Innovative Infrastructure Solutions (2019) 4:49

https://doi.org/10.1007/s41062-019-0236-x

STATE-OF-THE-ART PAPERS

Geosynthetic‑reinforced soil structures for railways and roads:

development from walls to bridges
Fumio Tatsuoka1

Received: 6 September 2019 / Accepted: 18 September 2019 / Published online: 1 October 2019
© Springer Nature Switzerland AG 2019

Abstract
The development and construction of various types of geosynthetic-reinforced soil (GRS) structures for railways, roads and
others, mainly for railways, for the last 35 years in Japan is described. In the 1980s, GRS retaining wall (RW) with full-
height rigid (FHR) facing was developed. The FHR facing is constructed firmly connected to reinforcement layers after the
reinforced backfill and subsoil has deformed sufficiently. In the early 1990s, GRS Bridge Abutment supporting one end of
a simple girder at the top of the FHR facing was developed. In the early 2000s, GRS Integral Bridge was developed, which
structurally integrates both ends of a continuous girder to the top of the FHR facings of a pair of GRS RWs. The total wall
length of these GRS structures became about 185 km by June 2019 with no problematic case. The use of FHR facing, the
staged construction of FHR facing and the structural integration for GRS Integral Bridge are the three major breakthroughs
for the development of these GRS structure technologies.

Keywords Bridge · Geosynthetic-reinforced soil · Rigid facing · Transportation · Wall

Introduction
This paper reports the development and construction for the
last 35 years of various types of geosynthetic-reinforced soil
(GRS) structures mainly for railways (including high-speed
railways, HSRs, Shinkansen) and partly for roads and oth-
ers in Japan. The development started with retaining walls
(RWs) and finally has been extended to bridge structures.
Based on researches that started in the beginning of 1980s,
GRS retaining wall (RW) with staged–constructed full-
height rigid (FHR) facing (Fig. 1a) was proposed at the end
of the 1980s [22, 24, 26]. It has been confirmed that the use
of FHR facing is essential for high wall stability with small
deformation against mechanical, hydraulic and fire attacks
to the wall face and against long-term traffic loads, severe
seismic loads, heavy rains, floods and so on. As shown in
Fig. 1a, after the deformation of backfill and subsoil has
taken place sufficiently, the FHR facing is constructed by
casting-in-place concrete directly on the temporary wall face
comprising gravel-filled bags wrapped around with geogrid
* Fumio Tatsuoka
tatsuoka@rs.noda.tus.ac.jp
1
Tokyo University of Science, 2641 Yamazaki, Noda City,
Chiba Prefecture 278‑8510, Japan
reinforcement. In this way, the FHR facing and the reinforce-
ment layers are firmly connected to each other while the con-
nection is not damaged by relative settlement between them
that would take place if the facing were constructed before or
simultaneously with wall construction. By June 2018, GRS
structures with FHR facing have been constructed for a total
length of about 185 km, all successful with no problematic
case during and after construction (Fig. 2).
Polymeric reinforcement (i.e., geosynthetic reinforce-
ment) is more extensible exhibiting a stronger trend of creep
deformation than metallic reinforcement (often called inex-
tensible reinforcement). However, it has been confirmed that
the backfill soil, which has essentially no strength and stiff-
ness in tension, can be effectively tensile-reinforced with
polymeric reinforcement, while post-construction creep
deformation of GRS structure can be effectively restrained
by proper structural design, good backfill compaction and
proper drain [30].
FHR facing can support other structures (e.g., noise bar-
rier walls, crash barrier walls, electric power supply facili-
ties, etc.) at its top. Taking advantage of this feature, Tat-
suoka et al. [31] proposed GRS Bridge Abutment, which
supports one end of a simple girder at the top of the FHR
facing of a GRS RW, after having attempted several other
types of GRS abutment and pier [18, 19, 27, 45, 46]. More

13
Vol.:(0123456789)
49 Page 2 of 18 Innovative Infrastructure Solutions (2019) 4:49

(a)

(a)
2,900 3,913 2,910
1,173 0.300

Yamanote line
21 x 300=6,300

FHR facing
384

41,484

Geogrid
6,914

Chuo
line
230
200 100

34,570 Gravel-filled bags

3,000 2,000 2,500
(all units in mm)
640 1,000

(b)

Fig. 1  a Staged construction of GRS RW with FHR facing; and b a

typical GRS RW near Shinjuku station, Tokyo, constructed during
1995–2000
(b)

than 60 GRS Bridge Abutments have been constructed or
are under construction. Extending this technology, Tatsuoka
et al. [33] proposed GRS Integral Bridge (Fig. 3), which
structurally integrates both ends of a continuous girder,
without using girder bearings, to the top of the FHR facings
of a pair of GRS RWs. Its design and construction codes
have been published [7, 47].
The full contents of this report are described in Tatsuoka
[44].
GRS retaining walls
From wrapped‑around facing to FHR facing
We learned the paramount importance of the stiffness (or
rigidity) of the facing for wall stability from the performance
of five full-scale test embankments that were consecutively
constructed in the 1980s. The wall facing was gradually
improved: from wrapped-around, via soil bags, shotcrete and
discrete panels, finally to full-height rigid (FHR) (Fig. 1a).
The first test embankment (Fig. 4) was constructed
in June 1982 at Chiba Experiment Station of Institute
Fig. 2  a Locations; and b statistics of GRS structures with FHR fac-
ing constructed as of June, 2019
of Industrial Science, University of Tokyo, to examine
whether stable near-vertical walls can be constructed
even by reinforcing clayey soil with very flexible rein-
forcement: i.e., a nonwoven geotextile (spun-bonded 100%
polypropylene) that is usually used as a drain material.
The backfill was on-site volcanic ash clay, called Kanto
loam, with a high water content wi = 100–129%; a high
degree of saturation Sr = 8– 90%; and a very low dry den-
sity ρ d = 0.55–0.69 g/cm 3. The wall face was wrapped
around with a nonwoven geotextile and nearly flat when
constructed. The wall on the left side in Fig. 4, in which
the initial vertical spacing between adjacent nonwoven
geotextile sheets was equal to 80 cm, deformed largely
already during construction and much more by heavy
rainfalls after completion. The wall at the other side (on
the right side in Fig. 4), having an initial vertical spacing
between nonwoven geotextile sheets equal to 40 cm, also
deformed noticeably, although it was less than the left-side
wall. We learned from this behavior that large deformation

13
Innovative Infrastructure Solutions (2019) 4:49 Page 3 of 18 49

Structural integration Initial (March 1984)

4. Girder 2. GRS wall Oct. 1986
Highly distorted zone
Soil
3. FHR facing Shear band bags

Firmly connected

Cracks
Non-woven geotextile
1. Ground improvement
(a) (when necessary)

Fig. 5  Cross sections of Chiba No. 2 embankment: 1 year after an

4. Girder
Structurally 2. Geosynthetic- artificial heavy rainfall test [21, 23, 29, 32]
integrated reinforced
3. FHR facing backfill

(a) Discrete concrete panels

(b) 1. Ground improvement (when necessary)

Shotcrete
Soil bags
Fig. 3  Schematic diagram of GRS integral bridge: a elevation; and b Drain hole
plan [33, 37–40]

Oct. 1985 Initial (June 1982)

(b)

Cracks
Non-woven
geotextile

Fig. 4  Cross section of Chiba No. 1 embankment of clayey soil con-

structed at the University of Tokyo [21, 23, 29, 32], observed when
demolished (about 2 years later after the construction)
starting from wall face could be one of the actual serious
problems with wrapped-around GRS RWs.
Chiba No. 2 embankment (Fig. 5) was constructed in
March 1984 using the same types of clayey soil and nonwo-
ven geotextile as No. 1 to evaluate the effects of soil bags
placed at the shoulder of each soil layer that are wrapped
around with the reinforcement on the constructability and
stability of GRS RW. The soil bags were made of the same
nonwoven geotextile as the reinforcement layers and filled
with the same clayey soil as the backfill. The post-con-
struction wall deformation for a period of one-and-a-half
years was much smaller than No. 1. To study the wall fail-
ure mechanism, the walls were brought to failure by sup-
plying in total 70 m3 of water from the crest for a period
of 8 days in October 1985. Although the two walls largely
deformed, in particular the right-side wall having very short
Fig. 6  Chiba No. 3 embankment: a cross sections at the center and at
the rear side; and b deformation of the cross sections at the center and
at the rear side [23, 29, 32]
reinforcement, they did not collapse. It was found that the
use of such soil bags is very effective for both good soil
compaction and high wall stability.
Chiba No. 3 embankment (Fig. 6) was constructed in
1986 also using the same types of clayey soil and nonwoven
geotextile as Nos. 1 and 2 to compare the behaviors of the
GRS RWs having the following three different facing struc-
tures: wrapped around as No. 1 wall (not shown in Fig. 6);
discrete concrete panels (50 cm × 50 cm × 5 cm thick with
a weight of 34 kgf each) connected to the reinforcement
layers w/o soil bags (the right-side wall in Fig. 6); and soil
bags wrapped around by the reinforcement layers and cov-
ered with a 8-cm-thick shotcrete layer (the left-side wall

13
49 Page 4 of 18 Innovative Infrastructure Solutions (2019) 4:49

in Fig. 6). Their behaviors were very different showing the 32.6
1.75
paramount effects of facing structure on the wall stability. 6.85 3 2 3 2 3 1 10

