Full-Scale Impact Test of Four Traffic Barriers on Top

of an Instrumented MSE Wall

K.-M. Kim1; J.-L. Briaud2; R. Bligh3; and A. Abu-Odeh4

Abstract: This paper presents the results of four full-scale impact tests against barriers placed on top of an instrumented mechanically
stabilized earth 共MSE兲 wall. The impact was created by a head-on collision of a 2,268-kg bogie going at about 32.2 km/h. The barriers
were New Jersey and vertical wall barriers with a 1.37-m-wide moment slab in 9.14-m-long sections. The wall was 1.52 m high with one
panel and two layers of reinforcement. The reinforcement was 2.44-m-long strips, 4.88-m-long strips, and 2.44-m-long bar mats. The
backfill was crushed rock. The instrumentation consisted of accelerometers, strain gauges, contact switch, displacement targets, string
lines, and high-speed cameras. The test was designed to represent a commonly used installation in current practice including an impact
load on the barrier at least equal to 240 kN. Most of the barriers sustained significant damage but overall the behavior of the wall was
satisfactory since the displacements of the panels were minimal 共less than 25 mm兲 and the panel damage was acceptable except possibly
in the case of the 4.88-m-long strips. The loads measured in the reinforcement indicate that the reinforcement was brought to its ultimate
capacity for the duration of the impact but since the impact duration was so short and since the displacements of the panels were within
tolerable limits of 25 mm, this is considered acceptable. The use of the longer strips 共4.88-m-long strips兲 leads to slightly smaller panel
displacements and higher panel stresses as evidenced by a bending crack in the panel. The 2.44-m-long strips permitted more displace-
ment of the wall panels, but the magnitude of the displacement was considered to be tolerable. The measured maximum dynamic loads
in the strips were found to be 3–5 times higher than the calculated maximum static loads by AASHTO guidelines.
DOI: 10.1061/共ASCE兲GT.1943-5606.0000232
CE Database subject headings: Impact tests; Barriers; Highway and road structures; Pullout; Reinforcement; Walls.
Author keywords: Impact text; Traffic barrier; Pullout; Reinforcement; Mechanically stabilized earthwalls.

Introduction 共bottom part of the L兲 共Briaud et al. 2008兲. The impact load also
generates forces in the MSE wall reinforcement and wall panels
Pavements are often built on top of mechanically stabilized earth in addition to the static loads due to gravity.
共MSE兲 walls. The most common case is the case of an MSE wall This paper presents the results of 4 full scale impact tests
supporting the access embankment for an overpass. Because cars against barriers placed on top of an instrumented MSE wall. The
and trucks travel on top of the MSE wall, traffic barriers are impact was created by a head-on collision of a 2,268-kg bogie
going at about 32.2 km/h. After a brief discussion of the design of
required. In the case of a concrete pavement, these barriers are
MSE walls for barrier impact, the results for the four tests are
rigidly tied to the pavement to provide the resistance needed
presented including accelerometers data, dynamic loads in the
when an impact load is generated by an errant car or truck. In the
reinforcement, dynamic and residual displacements of the wall,
case of an asphalt pavement, that resistance is not available and
and damage to the barrier and the wall. Conclusions are reached
the barrier must resist the impact load on its own. In this case an regarding current design practice. Guidelines for the design of
L shaped barrier-moment slab system is used and the resistance is barriers and MSE walls subject to impact loads will be finalized
generated by the inertia force required to lift the moment slab after the last part of study is performed: a full scale Test Level 3
共TL-3兲 vehicle crash test on an instrumented 2.29-m-high MSE
1
Graduate Research Assistant, Zachry Dept. of Civil Engineering, wall.
Texas A&M Univ., College Station, TX 77843-3136. E-mail: k2river@ While the literature on this topic is very limited these two
tamu.edu references are related to the topic. Tamura et al. 共1994兲 describe
2
Professor and Holder of the Buchanan Chair, Zachry Dept. of Civil the behavior of full scale laterally loaded columns embedded
Engineering, Texas A&M Univ., College Station, TX 77843-3136 共corre-
in the middle of a geosynthetic-reinforced MSE wall. Tateyama
sponding author兲. E-mail: [email protected]
3
Research Engineer, Texas Transportation Institute, Texas A&M Univ. et al. 共1994兲 presented the results of full scale laterally loaded
System, College Station, TX 77843-3135. E-mail: [email protected] columns embedded next to the facing of a geosynthetic-reinforced
4
Associate Research Scientist, Texas Transportation Institute, MSE wall.
Texas A&M Univ. System, College Station, TX 77843-3135. E-mail:
[email protected]
Note. This manuscript was submitted on May 4, 2009; approved on
Previous Crash Test of Barrier on Edge of MSE Wall
August 24, 2009; published online on August 28, 2009. Discussion period
open until August 1, 2010; separate discussions must be submitted for
individual papers. This paper is part of the Journal of Geotechnical and In 1982, Terre Armee Internationale 共TAI兲 in France, which is
Geoenvironmental Engineering, Vol. 136, No. 3, March 1, 2010. closely related to the Reinforced Earth Company 共RECO兲 in the
©ASCE, ISSN 1090-0241/2010/3-431–438/$25.00. United States, performed a crash test of a barrier on top of an

