grs structures recently developed and constructed for ... · fhr facing for yamanote line over chuo...

Figure 1. A typical cross-section of the GRS RW having a

FHR facing for Yamanote Line over Chuo Line in Tokyo,

constructed during 1995–2000.

1 INTRODUCTION

1.1 GRS retaining walls

For last about 25 years, several innovative rein-

forced soil structures were developed and construct-

ed as retaining walls (RWs), bridge abutments etc.

for transportations (i.e., railways, including high-

speed train lines, and roads) and others. In the late

1990’s, geosynthetic-reinforced soil (GRS) retaining

wall (RW) with staged-constructed full-height rigid

(FHR) facing was developed (Tatsuoka et al., 1997,

2007). Fig. 1 shows a typical case. This GRS RW is

characterized by staged-construction (Fig. 2a): i.e.,

firstly a vertical GRS RW is constructed with a help

of gravel-filled bags placed at the shoulder of each

soil layer. After major deformation of the supporting

ground and backfill has taken place, a full-height

rigid (FHR) facing (i.e., a lightly steel-reinforced

concrete layer) is constructed by casting-in-place

concrete on the wall face wrapped-around with

geosynthetic reinforcement (Fig. 2b). Therefore, suf-

ficiently high connection strength between the fac-

ing and the reinforcement layers can develop, which

results in high earth pressure on the back of the fac-

ing, therefore, a high stability of the backfill.

This new type GRS RW has been constructed at

more than 850 sites (Fig. 3a) and the total wall

length is now about 130 km (Fig. 3b) mainly for

GRS structures recently developed and constructed for railways and

roads in Japan

F. Tatsuoka Tokyo University of Science, Chiba, Japan

M. Tateyama Railway Technical Research Institute, Tokyo, Japan

J. Koseki

Institute of Industrial Science, University of Tokyo, Tokyo, Japan

Page limit 20

ABSTRACT: The construction of permanent geosynthetic-reinforced soil (GRS) retaining walls (RWs) with a

staged-constructed full-height rigid facing for railways, including high-speed train lines, roads and others

started about twenty five years ago in Japan. The total wall length is now more than 130 km, replacing con-

ventional gravity-type or cantilever steel-reinforced concrete (RC) RWs and steel-reinforced soil RWs. Many

were also constructed to reconstruct conventional type RWs and embankments that collapsed during recent

earthquakes, heavy rains, floods and storms. It is proposed to construct GRS coastal dykes with lightly steel-

reinforced concrete facings connected to geosynthetic reinforcement layers as tsunami barriers. By taking ad-

vantage of this GRS RW technology, a number of bridge abutments comprising geosynthetic-reinforced back-

fill were developed. The latest version for new construction, the GRS integral bridge, comprises a continuous

girder integrated to a pair of RC facing, not using bearings, while the backfill is reinforced with geosynthetic

reinforcement layers connected to the facings. Its high cost-effectiveness with a very high seismic stability is

described. The first prototype was constructed 2011 for a new high-speed train line. It is proposed to construct

GRS integral bridges and GR approach fills to restore conventional type bridges that collapsed by tsunami of

the 2011 Great East Japan Earthquake Disaster and also to newly construct many for new high-speed train

lines. Based on the GRS integral bridge technology, a new method to reinforce existing old conventional type

bridges was developed; the backfill is stabilized by large-diameter nails connected to the abutments and then

the girder is integrated to the abutments: i.e., nail-reinforced soil (NRS) integrated bridges.

railways and also for roads and other facilities. Any

problematic case, including poor performance dur-

ing severe rains and earthquakes, has not been re-

ported. This type of GRS RW has totally replaced

conventional RW types (i.e., gravity type and canti-

lever RC RWs) and steel strip-reinforced soil RWs

with discrete panel facing for new construction of

RWs for railways.

a)

b)

Figure 2. Staged construction of GRS RW: a) construction

steps; and b) details of connection between the facing and the

reinforced backfill (Tatsuoka et al., 1997, 2007).

The full acceptance of this GRS RW technology

by railway engineers is due to a very high cost-

effectiveness with a high-seismic stability as validat-

ed by high performance during the 1995 Kobe

Earthquake (Tatsuoka et al., 1998). This feature was

re-confirmed by high performance of a number of

GRS RWs of this type that had been constructed for

railway including a high-speed train line (Tohoku

Shinkan-sen) during the 2011 Great East Japan

Earthquake Disaster. On the other hand, a great

number of conventional type RWs and embankments

on level ground and slopes and in water-collecting

places were fully collapsed during these and other

earthquakes and heavy rains. Many of them were re-

constructed to GRS RWs of this type. The first

coastal GRS RW, which is about 10 m high facing

Pacific Ocean, was constructed for a length of about

1.0 km in 2010, replacing a gravity type RW that ful-

ly collapsed by scouring in the supporting ground

during a storm in the summer of 2007 (Tatsuoka &

Tateyama, 2012). Based on this technology, coastal

dykes and railway and road embankments having

nearly vertical wall faces or sloped slopes consisting

of geosynthetic-reinforced backfill with continuous

facing connected to the reinforcement layers as tsu-

nami barriers have been proposed.

In this paper, the characteristic features of this

GRS RW technology, typical case histories and les-

sons obtained from their behavior, in particular those

during earthquakes and heavy rains, are described.

a)

b)

Figure 3. a) Locations; and b) annual and cumulative lengths of

GRS RWs with FHR facing (as of March 2011).