That is, the wall with wrapped-around facing (w/o soil bags)
largely deformed during the subsequent several years after d f h
wall completion. On the other hand, as shown in Fig. 6b, the

6.9
(T) (C) (T) (C) (T) UNIT : m
deformation of the wall with facing of relatively small pre- c e g
cast concrete panels was much smaller. Yet, the deformation
was noticeable, showing that this type of facing is not rigid (T) test section; (C) control section
enough. Besides, it was very difficult to assemble these pan- (a)
els on site and compact backfill soil immediately behind the 36.0
facing ensuring a good wall face alignment. The deforma- 7.5 1 3 2 3 2 3 1 13.5
tion of the wall with facing of clay-filled soil bags wrapped
around with reinforcement of nonwoven geotextile that was c b a

6.9
covered with a shotcrete layer was also much smaller than (T) (C) (T) (C) (T) UNIT:m

the deformation of the rear side wall having wrapped-around

facing (w/o soil bags). Yet, the deformation was noticeable
to be used as a permanent RW, allowing a limited amount
of deformation, while the shotcrete facing was not aestheti-
cally acceptable. The performance of these two walls shown
in Fig. 6 indicates that it is feasible to construct GRS RWs
as permanent important RWs if the facing is more rigid and
aesthetically acceptable.
These experiences with the three Chiba test embankments
led us to the construction of GRS RW with staged–con-
structed FHR facing (Fig. 1a). JR No. 1 test embankment
(Fig. 7a) was constructed using Inagi sand during a period
of 1987–1988 to examine whether this type of GRS RW
can support important railways including high-speed rail-
ways (HSRs, Shinkansen) that require very high stability and
very small deformation. The reinforcement was a geogrid of
polyvinyl alcohol (PVA) having a tensile strength equal to
2.8 tonf/m. Two types of facings were employed to confirm
the importance of facing rigidity: i.e., five wall segments
c, d, e, f and g had FHR facing (Fig. 1a), but not steel-rein-
forced unlike the FHR facing currently used for the proto-
type GRS RWs, while only segment h had discrete-panel
facing fixed to gravel bags wrapped around by reinforcement
layers.
Subsequently, JR No. 2 test embankment of clay backfill
(Fig. 7b) was constructed in the beginning of 1988 to exam-
ine the effects of reinforcement stiffness and backfill soil
type on the stability of GRS RW with FHR facing. Volcanic
ash clay (called Kanto loam) similar to the one used to con-
struct the three Chiba test embankments was used. Three
test sections of the clayey soil embankment were reinforced
differently: i.e., section a–a with a nonwoven geotextile (as
Chiba Nos. 1, 2 and 3 test embankments); section b–b with
a polymeric geogrid (PVA) sandwiched between two thin
gravelly soil drainage layers; and section c–c with a geo-
composite consisting of a woven geotextile sheet for tensile-
reinforcing sandwiched by two layers of nonwoven geotextile
layers for drainage. Despite different reinforcement materials
and backfill soil types, all the segments having FHR facing
(T) test section; (C) control section
(b)
Scale of
deformation:
(c)
Fig. 7  Plan of a JR No. 1 test embankment with sand backfill; and b
JR No. 2 test embankment with clayey soil backfill, both constructed
at Railway Technical Research Institute; and c cross sections of sec-
tion g–h of JR No. 1 two years after wall construction [23, 29, 32]
of the two embankments performed very well for a period
of about 2 years after wall construction, as typically shown
in Fig. 7c. On the other hand, segment h (having discrete-
panel facing) of JR No. 1 exhibited noticeable deformation.
This case showed again the paramount importance of facing
rigidity for a high stability and small residual deformation of
GRS RW, while this factor could be much more important
than the reinforcement stiffness.
Segments d, f and h of JR No. 1 were vertically loaded
to examine the ultimate wall stability (Fig. 8). Segment
h exhibited the largest deformation with buckling of the
discrete-panel facing, as denoted by letter B in Fig. 8a.
This behavior indicates that this type of discrete-panel fac-
ing is not rigid enough to be used for permanent important
walls. Despite the use of the same relatively short geogrid
layers with a length of L = 2 m for a wall height H = 5 m,
segment d (with FHR facing) exhibited much smaller wall
deformation than segment h. With the same FHR facing,
segment f with shorter reinforcement (L = 1.5 m) exhib-
ited less stable behavior than segment d (L = 2.0 m). The
full-height concrete facings for segments d and f were not

13
Innovative Infrastructure Solutions (2019) 4:49
(a)
0 ble reinforcement should be used to sufficiently restrain the
q0
at the top of the facing, d (cm)
Lateral outward displacement
15
Yielding point
30 q forcement and metallic skin facing as flexible facing
45
50 200
(b) Average footing pressure, q (kPa)
Fig. 8  Vertical loading test of JR No. 1 sand embankment [23, 29,
32]: a view of wall segments d, f and h after loading test; and b lateral
displacement at the facing top versus footing load relations of wall
segments d, f and h (q0 = average pressure due to the weight of load-
ing apparatus)
steel-reinforced, unlike the current practice with prototype
GRS RWs. In the facing of these segments, a horizontal
crack took place at the upper construction joint (denoted
by letter CJ in Fig. 8a), which controlled the yield strength
of these segments (Fig. 8b). Obviously, if the facing had
been made stronger by using light steel reinforcement,
these segments would have been more stable.
Table 1  Classification of reinforced soil RW according to combinations of reinforcement type and facing structure type
Reinforcement
Stiff (or inextensible): e.g., metallic strip
Flexible (or extensible): e.g., polymeric and planar
(grid or sheet)
Page 5 of 18 49
In the current practice of GRS RW with FHR facing, a
biaxial geogrid of PVA, as used with JR Nos. 1 and 2 test
embankments, is usually used because PVA exhibits high
resistance against high PH environment by concrete and high
adhesiveness with concrete, while a grid structure exhib-
its good anchorage in the facing concrete and the backfill.
The vertical spacing between the geogrid layers is 30 cm to
ensure good backfill compaction in a lift of 15 cm and strong
integration of the FHR facing to the reinforced backfill.
Reinforcement and facing, stiff versus flexible
There are four combinations of facing (stiff and flexible)
and reinforcement (stiff and flexible) (Table 1). Previously
many engineers and researchers considered that inextensi-
wall deformation, while stiff (or rigid) facing is unnecessary
but flexible (or deformable) facing is sufficient. Polymeric
reinforcement, such as geogrid, was defined as extensible
reinforcement exhibiting large creep deformation.
A combination of metallic strips as inextensible rein-
δ
was employed with Terre Armée RWs at the initial stage
(Fig. 9a). Later, the metallic skin facing was replaced by
concrete panel facing (Fig. 9b). It was explained that this
400 600 replacement is for easier construction and better aesthetics.
The author considered, however, that this replacement was
due also to that the metallic skin facing cannot effectively
restrain the wall deformation. In parallel with the above,
with GRS RWs at the initial stage, flexible wrapped-around
facing (Fig. 9c) was often employed to construct not only
temporary walls but also permanent walls.
It was soon recognized that wrapped-around facing is
not suitable for permanent important GRS RWs because
of a low durability of wall face and large wall deforma-
tion starting from the wall face. In this respect, there was
a mixing-up of different issues: i.e., the stiffness of facing
and the stiffness of reinforcement. Many engineers and
researchers considered that large wall deformation, if it
takes place, is due to the use of extensible reinforcement
(i.e., polymeric reinforcement). It is true that the wall
deformation during construction may become somehow
Facing
Flexible, not developing high earth pressure on Stiff (i.e., rigid), developing high
the back of facing earth pressure on the back of facing
Metallic skin (Fig. 9a) e.g., Discrete concrete panel (Fig. 9b)
Wrapped-around (Fig. 9c) e.g., Modular block (Fig. 9d);
discrete concrete panel; or FHR
(Fig. 9e)
13
49 Page 6 of 18 Innovative Infrastructure Solutions (2019) 4:49

Reinforcement Reinforcement

Metal Discrete
skin Backfill panel Backfill
facing facing

(a) (b)
Reinforcement

Wrapped Backfill Modular

-around block
facing Geosynthetic facing Backfill
reinforcement
Fig. 10  Different behaviors by different connection forces between
the reinforcement and the facing [24])
(c) (d)
Reinforcement Advantages of FHR facing
FHR
facing
Backfill The use of FHR facing is effective for not only a high dura-
bility of wall face but also a high wall stability in many ways,
Gravel bags as discussed below.