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / MARCH 2010 / 431

 NORTH SOUTH
6 @ 10' = 60'
Moment slab Moment slab
30' 30'

16'
(a)
Bogie Bogie Bogie Bogie
8'
4.5'

10' 10' 10' 10' 10' 10'

NJ shape Vertical Wall Vertical Wall Vertical Wall
(a) /16-ft Strips /16-ft Strips /8-ft Strips /8-ft Bar mats

(b) (c) NORTH SOUTH

60'
LENGTH OF BARRIERS String Line
6 SPACES @ 10'
Fig. 1. TAI test: 共a兲 test vehicle; 共b兲 barriers; and 共c兲 MSE wall after
Bogie Concrete Pad
(1) (2) (3) (4) 137.2 for teo-system
NJ Vertical Vertical Vertical 61 15.2
16-ft Strip 16-ft Strip 8-ft Strip 8-ft Bar mats 25.4
test 1.9 5.1
20.3 77.5

1:1 TYP.
74.9

MSE wall 关Reinforced Earth Company 共RECO兲 1995兴. The test 59'-10 9/16"
LENGTH OF WALL PANELS
15.2 x 30.4 (TYP.) 67.3

vehicle was a 12,000-kg bus which impacted the barrier at a (b)

8.0' or 16.0'
STEEL STRIPS OR BAR MATS LENGTH
(c)
121.9
189.2
244 or 488

speed of 19.7 km/h and an angle of 20°. The impact was esti-
mated to be 30% larger than a Performance Level 2 AASHTO Fig. 2. Instrumented test MSE wall: 共a兲 plan view; 共b兲 elevation
loading condition 共AASHTO 2002兲. The barrier was a New Jer- view; and 共c兲 side view
sey 共N.J.兲 shape barrier approximately 0.81 m high. The barrier
reinforcement was minimal, consisting of two longitudinal Num-
ber 4 bars. The precast barrier units were 1.52 m long and tied to
the moment slab through rebars. The moment slab was cast in
place with a joint every 9.15 m. The width of the moment slab 共AASHTO 2004兲 was applied to the experimental instrumented
was 1.25 m, and its thickness was 25.4 cm. The 25.4 cm of cover MSE wall 共Fig. 2兲 to compare the forces expected on the rein-
over the moment slab consisted of compacted soil and a layer of forcement strips due to the impact loads to those measured in the
bituminous mix. dynamic impact experiments.
The MSE wall was 3.05 m high with two rows of 1.52-m-tall In AASHTO LRFD the following equation is used to calculate
panels. The reinforcement strips were 5 m long and the layers of the load T expected in each strip due to the soil weight and the
strips were located at depths of 38 cm and 1.14 m below the impact load:
bottom of the moment slab 共best guess兲 and were 76.2 cm apart in
the horizontal direction 共best guess兲. A horizontal gap of 1.9 cm
was purposely left between the coping and the traffic face of the T = At ⫻ 共␴h + ⌬␴h,max兲 共1兲
wall panels to avoid lateral contact with the wall panel during
impact. where At = panel tributary area of one strip; ␴h = horizontal stress
The test was considered successful. The bus was redirected due to the soil weight 共␴h = Kt ⫻ ␴v兲; Kr = horizontal earth pressure
and stayed upright. The barrier was damaged but the wall and the coefficient given by 1.7Ka; ⌬␴h,max = 2Ph1 / l1 = horizontal stress
moment slab were not damaged. The upper part of the barrier was due to the impact load Ph1 on the barrier; and l1 = depth of influ-
broken over a length of 2.2 m and a height of 50.8 cm. The top ence of the impact load down the wall face 关AASHTO 共2004兲,
panel of the wall moved 5 mm dynamically during the event and Fig. 3.11.6.3-2 a兴.
had 1.5 mm of residual movement after the impact. The bottom In the 2.43-m-long strip case, unfactored resistances were
panel did not move. No wall damage occurred. The maximum calculated to be 6.6 kN 共Fⴱ = 1.837兲 at the uppermost layer and
decelerations on the front and rear axles of the bus were 8 g 12.02 kN 共Fⴱ = 1.674兲 at the second layer. A density of three strips
共moving average兲 and 14 g, respectively. The maximum dynamic per layer per panel was used. The unfactored load per strip due
force recorded on the most loaded strip was 28.91 kN. to gravity was calculated to be 2.53 kN at the uppermost layer
The minimum reinforcement density for MSE walls gives a and 4.87 kN at the second layer. In this analysis, the traffic sur-
resistance of 42.3 kN/m of wall on the top layer of strips. Pulling charge was not considered. The unfactored load per strip due to
the strips out of the wall would require movement of the moment the impact was calculated to be 1.9 kN at the uppermost layer and
slab unit. For a joint spacing of the moment slab equal to 6.1 m, 1.3 kN at the second layer. Therefore, the total unfactored load
the maximum load that the strips can resist at impact is 6.1 m per strip was 4.43 kN at the uppermost layer and 6.16 kN at the
⫻ 42.3 kN/ m = 258 kN 共static兲. The 1982 TAI test leads to a load second layer. The ratios between the load and resistance are 1.49
of 28.91 kN⫻ 6.1 m / 0.76 m = 231.3 kN 共dynamic兲 if all strips at the uppermost layer and 1.95 at the second layer.
within the 6.1-m section of barrier and moment slab were stressed In the 4.88-m-long strip case, unfactored resistances were cal-
at the maximum observed value. 258 kN 共static resistance兲 is culated to be 13.19 kN 共Fⴱ = 1.837兲 at the uppermost layer and
much higher than the 44.5 kN 共static兲 required by AASHTO. 24.04 kN 共Fⴱ = 1.674兲 at the second layer. A density of two strips
Therefore RECO concludes that the minimum reinforcement den- per layer per panel was used. The unfactored load per strip due
sity is adequate to resist the impact load. Fig. 1 shows the damage to gravity was calculated to be 3.8 kN at the uppermost layer
of the test vehicle and barriers after test completed. and 7.31 kN at the second layer. In this analysis, the traffic sur-
charge was not considered. The unfactored load per strip due to
Design of MSE Wall for Barrier Impact: Current the impact was calculated to be 2.85 kN at the uppermost layer
Practice and 1.94 kN at the second layer. Therefore, the total unfactored
load per strip was 6.65 kN at the uppermost layer and 9.25 kN at
The load and resistance factor design 共LRFD兲 procedure outlined the second layer. The ratios between the load and resistance are
in the AASHTO Standard Specifications for Highway Bridges 1.98 at the uppermost layer and 2.6 at the second layer.

