings indicate
the optimal cost-effective solution for each highway scenario, for use when funding is a
limiting agent.
However, often times, other options may provide the minimum BC ratio threshold of 2.
Results referring to these alternatives were presented in Figures 6–8. It was found that RLM
and RGM presented BC ratios greater than 4 on freeways. On divided arterials, all systems
were cost-effective, except on scenarios with small lateral offsets and/or traffic volumes. On
scenarios with offsets less than 20 feet, nonredirecting sacrificial crash cushions could not
be economically justifiable, as shown in Figure 7. On divided arterials with traffic volumes
Figure 7. All cost-effective options for divided and undivided rural arterials. B/C =benefit/cost.
Downloaded by [University of Nebraska, Lincoln] at 13:56 03 November 2014
Analysis of Crash Cushion Systems 17
Figure 8. All cost-effective options for local highways. B/C =benefit/cost.
of 1,000 vpd and when a BC threshold of 4 was adopted, only RGM crash cushions
were cost-effective, as shown in Figure 7. Figure 8 shows that the do-nothing option and
redirecting crash cushions competed on local roads.
It is also important to stress that other factors, which may not have been considered in
this study, may play a role in selecting a crash cushion system. For example, the time between
the impact event and the time to repair the system should be taken in consideration. If there
is a significant time gap, this could indicate that motorists would be exposed to unprotected
hazards or less effective crash cushion systems. These conditions could potentially pose
unacceptable risks to motorists. This could ultimately increase the benefits of reusable crash
cushion systems, in a risk-adjusted basis, for certain roadway scenarios.
10. Limitations and Future Work
Installation, repair, and maintenance costs were based on limited data from the State DOTs
and manufacturers. These costs may vary from region to region and from system to system.
If the variation in cost is significant, a site-specific analysis would be required.
Posted speed limits along many highways, especially freeways, are above 55 mph
(88.5 km/h). However, RSAP cannot accurately treat higher posted speed limits because
the speed distributions were based on a study that investigated impact conditions in accident
reports in the 1970s (Mak et al., 1986), which was prior to the repeal of the national speed
limit of 55 mph (88.5 km/h). However, these speed distributions do allow for impact speeds
above 55 mph (88.5 km/h).
The highest modeled impact frequency in this report was 0.13 impacts per year, and that
was on a freeway with 100,000 vpd on a 4-degree curve and a lateral offset of 5 ft (1.5 m).
Most scenarios, especially low-volume scenarios, would experience impact frequencies
far less than 0.13 impacts per year. Therefore, if the accident frequency is known, the
BC analysis results contained herein should only be used at locations with fewer than the
maximum accident frequency recommended.
This article focuses on the modeled scenarios in the RSAP benefit-cost analysis, which
represented generic roadside configurations. “Black spots” and other anomalies, such as
gore areas, were not considered due to the impracticality of modeling the decision making
process of a human being, among other difficulties. Therefore, if the impact frequency is
known, and is relatively high, then a severe-duty crash cushion may be viable for impact
frequencies as low as one impact every 2.44 years (Schrum et al., 2013).
Downloaded by [University of Nebraska, Lincoln] at 13:56 03 November 2014
18 K. D. Schrum et al.
The economic analysis contained herein was limited to quantifiable parameters per-
taining directly to the crash cushions themselves. However, there may be other life cycle
costs that could be applicable to the analysis that were not incorporated, such as delay time
while a lane of traffic is closed, disposal costs of damage systems, and the risk to human
life associated with the task of repairing these systems. Where these costs may constitute a
significant portion of the life cycle costs, an in-depth case-by-case approach for conducting
a benefit-cost analysis should be adopted.
For future studies, States should consider recording not only repair times for each
system, but also the time between the impact event and the repair should be noted for
each incident. This information could be used to demonstrate the necessity for repairing
damaged crash cushions as quickly as possible.
Acknowledgments
The authors wish to acknowledge those who made a contribution: the States participating in
the Midwest States Regional Pooled Fund Program for providing cost and usage information
and Trinity Highway Products, Energy Absorption Systems, Inc., SCI Products Inc., and
Barrier Systems, Inc. for providing repair costs.