(e)
Fig. 9  Different combinations of flexible or stiff facing and flexible or
stiff reinforcement (refer to Table 1)
larger when using polymeric reinforcement than when
using metallic strip reinforcement. However, a limited
amount of wall deformation during construction is usu-
ally not a practical issue. In most cases, excessive post-
construction wall deformation becomes an actual issue and
this is due usually to the use of flexible facing, enhanced
by poor backfill compaction and poor drain. The perfor-
mance of many GRS RWs with modular block facing
(Fig. 9d) and FHR facing (Fig. 9e) indicated that properly
designed and constructed GRS RWs having relatively stiff
facing exhibit excellent post-construction performance
despite the use of the so-called extensible reinforcement.
Moreover, the post-construction time-dependent defor-
mation of GRS structure is due to visco-plastic deforma-
tions of not only geosynthetic reinforcement but also back-
fill. In actuality, the creep deformation of backfill is the
main cause of the creep deformation of GRS RW. Tatsuoka
et al. [30] and Kongkitkul et al. [9] showed a case history
of a high GRS wall in which the tensile forces working
in the geosynthetic reinforcement decreased with time
without showing a possibility of ultimate creep failure.
As a more fundamental point, with ordinary construction
materials including polymeric reinforcement, creep is not
a degrading phenomenon and the ultimate strength does
not decrease by creep deformation at intermediate load-
ing stages unless the mechanical properties deteriorate by
biochemical effects with time (e.g., [2, 8, 30]).
Non‑cantilever structure
Conventional-type RW is basically a cantilever structure sup-
ported at its base that resists the earth pressure activated by
the unreinforced backfill. Therefore, large internal moment
and shear forces are activated inside the facing, while large
overturning moment and lateral thrust forces develop at the
base of the facing, which usually makes necessary the use
of a pile foundation. As the lateral thrust force at the facing
base is basically proportional to the square of wall height (H)
and the overturning moment to H3, the conventional-type
RW become less cost-effective at an increasing rate with
an increase in H. With GRS RWs with FHR facing (Fig. 1),
on the other hand, a thin lightly steel-reinforced FHR fac-
ing without a pile foundation is usually sufficient. This is
because the FHR facing is a continuous beam supported
at many geogrid reinforcement layers with a small span,
typically 30 cm. Therefore, only small forces are activated
inside the FHR facing, which results in a much simpler fac-
ing structure, while insignificant overturning moment and
lateral thrust forces develop at the facing base, which makes
unnecessary the use of a pile foundation in usual cases.
High earth pressure
If the face is flexible or if the reinforcement layers are not
connected to stiff facing, only insignificant or no earth
pressure is activated at the wall face and only insignificant
or no connection forces are activated at the connection
between the facing and the reinforcement (Fig. 10a). A
significant reduction in the earth pressure at the wall face

13
Innovative Infrastructure Solutions (2019) 4:49 Page 7 of 18 49

from the one with conventional-type RWs having unrein-
forced backfill was often considered as one of the primary
advantages of reinforced soil RW. However, this notion
is theoretically wrong and quite misleading in practice.
That is, insignificant or no earth pressure at the wall face
results in insignificant or no lateral confining pressure in
the active zone, which results in very low strength and
stiffness of the backfill, in particular in the zone close
to the wall face, which may result in unacceptable large
wall deformation and a low wall stability. The reinforced
backfill can become stable without sufficient wall face
earth pressure only if the apparent cohesion due to matrix
suction is large with fine-grained backfill under unsatu-
rated conditions and/or if arching effect is large enough
with large-particle backfill reinforced with a small verti-
cal spacing between reinforcement layers. However, these
two factors are too volatile to ensure long-term stable wall
performance.
In this respect, Schlosser [17] reported the distributions
of tensile forces along metallic strip reinforcement in two
Terre Armée walls having standard discrete concrete panel
facing. With both walls, the ratio of the reinforcement ten-
sile force at the back face of facing (i.e., the connection
force) Tw to its maximum value T max is generally large,
not smaller than 50%, and is larger at lower elevations.
Schlosser [17] stated that “for standard reinforced con-
Short reinforcement
Figure 11a shows the active failure mechanism of unrein-
forced backfill retained by a conventional cantilever RW.
With reinforced soil RWs, the active and transitional zones
in the unreinforced backfill are stabilized with reinforce-
ment layers together with facing. When the facing is not
FHR (such as wrapped-around, modular block and discrete
panel), to restrain the shear deformation, overturning and lat-
eral translation of the reinforced zone, it is usual practice to
make the unreinforced transitional zone small enough usu-
ally by using reinforcement layers all truncated to an equal
length that is large enough (typically L ≥ 0.7H). When all
the reinforcement layers are shorter than this length, the
unreinforced transitional zone becomes too large, while the
reinforced zone becomes too narrow. In this case, when the
facing is not FHR (i.e., deformable), the shear deformation,
overturning and lateral translation of the reinforced zone
may become too large and the facing may buckle at low
Active failure plane Plane at repose
Retaining wall 0
(no friction on Transitional
Active
the back) zone
zone
z
Stationary
zone

crete panel facing, and for depths greater than 0.6H, the
Backfill (φ, c=0, γ)
ratio Tw/Tmax can approach one.” These results and notion φ

clearly indicate that discrete concrete panel facing can 45o+φ/2

(a) σh=KAγz
actually confine very well the active zone activating the
FHR facing 0.4
earth pressure that is nearly equal to the active earth pres- 0.36
sure in the unreinforced backfill. Tatsuoka [24] reports
0.8

Ballast track
many other similar cases of full-scale walls and model 6.5
walls showing a similar trend. With GRS RW with FHR
facing (Fig. 1), the gravel-filled bags wrapped around with 4.3
5.2
6.0

Plane at repose
geosynthetic reinforcement layers function as a temporary 3 (All units in m)
facing structure having some stiffness (or rigidity) that
can develop high earth pressure at the wall face during Length= 3
wall construction (Fig. 10b). As they are subsequently 2.5
(b)
integrated to the FHR facing, the reinforcement layers are D E C D E C
eventually firmly connected to the FHR facing and the
earth pressure that has been activated to the gravel bags is
fully transferred to the FHR facing. B
ӨB

As illustrated in Fig. 10b, high facing/reinforcement

45o+φ/2
connection force results in high earth pressure that is simi- ӨA
A
lar to, or could be larger than, the active earth pressure B
A
that would develop in the unreinforced backfill retained by (c) (d)
a conventional cantilever RW. Therefore, high confining
pressure develops in the active zone, which results in high Fig. 11  a Active failure mechanism in unreinforced backfill retained
stiffness and strength of the backfill, therefore much better by a conventional cantilever RW; b a typical GRS RW with FHR fac-
performance than walls having flexible facing (Fig. 10a). ing designed and two-wedge stability analysis following the current
design code for railway structures; c two-wedge analysis under gen-
That is, a substantial reduction in earth pressure is not the eral conditions; and d most likely critical failure planes when the fac-
target of the reinforced soil RW technology. ing is FHR

13
49 Page 8 of 18 Innovative Infrastructure Solutions (2019) 4:49

elevations in particular when loaded on the crest or by seis-

6. Backfilling
mic loading. When the facing is FHR, the shear deforma-
tion, overturning and lateral translation of the reinforced
5. Cantilever
zone become smaller and the facing is not likely to buckle. 3. Excavation
RC RW
However, when subjected to severe seismic loads, the shear
deformation, overturning and lateral translation of the rein-
forced zone may become large. Tatsuoka et al. [38] reported
an actual case where this mode of deformation/displacement
2. Anchor
took place during the 1995 Kobe earthquake and its numeri- 4. Pile
cal analysis. Koseki et al. [10] reported shaking table model 1. Sheet piles
tests simulating this case.
(a)
On the other hand, Fig. 11b shows a typical GRS RW
with FHR facing designed following the current design code 5. Backfilling
for railway structures. By extending several reinforcement
layers to the plane at repose when the backfill is unrein- 4. Modular
forced, the unreinforced transitional zone has disappeared. block
Besides, by taking advantage of a high pullout capacity of facing
planar reinforcement (i.e., geogrid), several reinforcement
layers at low elevations can be made relatively short while 2. Anchor
3. Excavation
maintaining a sufficiently high wall stability. The current
design code specifies that the allowable minimum length
of the basic short reinforcement is whichever the larger 1. Sheet piles
(b)
value of: (1) 35% of wall height; or (2) 1.5 m; or (3) the
length required for a sufficiently high stability against shear
deformation, overturning and lateral translation under high 2. Wall construction w/o FHR facing
seismic loads evaluated by the two-wedge limit equilibrium-
3. FHR
based stability analysis (Fig. 11c). The critical failure mech-
facing
anism that provides the minimum ratio of the total tensile
force of reinforcement available to the tensile force of rein- 1. Excavation
forcement required to maintain the global force equilibrium (c)
is sought by changing the locations of point A along the
back of facing AD and point B at an arbitrary location in Fig. 12  Construction of RW on an existing gentle slope: a conven-
the backfill and changing the angles θA and θB. With GRS tional cantilever RC RW; b reinforced soil RW with relatively long
RWs having FHR facing (Fig. 11d), point A is located at the reinforcement; and c GRS RW having FHR facing with several long
facing base, which substantially increases the wall stability, geogrid layers at high elevations and short basic geogrid layers (note:
the numbers denote construction sequences [26])
in particular against large loads applied near the wall face
on the crest and high seismic loads, when compared with
reinforced soil walls with flexible or deformable facing. The GRS RWs with modular block facing. L/H becomes even
stability analysis for global lateral translation, overturning larger when the reinforcement is metallic strips to ensure
and shear deformation under severe seismic loading condi- a sufficient pullout capacity. In this case, although a pile
tions taking into account strain-softening of the backfill from foundation is usually unnecessary, an anchored sheet pile
the peak shear strength toward the residual shear strength is may become necessary due to large slope excavation. On
explained in [10–12, 28, 34, 38]. the other hand, Fig. 12c shows the construction of a GRS
Figure 12a shows a conventional cantilever RC RW con- RW with FHR facing having relatively short geogrid layers
structed on an existing gentle slope of which the crest is at low elevations (Fig. 11b). In this case, slope excavation
used for railway or road or another facility to create new becomes much smaller; therefore, an anchored sheet pile
space on the crest. An anchored sheet pile is often used to becomes unnecessary in usual cases, which largely reduces
restrain the deformation of the slope caused by relatively the period and cost of slope reconstruction.
large excavation that is required to construct the RW.
Besides, a pile foundation is usually used for the RW. These High structural integrity
arrangements make this slope reconstruction very costly.
Figure 12b shows the construction of a reinforced soil RW With discrete-panel facing, local failure starting from some
having relatively long reinforcement, typically L ≥ 0.7H for place may not be effectively restrained by the whole wall.