432 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / MARCH 2010

 100% dowels. The moment slabs were 1.37 m wide from the back face
90% of the wall panel to the end of the moment slab.
80% Two different barrier types were used in the test plan: a N.J.
70% shape 共Test 1兲 and a vertical wall barrier 共Tests 2, 3, and 4兲. Tests
Percent passing (%)

60% 1 and 4 were conducted over the portion of the wall with 4.88-m
50% steel strip reinforcement. Tests 2 and 3 involved impacts into
40%
vertical wall barrier placed over wall segments with 2.44-m steel
strip and bar mat reinforcement, respectively 共Table 1兲.
30%
The reinforcement was instrumented with strain gauges to cap-
20%
ture the tensile forces transmitted into the reinforcement during
10%
each bogie vehicle impact. A total of eight strain gauges were
0% used for each test. The placement of these strain gauges was
0.01 0.10 1.00 10.00 100.00
selected to measure the maximum tensile load in each layer of
Gain size (mm)
reinforcement as well as give an indication of the distribution of
forces in lateral, longitudinal, and vertical directions. Five strain
Fig. 3. Particle-size distribution curve
gauges were used on the upper reinforcement layer, and three
strain gauges were placed on the lower reinforcement layer. Two
strain gauges were used on both layers of reinforcement adjacent
Design, Instrumentation, and Construction to the wall panel at the point of impact to provide some redun-
of the Test Wall dancy at the location expected to experience maximum tensile
loading. A contact switch was placed on the inside face 共traffic
The objectives of the tests were to quantify the movement of the side兲 of the concrete leveling pad cast on top of the wall panels to
barrier, the moment slab, and the wall panels during impact, and indicate the time at which the barrier came in contact with the
to measure the barrier impact force and the forces in the rein- wall panel.
forcement during impact. The wall panel below the point of impact on each barrier was
The wall barrier system was planned on the premise that mul- instrumented with five concrete strain gauges to capture normal
tiple impacts could be conducted on barrier segments connected strains in the panel at impact. Two of the five strain gauges were
to the same moment slab. An elevation of the test installation is placed in a horizontal direction just below the anchor point of
shown in Fig. 2共a兲. Half of the wall was constructed using two the upper layer of reinforcement 共region of maximum negative
types of 2.44-m-long reinforcement 共strips and bar mats兲 while moment兲. The other three strain gauges were placed in the vertical
the other half was constructed with 4.88-m-long reinforcement direction: two were adjacent to the anchorage locations for the
strips. The strips were 50 mm wide and 5 mm thick. Each layer of upper and lower layer of reinforcement at the point of impact,
bar mats consisted of two sections of 6 ⫻ 6-W 10⫻ W 10 bar and one was placed between in the middle of the panel between
attached to the wall panel using clevis loops. The 2.44-m-long the two layers of reinforcement 共region of maximum positive
reinforcement represents the minimum length allowed in current moment兲.
practice. Such lengths are commonly used in low height wall An accelerometer was mounted behind each barrier section at
segments such as at the beginning or ending of an elevated over- the height of impact to help analyze its dynamic response. An-
pass structure. At the minimum 2.44-m length, current design other accelerometer was placed at the end of each of the two
procedures typically require a density of six reinforcement strips 9.14-m-long moment slabs at their midpoints to measure any ac-
per wall panel 共three in each of two different horizontal layers of celeration imparted to the moment slab during impact. Addition-
reinforcement兲. The other half of the wall was constructed using ally, the bogie vehicle was instrumented with an accelerometer.
4.88-m-long reinforcement strips to quantify the effect of strip Rotation as well as vertical and horizontal displacements of
length. The density of the 4.88-m-long strips was four per panel. the barrier and wall panels were determined from high-speed
The backfill for the wall was crushed rock that met the speci- video operating at 1,000 frames/s. Displacement gauges were
fications for Texas DOT Type A backfill 共Texas DOT 2004兲. The placed at the top and bottom of the precast barrier-coping section
estimated friction angle for the crushed rock was 35° and the unit and at the upper and lower strip locations on the wall panel to
weight was 20 kN/ m3. The backfill was compacted in 0.15-m assist with the displacement analysis.
layers with 10 passes of a 1,320-kg, 89-cm wide drum roller. Independently secured string lines were placed 4 ft behind the
Also, the surface layer of soil was recompacted after each test. A barrier and wall to measure the permanent deflection of the bar-
grain size analysis was performed for the backfill material to de- rier and wall after impact. The permanent movement of the four
termine the relative proportions of different grain sizes as shown corners of the back of the barrier and the permanent movement of
in Fig. 3. The particle diameters corresponding to 10% fines, D10, five points on each panel were measured after each test.
and 60% fines, D60, were 0.075 and 6.8 mm, respectively. The The test sequence was selected such that the first two tests
coefficient of uniformity, Cu 共=D60/ D10兲 was determined to be involved impacting the barrier segments in the middle of each
90.67; therefore, the friction factor, Fⴱ, at ground level was deter- moment slab 共one N.J. barrier and one vertical wall兲. The other
mined to be 2.0 in accordance with AASHTO LRFD 关AASHTO two tests were conducted on vertical concrete barrier located
共2004兲, Fig. 11.10.6.3.2-1兴. at the end of each moment slab with different strip length and
Six precast barrier and coping sections were placed on top of density.
the wall panels. The panels were recessed inside the coping sec- A 2,268-kg bogie vehicle impacted each test section at a speed
tions a distance of 0.23 m comprised of 0.1 m of leveling concrete of approximately 35.4 km/h for the N.J. barrier and 32.2 km/h for
pad plus 0.13 m of engaged panel. The moment slab connecting the vertical concrete barriers. Loading each barrier near the ulti-
the precast barrier-coping sections was cast in two 9.14-m lengths mate load of the barrier ensured that the maximum impact load
connected to one another using two 0.91-m-long Number 9 shear was transferred into the MSE wall.