Funding
The authors wish to thank the Wisconsin Department of Transportation for sponsoring the
project.
References
Albuquerque, F. D. B., Sicking, D. L., & Stolle, C. S. (2009). Roadway departure and impact con-
ditions (Transportation Research Record No. 2195). Washington, DC: Transportation Research
Board.
American Association of State Highway and Transportation Officials. (1996). Roadside design guide.
Washington, DC: Author.
American Association of State Highway Transportation Officials. (2011a). Geometric design of
highways and streets. Washington, DC: Author.
American Association of State Highway and Transportation Officials. (2011b). Roadside design
guide. Washington, DC: Author.
Barrier Systems, Lindsay Transportation Solutions. (2013). TAU-II redirective non-gating crash cush-
ions: Product sheet. Vacaville, CA: Author. Retrieved from http://www.barriersystemsinc.com/
stuff/contentmgr/files/0/8dbe05c3a83d8546e2595a70837718dc/files/20120701 tauii pi.pdf.
Bureau of Economic Analysis, U.S. Department of Commerce. (2011). GDP & Personal Income.
Washington, DC. Retrieved from http://www.bea.gov/national/nipaweb.
Energy Absorption Systems, Inc. (2013a). QuadGuard Elite System: Product sheet.
Chicago, IL: Author. Retrieved from http://www.energyabsorption.com/products/PI%20Sheets/
QuadGuard%20Elite.pdf.
Energy Absorption Systems, Inc. (2013b). QuadGuard System: Product sheet. Chicago, IL: Author.
Retrieved from http://www.energyabsorption.com/products/PI%20Sheets/QuadGuard.pdf.
Energy Absorption Systems, Inc. (2013c). QUEST CEN System: Product sheet. Ipswich, UK: Author.
Retrieved from http://www.energyabsorption.com/products/PI%20Sheets/Quest.pdf.
Energy Absorption Systems, Inc. (2013d). REACT 350: Product sheet. Chicago, IL: Author. Retrieved
from http://www.energyabsorption.com/products/PI%20Sheets/REACT%20350.pdf.
Downloaded by [University of Nebraska, Lincoln] at 13:56 03 November 2014
Analysis of Crash Cushion Systems 19
Mak, K. K., Sicking, D. L., & Ross, H. E. (1986). Real world impact conditions for run-off-road ac-
cidents (Transportation Research Record No. 1065). Washington, DC: Transportation Research
Board, National Research Council.
Schrum, K. D., Albuquerque, F. D. B., Sicking, D. L., Lechtenberg, K. A., Faller, R. D., & Reid,
J. D. (2013). Guidelines for crash cushion selection (Final Report to Wisconsin Department of
Transportation, Transportation Research Report No. TRP-03–252–12 [revised]). Lincoln, NE:
University of Nebraska-Lincoln, Midwest Roadside Safety Facility.
SCI Products, Work Area Protection. (2013). Smart Cushion innovations: Product sheet. St. Charles,
IL: Author. Retrieved from http://www.workareaprotection.com/SCI100GM/Attenuator.pdf.
Sicking, D. L., Lechtenberg, K. A., & Peterson, S. M. (2009). Guidelines for guardrail implemen-
tation (National Cooperative Highway Research Program Report No. 638). Washington, DC:
Transportation Research Board.
Transportation Safety & Equipment Co. (2011). Impact attenuator sand barrels. Retrieved from
http://trafficsafetydirect.com/en/impact-attenuators/impact-attenuator.html.
Transportation Supply. (2011). Impact attenuator sand barrels. Retrieved from http://www.trans-
supply.com/p-320-sand-barrels.aspx.
Trinity Highway Products. (2013). TRACC: Product sheet. Dallas, TX: Author. Retrieved from
http://www.highwayguardrail.com/products/tracc.html.
Twin Discovery Systems, Inc. (2011). Traffic sand barrel with lid. Retrieved from
http://www.trafficsafetyexperts.com/trsaba.html.
Downloaded by [University of Nebraska, Lincoln] at 13:56 03 November 2014