13
Innovative Infrastructure Solutions (2019) 4:49 Page 9 of 18 49

Instead, local failure may develop to the overall wall failure. Electric Noise
Lee et al. [14] reported the failure of a set of reinforced soil power barrier Crash barrier
supply
RWs constructed on a natural slope in which metallic strips
Noise
at low elevations were relatively short to avoid large slope barrier
excavation. Some of reinforcement was pulled out, which Ballast track
RC RW
resulted in pushing out of a couple of standard concrete pan- RC RW
Unreinforced
els and this local failure developed to the collapse of the Unreinforced
backfill
backfill
whole wall. A similar failure took place in Japan. Although
the walls did not fully collapsed, standard discrete concrete
panel facings of several metallic strip-reinforced soil RWs
Pile foundation
deformed largely with multiple panels separated from each (a)
other while losing a wall face alignment during the 1995 Electric Noise Crash barrier
Kobe earthquake [25], the 2004 Niigata-ken Chuetsu earth- power barrier
supply
quake [5], the 1999 Kocaeli earthquake in Turkey [4] and the
Noise
2011 Great East Japan earthquake [13]. This type of failure barrier
RC slab
does not take place with GRS RWs having FHR facing due roadbed
to a high structural integrity of facing. FHR
FHR
When flexible or deformable facing is subjected to con- facing Backfill facing Backfill
centrated load immediately back of the facing on the crest
of the backfill, such local failure passing the wall face at an
intermediate elevation as shown in Fig. 13 may take place (b)
Geogrid reinforcement Geogrid reinforcement

[22]. On the other hand, with FHR facing connected to rein- Noise
Crash barrier
forcement layers, the failure plane reaches the facing base barrier

(Fig. 13) and all the reinforcement layers resist the applied
load, which significantly increases the wall stability. With
GRS RW with FHR facing (Fig. 1), one unit of FHR fac-
ing is basically 20 m long separated by vertical construc-
tion joins. The whole of each unit can effectively resist
concentrated loads applied to the facing unit. Fully taking
advantage of these features of FHR facing and the staged
wall construction (Fig. 1), GRS Bridge Abutment and GRS
Integral Bridge with FHR facings, which directly support a
continuous girder, were developed, as explained later.
In congested urban areas, elevated structures for railways
and roads are constructed for grade separation with cross-
ing other surface railways and roads and they are required
to occupy as narrow as possible space. To this end, in many
cases, the RC facing of conventional-type RW supports other
facilities, such as noise barrier, crash barrier, electric power
supply and others (Fig. 14a). GRS RW with FHR facing
(Fig. 1) maintains these important functions of wall facing
Concentrated
load
Local failure
when not using a
FHR facing
Failure plane when reinforcement
is connected to a FHR facing
Fig. 13  Failure of reinforced soil RW when subjected to concentrated
load near the wall face at the crest of a reinforced soil RW using, or
not using, a FHR facing connected to reinforcement layers
Additional fill
Modular
block
facing Backfill
Reinforcement
(c)
Fig. 14  Typical elevated transportation structures: a conventional
piled cantilever RC RW; b GRS RW with FHR facing; and c rein-
forced soil RW with modular block facing
structure in a more cost-effectively way (Fig. 14b). On the
other hand, when the facing is not FHR (such as facing of
modular blocks or discrete panels), some width from the
wall face on the crest is not reliable to arrange a railway track
or a road, while a foundation structure other than the facing
becomes necessary to support other facilities (Fig. 14c).
Besides, recent HSRs (i.e., Shinkansen) in Japan use
continuous RC slab tracks in place of ballasted tracks for a
lower life cycle cost resulting from a large reduction in the
maintenance cost despite a relatively high construction cost.
Continuous RC slab tracks are not allowed to be constructed
on ordinary embankments and the unreinforced backfill
retained by a conventional-type RW, as the residual settle-
ment may exceed a very small allowable limit. On the other
hand, the continuous RC slab track for HSRs is constructed
on the GRS RWs with FHR facing as the standard practice.

13
49 Page 10 of 18
This practice has been adopted based on very small residual
deformation and high wall stability, particularly against high
seismic loads, this type of GRS RWs. This feature can be
attributed to the use of selected backfill soil type and its
good compaction (enhanced by the staged construction and
geogrid reinforcement, Fig. 1a) in addition to the use of FHR
facing.
In summary, the construction cost for GRS RW with FHR
facing (Fig. 1) is basically lower than conventional cantilever
RC RW, in particular when a pile foundation is used with
conventional-type RW. Besides, when based on the life cycle
cost evaluated taking into account the features described
above as well as a high stability against severe earthquakes,
heavy rains and floods, GRS RW with FHR facing is much
more cost-effective than not only conventional-type cantile-
ver RC RW but also other types of reinforced soil RWs not
having staged–constructed FHR facing. For these reasons,
for the last about 30 years, GRS RW with FHR facing has
been adopted nearly fully in place of conventional-type RWs
at many places as summarized in Fig. 2.
Bridge abutments and bridges
Conventional‑type simple girder bridge
With a typical conventional-type bridge, both ends of a sim-
ple-supported girder are placed on a pair of bearings (i.e., a
fixed pin and a movable roller) arranged on the top of a pair
of piled RC walls retaining unreinforced approach fills. As
a result, the following several serious problems are often
encountered (Fig. 15).
1. As the abutment is a cantilever structure supported at its
base, with an increase in the abutment height and with
a decrease in the bearing capacity of subsoil, the RC
walls become more massive. Besides, usually a costly
Settlement by self-weight, traffic load & seismic load
5. Girder
3. Backfill
4. Bearings (fixed or movable)
→ Low seismic stability
2. RC
Displacement by seismic abutment Earth pressure
earth pressure
Ground
Ground settlement & lateral flow
by seismic load
1. Piles
1 two ends of a simple girder were placed on a pair of bear-
Fig. 15  Typical serious problems with conventional-type simple
girder bridge [33, 40]
13
Innovative Infrastructure Solutions (2019) 4:49
pile foundation becomes necessary to keep sufficiently
small the displacement caused by earth pressure from
the approach fill and by ground movements associated
with the construction of approach fills. These factors
become more serious when subjected to external distur-
bances (including seismic loads and scouring).
2. Installation and maintenance of girder bearings and
additional arrangements to prevent the dislodging of
the girder by seismic loads are rather costly.
3. The seismic stability of the girder at a roller bearing and
the unreinforced backfill is rather low.
4. A large bump may develop immediately back of the RC
wall gradually by self-weight of the backfill and long-
term traffic loads and suddenly by displacements of the
abutment and deformation of backfill and subsoil by
seismic loads.
Integral Bridge
To alleviate these problems with the conventional simple
girder bridges, Integral Bridge was developed in the UK
and the North America. Both ends of a continuous girder
are structurally integrated to the top of a pair of RC walls,
followed by the construction of unreinforced approach fills.
A great number of Integral Bridges have been constructed,
typically as bridges overpassing a highway, in the UK and
the North America. However, several problems due to that
the approach fill is unreinforced soil remain unsolved, while
a new problem arises as shown below.
Figure 16 shows the results from laboratory model tests
in which the top of FHR facing was subjected to small cyclic
lateral displacements simulating those by annual thermal
deformation of the girder of Integral Bridge. With cyclic
loading, an active wedge developed in the backfill of air-
dried Toyoura sand, which resulted in noticeable settlements
in the backfill producing a bump immediately back of the
FHR facing. At the same time, the passive earth pressure
at the back of the facing increased significantly with cyclic
loading. These two phenomena, which are independent from
each other, make rather complicated the design and con-
struction of Integral Bridge.
GRS Bridge Abutment
GRS Bridge Abutment was developed to alleviate these
problems with the conventional simple girder bridge illus-
trated in Fig. 15 [26, 31]. Figure 17a shows one of the GRS
Bridge Abutments of the first generation, which was con-
structed in 1993 for a busy urban railway in Tokyo. The
ings (a movable roller and a fixed pin) arranged on a pair
of small foundation placed immediately back of the FHR
facing on the crest of the reinforced backfill. These GRS
Innovative Infrastructure Solutions (2019) 4:49 Page 11 of 18 49

D L Sg Bridge Abutments were constructed directly on a deposit of

intact volcanic ash clay (Kanto loam) without using a pile
foundation.
K 50.5 cm
H=50.5 cm
The GRS Bridge Abutments were designed in such that
the FHR facing effectively resists the lateral seismic loads
0.6 0.6 from the girder. This capacity was confirmed by performing
0.5 0.5 full-scale loading tests applying a lateral outward load up to
98 kN to a small girder foundation of another GRS Bridge

d/H (%)
0.4 0.4
d/H (%)