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / MARCH 2010 / 433

Table 1. Test Matrix and Results
Test 1 Test 2 Test 3 Test 4a
Test installation
Barrier type New Jersey Vertical wall Vertical wall Vertical wall
Reinforcement 4.88-m-long strip 2.44-m-long bar mat 2.44-m-long strip 4.88-m-long strip
共four per panel兲 共four per panel兲 共six per panel兲 共four per panel兲
Speed of bogie 35.08 km/h 32.67 km/h 32.49 km/h 32.49 km/h
Test results
Peak acceleration
Bogie 14.45 g 13 g 13.82 g 12.69 g
Barrier 7.36 g 10.71 g 10.16 g 13.04 g
Moment slab 1.84 g N/A 1g N/A
Impact force 326.5 kN 294 kN 312.13 kN 286.55 kN
Displacement
Top of barrier
Dynamic 156 mm 153.42 mm 131.32 mm 152.91 mm
Permanent 76.2 mm 101.6 mm 63.5 mm 76.2 mm
Bottom of coping
Dynamic 28.45 mm 23.62 mm 29.46 mm 17.53 mm
Permanent 14 mm 12.7 mm 15.24 mm 5.6 mm
Panel 共upper layer兲
Dynamic 16 mm 9.4 mm 23.37 mm 7.62 mm
Permanent 6.1 mm 5.1 mm 14 mm 1.78 mm
Panel 共second layer兲
Dynamic No movement 2.54 mm 4.83 mm 1.78 mm
Permanent No movement 0.51 mm 4.57 mm No movement
Load in strips
Upper layer
Max. 50 msec 32.16 kNb 6.85 kNb 9.47 kNb 33.18 kNb
Design load 共kN兲 23.53 kN 7.47 kN 7.3 kN 27.8 kN
Design load 共kN/m兲 31.65 kN/mc 31.21 kN/mc 14.73 kN/mc 37.48 kN/mc
Second layer
Max. 50 msec ⫺5.34 kNb 0.36 kNb 5.29 kNb 0.67 kNb
Design load 共kN兲 ⫺3.91 kN 0.36 kN 4.09 kN 0.58 kN
c c c
Design load 共kN/m兲 ⫺5.25 kN/m ⫺1.75 kN/m 8.31 kN/m 0.73 kN/mc
d d d
Total design load 26.4 kN/m 32.96 kN/ft 23.04 kN/m 38.21 kN/md
a
Test Section 4 was between Test 1 and Test 3 and Test 4 was carried out after Test 1 and Test 3. Residual deformations from Test 1 and Test 3 may have
influenced the results of Test 4.
b
Average of top and bottom load per one strip.
c
Average of top and bottom load per unit width of panel.
d
Total design load of strips in upper and second layers per unit width of panel.

Impact Tests Results

Four impact tests were carried out. Each time the 2,268-kg bogie
vehicle, shown in Fig. 4, impacted the barrier head-on at a speed
ranging from 32.5 to 35 km/h 共Table 1兲. The impact point for each
test is shown in Fig. 2. Note that while the impact point was close
to the center of the barrier, it was not exactly at the center because
(a)
it was desirable to position the impact point directly above the
reinforcement that was instrumented to record maximum impact
forces. The test results include accelerometer data and associated
forces and displacements 共bogie, barrier, and moment slab兲,
high-speed film analysis, string line measurements, and associated
displacements 共barrier and wall panel兲, strain gauges and associ- (b)
ated forces in the reinforcement, and accumulated damage of the
barrier, soil, and wall panels. Fig. 4. MSE wall installation: 共a兲 before; 共b兲 after test