0.3 0.3
0.2
Constant amplitude, D/H
0.2 Abutment at the site (Fig. 17b). The maximum lateral move-
0.1 0.1 ment near the top of the facing (level U) was only 0.9 mm.
0.0
4
0.0
Besides, the displacement near the facing bottom (i.e., at
L= 35 cm
level L) was approximately one-half of that at level U. These
Total earth pressure

0.0
Initial state
L= 20 cm
3 (K0=0.5)
0.5 trends of behavior show that the whole of the FHR facing
Sg/H (%)
coefficient, K

and all the reinforcement layers, connected to the FHR fac-

2 L= 10 cm
1.0

1
1.5 Distance from
ing, resisted the applied load. A number of similar GRS
0
the back of facing, L= 5 cm
RWs with modular block facing were constructed as abut-
2.0
0 5000 10000 15000 0 5000 10000 15000
ments supporting a simple girder in the USA, and their good
Elapsed time (second) Elapsed time (second)
performance is reported (e.g., [1, 15, 49]). However, with
such GRS Bridge Abutment as depicted in Fig. 17 and those
Fig. 16  Increase in the coefficient K of the total earth pressure on the
back of the facing and the settlement Sg at the crest of dense air-dried
similar to that, as the girder becomes longer and heavier, it
Toyoura sand (Dr = 90%) caused by cyclic lateral displacements at the becomes more difficult to keep small enough the settlement
top of the FHR facing in a small model test [33, 35, 36] of the girder foundations arranged on somehow vertically
compressible reinforced backfill and to ensure a high seis-
mic stability of girder foundations having a small mass. In
particular, seismic displacements at a movable bearing may
increase.
Foundation
In view of the above, GRS Bridge Abutment of the sec-
Track ond generation was developed (Fig. 18a) to alleviate the
drawbacks due to the use of small girder foundations and
a movable bearing with GRS Bridge Abutment of the first
generation described above. That is, one end of a simple
FHR facing

Geogrid girder is placed on a fixed pin bearing arranged on the top

(TTR= 58.8 kN/m)
of the FHR facing of a GRS RW, while the other end of the
girder is placed on a movable bearing arranged on the top
of the adjacent RC pier. To ensure a high seismic stabil-
(a) [All dimensions: m]
ity and for essentially zero bump immediately back of the
FHR facing, the approach block (zone A) consists of well-
compacted lightly cement-mixed well-graded gravelly soil
that is reinforced with geogrid layers connected to the back
of the FHR facing [42]. For a continuous increase in the
thickness of unbound fill behind (zone B) from zero to the
full wall height and for a higher stability of the FHR facing
when subjected to large lateral seismic inertial loads of the
girder at the top, the shape of zone A is trapezoidal with the
base wider than the crest.
The first GRS Bridge Abutment of the second genera-
tion (Fig. 18a) was constructed in 2002 at Takada along
Kagoshima Route of Kyushu Shinkansen (HSR) [31, 37, 40,
(b)
41]. Figure 19 shows the current Shinkansen network. Since
the construction of the first one, in total 39 GRS Bridge
Fig. 17  GRS Bridge Abutments (the first generation) near Sakuradai Abutments including the one shown in Fig. 18b have been
station, Seibu Ikebukuro line, Tokyo; a elevation of the bridge; and
b lateral loading test of another GRS Bridge Abutment performed on constructed until today. For Nishi-Kyushu Route of Kyushu
June 24, 1993 Shinkansen now under construction, in total 80 GRS Bridge

13
49 Page 12 of 18 Innovative Infrastructure Solutions (2019) 4:49

4. Girder 1. GRS RW Abutments were adopted fully in place of conventional-type

bridge abutments (Table 2), many at tunnel entrances.
3. Movable 3. Fixed
bearing bearing
B
GRS Integral Bridge
2. FHR facing A
Pier
Structural integration of the girder to the FHR facing

GRS Bridge Abutment (Figs. 17 and 18) is not free from

(a)
several problems due to the use of bearing, in particular
when both ends of a simple girder are supported by a pair
of GRS Bridge Abutments (Fig. 17). In view of the above,
GRS Integral Bridge (Fig. 3) was then developed to alleviate
these problems while maintaining the advantageous features
of GRS Bridge Abutment. That is, after the deformation of
the backfill and subsoil has taken place sufficiently, FHR fac-
ings are constructed. Then, a continuous girder is arranged
with its both ends structurally integrated to the top of a pair
Oct. 2011 of FHR facings (without using bearings). Due to the specific
staged construction method described in Fig. 1a, the FHR
facings are firmly connected to geogrid layers reinforcing
the backfill. In this way, the connections between the FHR
facing and the geogrid layers and between the girder and the
FHR facings become free from damage due to the deforma-
tion of the backfill and subsoil that may take place associated
with the construction of GRS RW. As a result, GRS Integral
Bridge alleviates all of the serious drawbacks with conven-
tional simple girder bridges illustrated in Fig. 15.
Aug. 2012
(b) Research

Fig. 18  GRS Bridge Abutment (the second generation): a struc-

A series of static and dynamic loading tests were performed
ture (the numbers denote the construction sequences); and b a on small models in the laboratory and on a full-scale model
13.4-m-high GRS Bridge Abutment at Mantaro for Hokkaido Shin- constructed at Railway Technical Research Institute in 2009
kansen (Fig. 20) to establish the design and construction procedures
of GRS Integral Bridge. The full-scale model comprised a
In service
14.75-m-long and 3-m-wide girder that was integrated to
In total 2,764.5 km
Hokkaido SKS*
Shin-Otaru the top of a pair of 5.55-m-high RC walls. The results of
211.5 km (tentative)
(Shin-HakodateHokuto - Sapporo)
Hokuriku SKS Sapporo
these studies showed that GRS Integral Bridge is highly
125.2 km
Under (Kanazawa – Tsuruga)
construction Kyushu SKS, Nish-Kyushu Route Hokkaido SKS
cost-effective while highly stable against long-term thermal
66.0 km
(Takeo Onsen - Nagasaki）
Shin-Hakodate deformation as well as against severe seismic loads, heavy
*SKS: ShinKanSen (High Speed Railway) Shin-Aomori Hokuto
Hachinohe

Aomori
Kyushu SKS, Nishi-Kyushu Route Hakusan
(to be completed in 2022) General Rolling
Niigata Sendai
Hokuriku SKS Kanazawa
Fukushima Table 2  GRS structures for Nishi-Kyushu route, Kyushu Shinkansen
Okayama
Takeo-
Shin-Shimonoseki Toyama
Nagano [20]
Hakata Hiroshima Tsuruga Fukui Omiya
Onsen Takasaki
Nagoya
Isahaya Shin-Tosu Shin-Kobe
Shin-Osaka
Tokyo
Saga pref. Nagasaki pref. Total
Nagasaki Kumamoto
Shin-Yatsushiro
1 GRS RW with FHR facing (m) 10009 3979 4988
Kagoshima-Chuo
2 GRS tunnel entrance protec- 15 40 55*
tion
3 GRS Bridge Abutment 34 46 80
Fig. 19  Network of high-speed railways (HSRs; Shinkansen) in Japan
4 GRS Integral Bridge 2 5 7
[20])
*89% of the total tunnel entrance protections (62)

13
Innovative Infrastructure Solutions (2019) 4:49 Page 13 of 18 49

60 Dislodging of girder

Lateral displacement at top,

Conventional GRS bridge abutment
(gravity) (1st. generation)
40
Lateral (displacement at the

dT (mm)
dT hinged bearing)
displacements
in shaking table tests dB GRS
20
Integral

Integral
0

Conventional (gravity type

Fig. 20  Full-scale model of GRS Integral Bridge constructed at Rail- abutment w/o a pile) 60 Integral
way Technical Research Institute [6, 7, 40]

Lateral displacement at bottom,

GRS
40 Integral
GRS bridge abutment

dB (mm)
(1st generation)

rains and floods. Several key findings from the research are 20 Conventional
(gravity)
presented below. Integral
Figure 21 shows the settlement immediately back of 0
GRS bridge abutment
Most stable
the facing in the backfill of air-dried Toyoura sand caused 0 200 400 600 800 1000 1200
GRS Integral Base acceleration, αmax (gal)
by cyclic lateral displacements at the top of FHR facing
observed in laboratory model tests. The FHR facing was

Settlement of the backfill, S5 (mm)

0
GRS
hinged at the bottom; H is the wall height = 50.5 cm and D GRS bridge
abutment
Integral
20
is the double amplitude displacement. The following three Integral
model walls were tested: NR (the backfill was unreinforced); 40 Out of
S5
R&NoC (the backfill was reinforced with geogrid layers that
measument
Conventional range
(gravity)
were not connected to the FHR facing); and R&C (the back- Residual settlement
at the backfill in
5 cm 60

fill was reinforced with geogrid layers that were connected shaking table tests
0 200 400 600 800 1000 1200

to the FHR facing simulating the abutment of GRS Integral Base acceleration, αmax (gal)