434 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / MARCH 2010

 gets are shown in Fig. 7. The maximum dynamic horizontal dis-
placement at the top of the barrier varied from 131 to 156 mm.
The maximum dynamic horizontal displacement at the bottom of
the barrier varied from 18 to 30 mm. The maximum dynamic
horizontal displacement of the panel at the top layer of reinforce-
ment varied from 8 to 23 mm. The maximum dynamic horizontal
displacement of the panel at the bottom layer of reinforcement
varied from 0 to 5 mm.
The permanent movements of the target locations were ob-
Fig. 5. Impact force from bogie acceleration data tained in two ways: high-speed film analysis and distances from
the reference string line stretched in front of the wall. The string
line permanent measurements consisted of measuring the distance
Data from Accelerometers and Associated Forces
with a tape measure from the target to the string before and after
and Displacements
each test. The permanent horizontal displacement at the top of the
Data obtained from the bogie mounted accelerometer were ana- barrier varied from 63 to 102 mm. The permanent horizontal dis-
lyzed and a 50-ms moving average was generated. Based on the placement at the bottom of the barrier varied from 6 to 15 mm.
acceleration and the mass of the bogie, the impact force was The permanent horizontal displacement of the panel at the level
calculated as a function of time for the four tests 共Fig. 5兲. Note of the top row of reinforcement varied from 1.78 to 14 mm. The
that the peak impact forces vary from 286.55 to 324.72 kN and permanent horizontal displacement of the panel at the level of the
are all higher than the 240-kN dynamic force associated with the bottom row of reinforcement varied from 0 to 4.57 mm.
design of barriers for TL-3 and TL-4 levels. Even though the wall systems were subjected to higher than
Data obtained from the accelerometer mounted at the top of design conditions, all movements were considered acceptable
the barrier were analyzed and a 50-ms moving average was gen- from a performance point of view. It is thought that a permanent
erated 共see Fig. 6兲. The horizontal acceleration-time history was displacement of 25 mm or less is acceptable for these structures;
integrated to obtain horizontal velocity versus time. The however the integrity of the structural elements must also be con-
displacement-time history at the top of the barrier was obtained served after the impact. The wall system comprised of the 2.44-m
through double integration of the acceleration history. The strip reinforcement 共Test 3兲 lead to the highest panel movements,
displacement-time history at the top of the barrier was also ob- while the lowest movements were recorded for the configuration
tained through analysis of the high-speed film. Note that for Tests that incorporated 4.88-m strips and the vertical wall parapet 共Test
1 and 2, the acceleration measurements were truncated when the 4兲. However, the Test 4 configuration also had the most extensive
acceleration exceeded the range of the device. Therefore, the panel damage. In Test 4, the top panel exhibited a horizontal
high-speed film displacements are more reliable than the accel- hairline fracture crack along a line corresponding to the location
eration derived displacements for these two tests. Overall the of the top layer of reinforcement. It is possible that this damage
horizontal displacement at the top of the barrier based on the film occurred as a result of accumulated and repeated movement of the
analysis varied from 131 to 156 mm. coping, thereby decreasing the clearance between the coping and
Data obtained from the accelerometer mounted at the edge of the panel. As a result the coping engaged the panel earlier in the
the moment slab were analyzed and a 50-ms moving average was dynamic event and may have led to higher load.
generated. The vertical displacement-time histories of the moment
slab were obtained by double integration of the acceleration data
for the four tests. The vertical displacements of the edge of the Load in Reinforcement Strips
moment slab were 12.7 and 24.1 mm for Test 1 and Test 3, re-