Bridge). Only R&C model exhibited nearly zero settlement

in the backfill by preventing the development of an active
wedge in the backfill.
Figure 22 shows the residual lateral displacements at the
top and bottom of the FHR facing and the residual back-
fill settlement at 5 cm back of the facing in shaking table
NR: not reinforced
-0.5
at 5 cm back of the facing, Sg/H (%)
R&NoC: reinforced,
but no connection
Fig. 22  Residual lateral displacements at the top and bottom of the
FHR facing and residual settlement at 5 cm back of the FHR facing in
shaking table tests of four bridge models (αmax: the amplitude of input
acceleration at a frequency of 5 Hz) [33]
tests of four bridge models, plotted against the amplitude of
input acceleration at the shaking table, αmax. Twenty sinu-
R&C: reinforced; soidal waves at an input frequency fi = 5 Hz were applied at
and connection each stage increasing stepwise αmax. Already at a low αmax
(= 200 gal), the model of conventional simple girder bridge
with gravity-type abutments not supported by a pile started
exhibiting large lateral displacements first at the top and
then at the bottom of the abutment, which resulted in large

R & C: D/H= 0.6 %

Residual settlement of the backfill

0.0
Reinforced & Connected (R & C): settlement in the backfill. The bridge model comprising a
0.5 D/H= 0.2 % pair of GRS Bridge Abutment (1st generation) supporting
1.0 Reinforced & No Connected
the two ends of a simple girder via a pair of fixed pin and
(R & NoC): D/H= 0.2 % movable roller bearings on a pair of small foundation placed
1.5 NR:
D/H= 0.6 % on the crest of the reinforced backfill and the model of Inte-
R & NoC:
2.0 S5 D/H= 0.6 % gral Bridge (with unreinforced backfill) were more stable.
NR (no reinforcement):
2.5
5 cm
D/H= 0.2 % However, they were not as stable as GRS Integral Bridge
model, which lost its stability due to the rupture at the fac-
3.0
0 50 100 150 200 ing/geogrid connection. The model became more stable by
Number of loading cycles, N (cycles) increasing this connection strength [43]. The relevant struc-
ture of this connection, which becomes more important with
Fig. 21  Settlement at 5 cm back from the FHR facing in the backfill an increase in the bridge span, was studied by performing
of dense air-dried Toyoura sand (Dr = 90%) by cyclic lateral displace- full-scale loading tests [7]. In addition, with the GRS Inte-
ment at the top of FHR facing (H = 50.5 cm) hinged at the bottom in
three model walls (see Fig. 16 for the details of the model tests) [33, gral Bridge model, the settlement in the backfill was kept
35] very small even after lateral displacements at the bottom of

13
49 Page 14 of 18 Innovative Infrastructure Solutions (2019) 4:49

the facing started increasing significantly. This is one of the where f0 = fi. The αmax value when GRS Integral Bridge
most advantageous features of GRS Integral Bridge when model reached the resonance state for fi = 5 Hz exceeded
used for railways and roads. 1000 gal, which was substantially higher than the value
A very high seismic stability of GRS Integral Bridge when the conventional-type bridge model reached the reso-
when compared with other bridge types is due to the fol- nant state, equal to about 200 gal. Thirdly, GRS Integral
lowing four specific mechanisms resulting from its high Bridge exhibits a large damping ratio at the time of failure
structural integrity. Firstly, the initial value (i.e., the value that is usually immediately after having reached the reso-
at low seismic loads) of the natural frequency f0 of GRS nant state, which results in a smaller response acceleration
Integral Bridge is basically much higher than the predomi- ratio at the time of failure. This trend of high damping ratio
nant frequencies of typical severe seismic motions, which is due to a high dissipation rate of the dynamic energy of
were simulated by sinusoidal waves with an input frequency the girder and FHR facings toward the backfill and subsoil
fi = 5 Hz in these model tests (Fig. 23). This factor resulted resulting from a strong connection between the FHR facing
in a low initial response acceleration when subjected to low and the reinforced backfill. Lastly, GRS Integral Bridge has
seismic loads. In comparison, the initial f0 value of the con- a large strength against response acceleration. In particular,
ventional simple girder bridge model was much lower, about the dynamic stability of the continuous girder structurally
one-third of the value of GRS Integral Bridge model, which integrated to the FHR facings is substantially higher than
resulted in a much higher initial response acceleration. Sec- a simple girder of conventional-type bridge (Fig. 22). As
ondly, GRS Integral Bridge model exhibited a low decreas- shown in Fig. 23, the trend of f0 − αmax behavior and associ-
ing rate of f0 with an increase in the input acceleration, αmax, ated dynamic performance of a bridge model comprising
due to a small structural damage by dynamic loading. This a pair of GRS Bridge Abutment of the first generation was
behavior resulted in a slow approach to the resonance state, between those of the conventional simple girder bridge
model and GRS Integral Bridge model.
40

Resonance state for

: Resonance an
state Construction
Natural frequency of bridge, f0 (Hz)

35 input frequency fi= 5Hz

(fi= 5Hz)

30
GRS bridge abutment
The construction of the first prototype GRS Integral Bridge
25
(1st generation) was started in 2012 and completed by the end of 2013 as
an over-road bridge at Kikonai for Hokkaido Shinkansen
20
GRS-RW
GRS integral bridge
GRS-IB (see Fig. 2a for the location and Table 3 for the statistics).
15 Yonezawa et al. [48] reported that the construction cost of
CB
10 VII this bridge was estimated to be about one-half of that of the
I
VIII
IX
equivalent conventional-type simple girder bridge.
Conventional
5 II
III IV
X
By the great tsunami of the 2011 Great East Japan
V XI
bridge 0
VI XII
earthquake, more than 340 bridges located in coastal
(gravity) 0 200 400 600 800 1000 1200 1400

αmaxüb] (gal)
Base acceleration amplitude, Amp[ü
areas, mostly simple girder bridges, collapsed. Kawabe
et al. [3] reported the results of model tests performed
to evaluate the stability of GRS Integral Bridge against
Fig. 23  Decrease in the natural frequency f0 of different bridge mod-
els with an increase in the input acceleration at a frequency fi equal to a deep overtopping tsunami flow and compared it with
5 Hz in the shaking table tests described in Fig. 22 [16] the one of the conventional simple girder bridges. It was

Table 3  Major GRS integral bridges (other than those listed in Table 4)
Railway Bridge name Span Girder structure Note

Hokkaido SKS, between Shin- Tyugakkousen overbridge (at 12.00 m RC slab First prototype
Aomori and Shin-Hakodate- Kikonai)
Hokuto stations
Sanriku railway (local and ordi- Matsumaegawa bridge 27.40 m RC slab Continuous
nary) between Shimanokoshi and 13.7 m + 13.7 m girder with
Tanohata stations Koikorobesawa bridge 39.86 m RC slab two spans
19.93 m + 19.93 m
Haipesawa bridge 60.00 m SRC* through girder
32.16 m + 27.84 m

*Steel-framed steel-reinforced concrete

13
Innovative Infrastructure Solutions (2019) 4:49 Page 15 of 18 49

Geogrid-reinforced
Cement-mixed
27. 8 m 32.16 m simple girder railway bridges of Sanriku Railway were
fully washed away by a deep overtopping tsunami flow
A2 P1 A1
gravelly soil 2.1 m To south
2.2 m

(4.4–8.2 m deep) of the 2011 Great East Japan earthquake.

4.7 m
4.7 m
Local road

F: Foundations of
F F
Haipe
stream
F
They were reconstructed to three GRS Integral Bridges
the collapsed
bridge Ground 8.5 m
(Figs. 24 and 25; see Fig. 2a for the locations and Table 3
(a) 4.5 m Bed rock 8.5 m
for the statistics). These bridges were designed to be stable
improvement

against the overtopping tsunami flow by which the previ-

ous bridges were washed away. It was confirmed that it
is not feasible to cost-effectively construct other types of
bridges that can withstand such a deep overtopping tsu-
(b) nami flow.
Table 4 lists the GRS Integral Bridges that have been
Fig. 24  GRS Integral Bridge at Haipe, Sanriku Railway (the continu- constructed, or will be constructed, for a new HSR under
ous girder is vertically supported by a central pier): a structure; and b
April 2014 [39–41] construction, Nishi-Nagasaki Route, Kyushu Shinkansen
(see Figs. 2a and 19 for its location). Figure 26 shows
the one typical of them. In the same way as GRS Bridge
Geogrid-reinforced 19.93 m 19.93 m Abutment, the approach block comprises well-compacted
Cement-mixed gravelly soil A2 P1 A1
1.8 m 1.2 m 1.2 m
To south
lightly cement-mixed well-graded gravelly soil that is
4.7 m
reinforced with geogrid layers firmly connected to the
0.6m7.6m

Koikorobe
Local road stream
F: Foundations
F
Ground FHR facing to ensure a very high seismic stability of the
improvement
of the collapsed
bridge
F
approach block and for essentially zero bump immediately
(a) 6.5 m 5.0 m Bed rock 6.5 m
behind the FHR facing.
The GRS Integral Bridges listed in Tables 3 and 4 are
all for railways (mostly HSRs, Shinkansen), for which the
geogrid-reinforced gravelly soil of the approach blocks is
lightly cement-mixed. In this respect, the geogrid-rein-
forced well-compacted well-graded gravelly soil of the two
(b) approach blocks of the full-scale model (Fig. 20) is either
lightly cement-mixed (the left-side one) or unbound (the
right-side one).
Fig. 25  GRS Integral Bridge at Koikorobe, Sanriku Railway (the con-
tinuous girder is vertically supported by a central pier): a structure; Their long-term behaviors and the results of full-scale
and b April 2014 [39–41] loading tests simulating annual thermal displacements and
strong seismic loads showed that the approach block com-
prising geogrid-reinforced unbound gravelly soil behaves
confirmed that, due to a high structural integrity of the satisfactorily in a similar way as the one comprising lightly
three major components, the girder, the FHR facings and cement-mixed gravelly soil. So, it is very likely that, when
the approach blocks, GRS Integral Bridge is substantially the allowable settlement is not very strict as those sup-
more stable than simple girder bridges. Three conventional porting a continuous RC slab for a HSR, the approach