spectively. The data for Tests 2 and 4 are not reliable because the The wall reinforcement was instrumented with a total of 32 strain
accelerometers were damaged in the previous tests. gauges 共8 for each test兲 to capture the tensile forces transmitted to
the reinforcement during the dynamic bogie vehicle impacts. The
eight strain gauges were placed on six strips in two locations:
High-Speed Film Analysis and Associated 0.18 and 1.09 m behind the connection to the panel. Two strips
Displacements receiving a pair of gauges 共top and bottom兲 and the other four
Targets were affixed to the top of the barrier, the bottom of the strips received only one strain gauge 共top only兲. Because a fair
barrier, and the wall panel at a location corresponding to the top amount of bending unexpectedly took place in the strips during
and the bottom layers of reinforcement. The displacement-time the impact, the positions which had only one gauge did not lead to
histories obtained from high-speed film analysis of the four tar- reliable data. This bending is believed to be due to a combination
of axial tension and downward movement of the strips-panel con-
nections near the face of the wall. The positions which had two
gauges permitted cancellation of the bending stresses so that the
true tensile load in the strip could be obtained. As a result there
are only two strip loads reported for each test in Fig. 8. For the
bar mats, the strain gauges were placed on the bars in similar
fashion. The load-time history in the reinforcement is a 50-ms
average. Note that the strain gauges were zeroed after the MSE
wall was built and, therefore, the loads shown do not include the
static load necessary to retain the soil mass.
The maximum dynamic load in the strips 共50-ms average兲 var-
Fig. 6. Acceleration at top of barrier ied from 6.85 kN 共Test 2兲 to 33.18 kN 共Test 4兲. The higher loads

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / MARCH 2010 / 435

 Fig. 7. Barrier and panel displacement 共film analysis兲: 共a兲 Test 1; 共b兲 Test 2; 共c兲 Test 3; and 共d兲 Test 4

were associated with the 4.88-m-long strips while the lower loads glance. The reason they are higher than unity is that the strips
were associated with the 2.44-m-long reinforcement. have ribs which engage the bearing capacity of the surrounding
The back-calculated Fⴱ values ranged from 2.58 to 4.51. These soil in addition to the friction. So, the pull out of strips is not a
measured instantaneous values are much higher than the design pure friction phenomenon. The bearing capacity effect is also
recommendation for this case which is 1.837. This is not unusual lumped into the recommended friction factor. Note also that the
since the recommended values represent a lower bound of the sum of the strip forces does not have to add to the impact force
values measured by many researchers. Note that these friction because other forces contribute to the resistance including the
values are much higher than might seem appropriate at first barrier inertia force and the barrier-soil friction force.
A tape switch was adhered to the back face of the leveling pad
extension on top of the wall panels. The idea was to identify
324.7 kN Top Layer 294 kN whether or not the barrier-coping contact and engage the panel
240 kN Bottom Layer 240 kN Top Layer
8
Bottom Layer during an impact event. The contact duration is indicated in Fig. 8
30
6 by an arrow labeled “tape switch.” Note that sometimes there are
Load (kN)

Load (kN)

20
Tape Switch 4 two time periods over which the contact took place indicating that
10 2 Tape Switch a bouncing phenomenon developed.
0 0 As previously noted, the peak dynamic loads against the bar-
-10 -2 riers were higher than 240 kN which is the design load prescribed
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.00 0.05 0.10 0.15 0.20 0.25 0.30 by AASHTO. To obtain the load on the strips for an impact force
(a) Time (sec) (b) Time (sec)
equal to 240 kN, the loads were reduced by the ratio between the
312 kN
240 kN
287 kN
240 kN Top Layer
peak dynamic load actually measured and the 240-kN load.
Top Layer
10 Bottom Layer
30
Bottom Layer Therefore, for Test 1 for example, the load in the reinforcement
8
6
corresponding to a 240-kN barrier impact load would be obtained
Tape Switch
Load (kN)