Table 4  GRS integral bridges for Nishi-Kyushu line, Kyushu Shinkansen [20]

Location Bridge name Span (m) Girder structure

Between Takeo-Onsen and Ureshino-Onsen stations Momoki No. 1 overbridge 12.00 RC slab
Tsubakihara overbridge 10.00
Between Ureshino-Onsen and Shin-Omuru stations Onibashi No. 1 overbridge (for a line to 10.10
Omuru general rolling)
Between Isahaya and Nagasaki stations Genshu overbridge* 30.00 Four PPC
T-shaped main
girders
Genshu bridge 20.00 RC slab
Kaizu bridge 15.00
Funaishi No. 4 overbridge 15.00

*See Fig. 26

13
49 Page 16 of 18 Innovative Infrastructure Solutions (2019) 4:49

and construction concepts of GRS structures. The following

three technical breakthroughs were essential for the above:

1. Use of full-height rigid (FHR) facing: This methodology

changed the earth pressure that is taken into account in
wall design from low values to high values, while this
changed the facing from a secondary, thus deformable,
structural component to a primary, thus stiff, structural
component.
(a) 2. Staged construction: This technology reversed the con-
11.3m struction sequence from the backfill last to the backfill
2.1m
first.
3. Structural integration of the girder to the FHR facings
(b) Four PPC T-shaped girders
for GRS Integral Bridge: This design concept changed
For Takeo-Onsen
the bridge structure from a statically determinate but
For Nagasaki unstable one to a statically indeterminate but stable one.

The use of FHR facing firmly connected to the reinforce-

GRS abutment ment layers ensures not only a high durability of wall face
& approach fill
but also a high wall stability by: developing high earth pres-
sure on the back of the facing, which results in high con-
Highway
fining pressure, thereby high stiffness and strength of the
(c) backfill; and making monolithic the wall behavior by pre-
-mixed soil

venting local failure and its development to global wall fail-

(d)
Fig. 26  GRS Integral Bridge at Genshu, Nishi-Nagasaki route,
Kyushu Shinkansen: a structure; b cross section of the girder; c com-
pleted bridge; d after construction of RC slab track (Feb. 2019) (a–c:
[20])
blocks for GRS Integral Bridge could be constructed using
geogrid-reinforced unbound gravelly soil.
Summary
The development and construction for about 35 years of
various types of geosynthetic-reinforced soil (GRS) struc-
tures is briefly described. The development of these GRS
structures started with retaining walls (RWs) and has been
extended to GRS Integral Bridges. For their development, it
was required to change and reverse our conventional design
ure. Besides, unlike conventional cantilever RWs, the FHR
facing behaves as a continuous beam laterally supported
by many geogrid layers. As a result, the shear forces and
moments activated inside the FHR facing and the overturn-
ing moment and lateral thrust forces activated at the facing
base do not become as large as those in the cantilever fac-
ing, leading to a much lighter facing structure and making
unnecessary the use of a pile foundation under ordinary
conditions.
In the staged construction, a GRS RW without FHR fac-
ing is first constructed with the help of gravel bags (or its
equivalent) placed at the shoulder of each soil layer. After
the deformation of the backfill and subsoil has taken place
sufficiently, FHR facing is constructed by casting-in-place
concrete on the temporary wall face. In this way, relative dis-
placements between the facing and the backfill that damage
the connection between the FHR facing and the reinforce-
ment layers (and between the girder and the FHR facings
with GRS Integral Bridge) do not take place.
FHR facing contributes significantly to the stability of
GRS structures in many ways, while the FHR facing can
be used as a foundation structure supporting other struc-
tures (e.g., noise barrier walls, crash barrier walls, electric
power supply facilities, etc.) at its top. Taking advantages
of the features of the FHR facing described above, GRS
Bridge Abutment (the second generation) supports one end
of a simple girder on a fixed bearing arranged at the top
of the staged–constructed FHR facing of a GRS RW. As
a bridge comprising a GRS Bridge Abutment is statically