Load (kN)

20
4
Tape Switch
as follows:
2 Tape Switch
10
0
-2
Tape Switch 0
-4 estimated strip load for 240 kN = 240/326.5
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.00 0.05 0.10 0.15 0.20 0.25 0.30
(c) Time (sec) (d) Time (sec) ⫻ maximum measured strip load 共2兲
This linear proportionality is made possible by the fact that no
Fig. 8. Load in the reinforcements 共50-ms average兲: 共a兲 Test 1; 共b兲 rate effect was found on the strip capacity as shown below. Note
Test 2; 共c兲 Test 3; 共d兲 and Test 4 that the peak dynamic impact load on the barrier occurred before

436 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / MARCH 2010

 100 els. This spalling took place in all four tests. In addition, in Test 4
Strip, Unsaturated

Pull-out Force (kN, log)

Bar mat, Unsaturated
共vertical wall with 4.88-m-long strips兲, a flexural crack developed
horizontally across the wall panel at the level of the first row of
reinforcement 关Fig. 4共b兲兴. While the decision to replace MSE wall
10
panels after an impact should be left to the responsible parties, it
is felt in this case that none of the panels would need to be
replaced with the exception of the cracked panel in Test 4. The
1 4.88-m strips can resist a larger load than the 2.44-m strips but
0.01 1 100 10000 represent a much stiffer system which can lead to the damage of
Time (sec, log) the wall panel.
Observation of any damage to the moment slab, the front side
Fig. 9. Pull-out load at failure versus time to failure of the coping, and the wall reinforcement would require proper
excavation and inspection. Although this has not been done, there
is no external evidence of any failure of these components. The
the maximum load in the strips. This seems to indicate that there high-speed films did indicate that the edge of the moment slab
is a lag time between the barrier movement and the soil deforma- lifted during impact. A small gap could be seen after the tests at
tion which engages the strip loads. the interface between the moment slab and the soil behind it.
The second row of reinforcement experienced much lower Furthermore for the tests near the connection between the two
loads including compression loads in some cases. The reason for 9.14-m-long moment slabs 共Tests 3 and 4兲, the film showed that
this is attributed to the rotation of the panel around the first an- the edge of the moment slab lifted over the entire 18.29 m indi-
chor point provided by the top row of reinforcement. This rotation cating that the dowels engaged both 9.14-m lengths.
creates a possible inward movement of the bottom of the panel.
This compression can be resisted by a combination of strip load
and by pressure against the soil mass. Conclusions

The following conclusions are based on the four dynamic impact

Rate of Loading Effect on Pull-Out Reinforcement tests performed in this study:
Capacity 1. The test was designed to represent a commonly used instal-
A separate study was performed in the laboratory to evaluate the lation in current practice including a 1.37-m-wide moment
possible increase in reinforcement resistance associated with an slab and an impact load on the barrier at least equal to
increase in pull-out rate. The experiments were conducted in a 240 kN. Most of the barriers sustained significant damage
box filled with moist uniform sand 共D50= 0.4 mm兲. The box was but overall the behavior of the wall was satisfactory since the
0.7 m wide by 0.37 m high by 1.31 m long. A 1.31-m-long displacements were small and the panel damage was accept-
strip was placed in the center of the box and a dead weight was able 共except possibly for one case with 4.88-m-long strips兲;
placed on top of the surface to generate 14.36 kPa of vertical 2. The loads measured in the reinforcement indicate that the
pressure on the strip to represent soil overburden. Three tests reinforcement was brought to its ultimate capacity for the
were performed where the strip was pulled to failure at rates such duration of the impact 共⬃100 ms兲. Because the impact du-
that the times to failure were 0.05, 5, and 1,287 s. The results ration was so short and because the displacements were
showed that there was no significant rate effect on the strip ulti- within tolerable limits, this is considered acceptable. There-
mate capacity 共Fig. 9兲. Another three tests were performed on bar fore the reinforcement can be designed to be at failure for
mats with similar results. this condition 共factor of safety of 1兲;
3. While the use of the longer strips 共4.88-m-long strips兲 led to
slightly smaller panel displacements, it also lead to higher
Component Damage
panel stresses 共a bending crack was observed in one test兲.
In all tests, the most severe damage occurred in the barriers. In a The 2.44-m-long strips appear to be a more flexible system
real situation, these barriers would have to be replaced. However, placing less stress in the panels while still resulting in accept-
it should be noted that the impact load was designed to fail the able movements;
barrier and exceeded the 240-kN design load. The N.J. barrier in 4. When comparing the behavior of the strips and bar mats, it
Test 1 failed along a horizontal line located at the change in was observed that the barrier displacement was about the
thickness of the cross section with some evidence of V-shaped same but the panel displacement was about three times less
cracks. The vertical wall barrier in Test 2 failed along a combi- for the bar mats than for the strips. The load in the strips and
nation of half V half horizontal line; the reason is that the point of the bar mats during impact per tributary area of panel was
impact was closer to one edge of the 3.05-m-long barrier to po- about the same;
sition the point of impact over one of the bar mats. The vertical 5. The measured maximum dynamic loads in the strips are three
wall barrier in Test 3 failed in a V shape pattern with a horizontal to five times higher than the calculated maximum static loads
line at its base and so did the vertical wall barrier in Test 4. None by AASHTO guidelines. Since the rate effect study showed
of the barriers failed in the assumed V shape associated with a no practical difference between 0.05 and 1,200 s, the rate
traditional yield line failure but rather in a V shape combined with effect cannot be the reason for the three to five times ratio.
a horizontal flexure mode. That ratio is explained by the fact that the AASHTO guide-
The damage to the MSE wall panels consisted of spalling lines are based on Fⴱ values 共friction coefficient between the
around the connection between the top of the panel and the bot- strips and the soil兲 which are lower bounds of measured val-
tom of the barrier. This is attributed to the shearing process that ues, and the crushed rock used as backfill in the experimental
broke the bond between concrete leveling pad and the wall pan- study is probably one of the highest friction generating back-

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / MARCH 2010 / 437

 fills available. Further, the design values have a factor of References
safety of 1.5, while the measured values are ultimate values;
and AASHTO. 共2002兲. Standard specifications for highway bridges, 17th Ed.,
6. The AASHTO guidelines are being revised in light of these AASHTO, Washington, D.C.
tests and a full scale TL3 test and being published in Bligh AASHTO. 共2004兲. AASHTO LRFD bridge design specifications, 3rd Ed.,
et al. 共2009兲. AASHTO, Washington, D.C.
Bligh, R. P., Briaud, J. L., Kim, K. M., and Abu-Odeh, A. 共2009兲. “De-
sign of roadside barrier systems placed on MSE retaining walls.”
Acknowledgments Revised Final Rep. Prepared for National Cooperative Highway Re-
search Program Project 22-20, Transportation Research Board, Na-
The content of this paper is the partial result of a project spon- tional Research Council, performed by Texas Transportation Institute,
sored by the National Cooperative Highway Research Program College Station, Tex.
共Grant No. NCHRP 22-20兲. The opinions expressed in the paper Briaud, J.-L., Bligh, R., Abu-Odeh, A., and Kim, K. 共2008兲. “Stability
are those of the writers and not necessarily those of NCHRP. analysis and full scale test of a traffic barrier-moment slab system
bridges.” Transp. Res. Rec., 2050, 26–38.
The writers thank the NCHRP panel members for their input and
Reinforced Earth Company 共RECO兲. 共1995兲. “Crash testing of a precast
in particular Mark McClelland, chair of the panel. The writers
traffic barrier atop a reinforced earth wall.” Technical Rep. No.
also thank the Reinforced Earth Company 共Pete Anderson, John
MSE-8, Va.
Sankey兲 for providing components for construction of the test Tamura, Y., Nakamura, K., Tateyama, M., Murata, O., Tatsuoka, F., and
wall, sharing data, and answering many questions. The writers Nakaya, T. 共1994兲. “Full-scale lateral loading tests of column founda-
also thank Foster Geotechnical 共William Neely兲 for sharing their tions in geosynthetic-reinforced soil retaining walls.” Recent case his-
experience on MSE walls. A special note of thanks to Geotesting tories of permanent geosynthetic-reinforced soil retaining walls,
Express in Atlanta for conducting the rate effect study on pull-out Tatsuoka and Leshchinsky, Balkema, 277–286.
resistance of the wall reinforcement. Tateyama, M., Murata, O., Tamura, Y., Tatsuoka, F., and Nakaya, T.
The precast barrier-coping sections, concrete wall panels, and 共1994兲. “Lateral loading tests on columns on the facing of
steel strip wall reinforcement were provided by RECO at no cost geosynthetic-reinforced soil retaining wall.” Recent case histories of
to the project. RECO also provided supervision of the construc- permanent geosynthetic-reinforced soil retaining walls, Tatsuoka and
tion of the wall. The bar mat reinforcement that was used in one Leshchinsky, Balkema, 287–294.
of the tests was provided by Foster Geotechnical at no cost to the Texas DOT. 共2004兲. “Standard specifications for construction and main-
project. tenance of highways, streets, and bridges.” Austin, Tex.

438 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / MARCH 2010

Copyright of Journal of Geotechnical & Geoenvironmental Engineering is the property of American Society of
Civil Engineers and its content may not be copied or emailed to multiple sites or posted to a listserv without the
copyright holder's express written permission. However, users may print, download, or email articles for
individual use.