13
Innovative Infrastructure Solutions (2019) 4:49
determinate, its design and construction is not complicated
and quite feasible in ordinary practice.
GRS Integral Bridge was developed also taking advan-
tage of the features of staged–constructed FHR facing. With
this bridge system, at the last stage of construction, both
ends of a continuous girder are structurally integrated to
the top ends of a pair of FHR facing of GRS RW. As GRS
Integral Bridge is a statically indeterminate structure, its
design is more complicated than conventional simple girder
bridges, which are basically statically determinate structures.
However, this issue is not serious in today’s computerized
design practice. Rather, we can obtain significant benefits
by structural integration of the three components (i.e., the
girder, the FHR facings and the approach fills) in increasing
the stability and decreasing the residual deformation. All
these benefits result in a much higher life cycle cost-effec-
tiveness. The design and construction codes and manuals
have been published.
In addition to the GRS structures described above, GRS
Box Culvert integrated to GRS RWs on both sides and GRS
Tunnel Exit Protection were also developed. The develop-
ment of all these GRS structures is one of recent significant
technological advances of Geosynthetic Engineering disci-
pline exploring new applications of geosynthetics.
Acknowledgements The author sincerely appreciates the efforts and
help of his previous and current colleagues who were and are being
involved in this long research and construction project at University
of Tokyo, Tokyo University of Science, Railway Technical Research
Institute, Japan, Japan Railway Construction, Transport and Technol-
ogy Agency, a number of railway companies and Integrated Geotech-
nology Institute, Ltd. and other consulting and construction companies,
in particular Dr. Tateyama, M. and Prof. Koseki, J.
References
1. Abu-Hejleh N, Zornberg JG, Wang T, Watcharamonthein J (2002)
Monitored displacements of unique geosynthetic-reinforced soil
bridge abutments. Geosynth Int 9(1):71–95
2. Hirakawa D, Kongkitkul W, Tatsuoka F, Uchimura T (2003)
Time-dependent stress-strain behavior due to viscous property of
geosynthetic reinforcement. Geosynth Int 10(6):176–199
3. Kawabe S, Kikuchi Y, Watanabe K, Tatsuoka F (2015) Model tests
on the stability of GRS integral bridge against tsunami load. In:
Proceedings of 15th Asian regional conference on soil mechanics
and geotechnical engineering, Fukuoka
4. Kempton G, Özçelik H, Naughton P, Mum N, Dundar F (2008)
The long-term performance of polymeric reinforced walls under
static and seismic conditions. In: Proceedings of 4th European
conference on geosynthetics engineering (EuroGeo 4), Edinburgh,
paper no. 181
5. Kitamura Y, Nagao K, Matsuzawa K, Nagakura H (2005) Primary
field investigation on a Terre Armée Wall subjected to strong seis-
mic load—evaluation of the internal condition and damage state.
In: Proceedings of 41st Japan national conference on geotech-
nical engineering, Hakodate, paper no. 981, pp 1959–1960 (in
Japanese)
Page 17 of 18 49
6. Koda M, Nonaka T, Suga M, Kuriyama R, Tateyama M, Tat-
suoka F (2013) Lateral cyclic loading tests of a full-scale GRS
integral bridge model. In: Ling et al. (eds) Proceedings of inter-
national symposium on design and practice of geosynthetic-
reinforced soil structures, Oct. 2013, Bologna, pp 157–174
7. Koda M, Nonaka T, Suga M, Kuriyama R, Tateyama M, Tat-
suoka F (2018) Lateral cyclic loading tests of a full-scale GRS
integral bridge model. In: Proceedings of 11ICG, Seoul (this
conference)
8. Kongkitkul W, Hirakawa D, Tatsuoka F, Uchimura T (2004)
Viscous deformation of geogrid reinforcement under cyclic
loading conditions and its model simulation. Geosynth Int
11(2):73–99
9. Kongkitkul W, Tatsuoka F, Hirakawa D, Sugimoto T, Kawahata
S, Ito M (2010) Time histories of tensile force in geogrid arranged
in two full-scale high walls. Geosynth Int 17(1):12–32
10. Koseki J, Tateyama M, Watanabe K, Nakajima S (2008) Stability
of earth structures against high seismic loads. In: Keynote lecture,
Proceedings of 13th ARC on SMGE, vol 2. Kolkata, pp 222–241
11. Koseki J, Nakajima S, Tateyama M, Watanabe K, Shinoda M
(2009) Seismic performance of geosynthetic-reinforced soil
retaining walls and their performance-based design in Japan,
theme lecture. In: Proceedings of international conference on
performance-based design in earthquake geotechnical engineer-
ing—from case history to practice, Tsukuba, pp 149–161
12. Koseki J (2012) Use of geosynthetics to improve seismic per-
formance of earth structure. Mercer lecture 2011. Geosynth
Geomembr 34:51–68
13. Kuwano J, Koseki J, Miyata Y (2014) Performance of reinforced
soil walls during the 2011 Tohoku earthquake. Geosynth Int
21(3):1–18
14. Lee K, Jones CJFP, Sullivan WR, Trollinger W (1994) Failure and
deformation of four reinforced soil walls in eastern Tennessee.
Géotechnique 44(3):397–426
15. Lee KZZ, Wu JTH (2004) A synthesis of case histories on
GRS bridge-supporting structures with flexible facing. Geotext
Geomembr 22(4):181–204
16. Munoz H, Tatsuoka F, Hirakawa D, Nishikiori H, Soma R, Tatey-
ama M, Watanabe K (2012) Dynamic stability of geosynthetic-
reinforced soil integral bridge. Gesynth Int 19(1):11–38
17. Schlosser F (1990) Mechanically stabilized earth retaining struc-
tures in Europe. In: Lambe PC, Hansen LA (eds) Design and
performance of earth retaining structures. Geotechnical Special
Publication No. 25, ASCE, pp 347–378
18. Shinoda M, Uchimura T, Tatsuoka F (2003) Increasing the stiff-
ness of mechanically reinforced backfill by preloading and pre-
stressing. Soils Found 43(1):75–92
19. Shinoda M, Uchimura T, Tatsuoka F (2003) Improving the
dynamic performance of preloaded and prestressed mechanically
reinforced backfill by using a ratchet connection. Soils Found
43(2):33–54
20. Soga D, Takano Y, Yonezawa T, Koda H, Tateyama M, Tatsuoka
F (2018) Design and construction of various type GRS structures
for a new high-speed railway. In: Proceedings of 11ICG, Seoul
(this conference)
21. Tatsuoka F, Yamauchi H (1986) A reinforcing method for steep
clay slopes using a non-woven geotextile. Geotext Geomembr J
IGS 4(3/4):241–268
22. Tatsuoka F, Tateyama M, Murata O (1989) Earth retaining wall
with a short geotextile and a rigid facing. In: Proceedings of 12th
international conference on SMFE, vol 12, no 2. Rio de Janeiro,
pp 1311–1314
23. Tatsuoka F, Murata O, Tateyama M, Nakamura K, Tamura Y, Ling
H-I, Iwasaki K, Yamauchi H (1990) Reinforcing steep clay slopes
with a non-woven geotextile. In: Proceedings of international rein-
forced soil conference, Glasgow, Sept 1990, pp 141–146
13
49 Page 18 of 18
24. Tatsuoka F (1992) Roles of facing rigidity in soil reinforcing,
keynote lecture. In: Proceedings of earth reinforcement practice
(IS-Kyushu ‘92), vol 1, pp 831–870
25. Tatsuoka F, Koseki J, Tateyama M (1997) Performance of earth
reinforcement structures during the Great Hanshin earthquake.
In: Ochiai et al. (eds) Special lecture, proceedings of interna-
tional symposium. on earth reinforcement, IS Kyushu ‘96, vol 2.
Balkema, pp 973–1008
26. Tatsuoka F, Tateyama M, Uchimura T, Koseki J (1997) Geo-
synthetic-reinforced soil retaining walls as important permanent
structures, 1996–1997 mercer lecture. Geosynth Int 4(2):81–136
27. Tatsuoka F, Uchimura T, Tateyama M (1997) Preloaded and pre-
stressed reinforced soil. Soils Found 37(3):79–94
28. Tatsuoka F, Koseki J, Tateyama M, Munaf Y, Horii N (1998)
Seismic stability against high seismic loads of geosynthetic-rein-
forced soil retaining structures, keynote lecture. In: Proceedings
of 6th international conference on geosynthetics, vol 1. Atlanta,
pp 103–142
29. Tatsuoka F, Tateyama M, Tamura Y, Yamauchi H (2000) Lessons
from the failure of full-scale models and recent geosynthetic-rein-
forced soil retaining walls. In: Proceedings of the second Asian
geosynthetics conference, GeoAsia 2000, vol 1. Kuala Lumpur,
pp 23–53
30. Tatsuoka F, Hirakawa D, Shinoda M, Kongkitkul W, Uchimura
T (2004) An old but new issue; viscous properties of polymer
geosynthetic reinforcement and geosynthetic-reinforced soil struc-
tures, keynote lecture. In: Proceedings of GeoAsia04, Seoul, pp
29–77
31. Tatsuoka F, Tateyama M, Aoki H, Watanabe K (2005) Bridge
abutment made of cement-mixed gravel backfill. In: Indradratna
B, Chu J (eds) Ground improvement, case histories, vol 3. Elsevier
geo-engineering book series. Elsevier, Amsterdam, pp 829–873
32. Tatsuoka F (2008) Geosynthetic-reinforced soil structures: a cost-
effective solution combining two engineering disciplines. In: 19th
Carrillo lecture, Mexican society for soil mechanics, Aguascali-
entes, 29 Nov 2008
33. Tatsuoka F, Hirakawa D, Nojiri M, Aizawa H, Nishikiori H,
Soma R, Tateyama M, Watanabe K (2009) A new type integral
bridge comprising geosynthetic-reinforced soil walls. Gesynth Int
16(4):301–326
34. Tatsuoka F, Koseki J, Tateyama M (2010) Introduction to Japanese
codes for reinforced soil design, panel discussion on reinforced
soil design standards. In: Proceedings of 9th international confer-
ence on geosynthetics, Brazil, pp 245–255
35. Tatsuoka F, Hirakawa D, Nojiri M, Aizawa H, Nishikiori H, Soma
R, Tateyama M, Watanabe K (2010) Closure to discussion on “a
new type of integral bridge comprising geosynthetic-reinforced
soil walls”. Gesynth Int 17(4):1–12
36. Tatsuoka F, Munoz H, Kuroda T, Nishikiori H, Soma R, Kiyota
T, Tateyama M, Watanabe K (2012) Stability of existing bridges
improved by structural integration and nailing. Soils Found
52(3):430–448
13
Innovative Infrastructure Solutions (2019) 4:49
37. Tatsuoka F, Tateyama M, Koseki J, Yonezawa T (2014) Geo-
synthetic-reinforced soil structures for railways in Japan. Transp
Infrastruct Geotechnol 1(1):3–53
38. Tatsuoka F, Koseki J, Kuwano J (2014) Natural disasters miti-
gation by using construction methods with geosynthetics (earth-
quakes), keynote lecture. In: Proceedings of 10th international
conference on geosynthetics, Berlin, September
39. Tatsuoka F, Tateyama M, Koda M Kojima K, Yonezawa T, Shindo
Y, Tamai S (2015) Recent research and practice of GRS inte-
gral bridges for railways in Japan. In: Proceedings of 15th Asian
regional conference on soil mechanics and geotechnical engineer-
ing, Fukuoka
40. Tatsuoka F, Tateyama M, Koda M, Kojima K, Yonezawa T,
Shindo Y, Tamai S (2016) Research and construction of geosyn-
thetic-reinforced soil integral bridges. Transp Geotech 8:4–25
41. Tatsuoka F, Watanabe K (2015) Design, construction and per-
formance of GRS structures for railways in Japan. In: Indraratna
B et al (eds) Ground improvement case histories—compaction,
grouting and geosynthetics. Elsevier, Amsterdam, pp 657–692
42. Tatsuoka F, Furusawa S, Kataoka T, Watanabe K, Lohani TN
(2017) Strength and stiffness of compacted cement-mixed gravelly
soil controlled by the degree of saturation. In: Proceedings of 19th
ICSMGE, Seoul, pp 1253–1256
43. Tatsuoka F, Soma R, Nishikiori H, Watanabe K, Hirakawa D
(2018) High seismic performance of GRS integral bridge with
approach fills of geogrid-reinforced cement-mixed gravelly soil.
In: Proceedings of 11ICG, Seoul (this conference)
44. Tatsuoka F (2018) Geosynthetic-reinforced soil technology in
railway applications—from walls to bridges, prestigious lecture.
In: Proceedings of 11ICG, Seoul
45. Uchimura T, Tateyama M, Tanaka I, Tatsuoka F (2003) Perfor-
mance of a preloaded-prestressed geogrid-reinforced soil pier for
a railway bridge. Soils Found 43(6):33–50
46. Uchimura T, Tamura Y, Tateyama M, Tanaka I, Tatsuoka F (2005)
Vertical and horizontal loading tests on full-scale preloaded
and prestressed geogrid-reinforced soil structures. Soils Found
45(6):75–88
47. Yazaki S, Tatsuoka F, Tateyama M, Koda M, Watanabe K, Dut-
tine A (2013) Seismic design of GRS integral bridge. In: Ling
et al. (eds) Proceedings of international symposium on design
and practice of geosynthetic-reinforced soil structures, Oct. 2013,
Bologna, pp 142–156
48. Yonezawa T, Yamazaki T, Tateyama M, Tatsuoka F (2014) Design
and construction of geosynthetic-reinforced soil structures for
Hokkaido high-speed train line. Transp Geotech 1(1):3–20
49. Zornberg JG, Abu-Hejleh N, Wang T (2001) Geosynthetic-rein-
forced soil bridge abutments—measuring the performance of
geosynthetic reinforcement in a Colorado bridge structure. GFR
magazine (geosynthetics), industrial fabric association interna-
tional, vol 19, no 2, March