1.2 GRS integral and NRS integrated bridges

The second category of innovative GRS structure is

a series of new type bridge systems developed based

on the technology of GRS-RW with FHR facing, de-veloped to alleviate several serious problems with

conventional type bridges (explained later). During

the first half of 1990’s, a number of GRS bridge

abutments were constructed with the girder placed

on a pair of sill beams, via a pair of fixed and mova-

ble bearings, placed immediately behind the facing

on the crest of the reinforced backfill. To alleviate

the problems of long-term settlement and unstable

behavior during earthquakes of the sill beams, this

abutment type was modified by placing the girder on

the crest of facings via a pair of bearings (Tatsuoka

et al., 2005). Finally, a more innovative bridge sys-

tem was developed, called the GRS integral bridge,

which comprises an integral bridge (with the girder

integrated to a pair of RC facings without using

bearings) and the backfill reinforced with

geosynthetic reinforcement layers connected to the

5)5) Completion of

wrapped-around wall

4)4) Second layer3)3) Backfilling & compaction

2) Placing geosynthetic &

gravel gabions

Gravel gabionGeosynthetic

1) Leveling pad for facing

Drain hole

6)6) Casting-in-place

RC facing

Lift = 30 cm

5)5) Completion of

wrapped-around wall

4)4) Second layer3)3) Backfilling & compaction


gravel gabions



Drain hole


RC facing

5)5) Completion of

wrapped-around wall

5)5) Completion of

wrapped-around wall

4)4) Second layer4)4) Second layer3)3) Backfilling & compaction3)3) Backfilling & compaction


gravel gabions



gravel gabions



Drain hole


Drain hole


RC facing


RC facing

Lift = 30 cm

30

Annual w

all

length

（km

）

10

0

20

Tota

l （

km

）

1995 20001990 2005

1995 Kobe EarthquakeTotal

Annual

Restart of construction of

new bullet train lines

2004 Niigata-ken-Chuetsu EQ

2010

140

40

120

100

80

60

130 km

2011 Great

East Japan

EQ

facings. The GRS integral bridge is much more sta-

ble against seismic loads and cyclic lateral dis-

placements at the top of the facing caused by sea-

sonal thermal deformation of the girder. From these

features and because of no use of bearings, this

bridge is much more cost-effective than the other

bridge types described above.

Referring to high cost-effectiveness with high per-

formance of GRS integral bridge, a new method to

reinforce existing old conventional type bridges was

developed: the backfill is stabilized by large-

diameter nails connected to the abutments, then by

integrating the girder to the abutments.

2 GRS RETAINING WALLS

2.1 Characteristic features

The three major characteristic features of the GRS

RW system (Figs. 1 & 2) are summarised below:

(1) Staged construction procedure (Fig. 2a): The first

advantage of the staged construction is that the con-

nection between the reinforcement and the facing is

not damaged by differential settlement between the

facing and the backfill during wall construction.

Then, construction on relatively compressible sub-

soil without using heavy piles for the facing and/or

using relatively deformable backfill (e.g., on-site soil

with a high fines content) becomes possible. Sec-

ondly, good compaction immediately at the back of

the wall face becomes possible, allowing sufficient

outward movements at the wall face mobilizing suf-

ficient tensile forces in the reinforcement during

wall construction.

A firm connection between the RC facing and the

geosynthetic-reinforced backfill can be ensured as

follows (Fig. 2b). Firstly, fresh concrete can easily

enter the inside of the gravel-filled soil bags

wrapped-around with geogrid reinforcement through

the aperture of the geogrid. Secondly, extra water

from fresh concrete is absorbed by gravel filling the

soil bags, which reduces negative effects of the

bleeding phenomenon of concrete. The soil bags

function as a temporary but stable facing during

construction, resisting against earth pressure gener-

ated by compaction works and further backfilling at

higher levels. Then, backfill-compaction becomes

more effective. The soil bags also function as a drain

after construction and as a buffer protecting the con-

nection between the FHR facing and the reinforce-

ment against relative displacement if it takes place

after construction. Moreover, to construct a conven-

tional cantilever RC RW, concrete forms on both

sides of the facing and its propping are necessary

and they become more costly at an increasing rate

with an increase in the wall height. With this type of

GRS RW (Fig. 2a), on the other hand, only an exter-

nal concrete form anchored in the backfill is used

without using an external propping is necessary,

while not using an internal concrete form (Fig. 2b).

Figure 4. Effects of firm connection between the reinforcement

and the facing (Tatsuoka, 1992).

(2) Use of a FHR facing constructed by If the wall face is loosely wrapped-around with geosynthetic re-inforcement without using soil bags, or their equiva-lent, at the shoulder of respective soil layers, or if the reinforcement layers are not connected to a rigid facing, no or only very low tensile forces are acti-vated at the connection between the facing and the reinforcement (Fig. 4a). Then, no or only very small earth pressure is mobilized at the wall face, which results in no significant lateral confining pressure in the active zone. This results in low stiffness and low strength of the active zone, which may lead to intol-erably large deformation of the active zone. On the other hand, with this new GRS RW system, the soil bags function as a temporary facing structure and high earth pressure can be activated at the wall face already before placing a FHR facing (Fig. 4b). As the wrapping-around geosynthetic reinforcement at the wall face is buried in the fresh concrete layer (i.e., the facing), eventually the reinforcement layers are firmly connected to the FHR facing. Then, the earth pressure that has been activated on the tempo-rary facing structure comprising soil bags is eventu-ally transferred to the FHR facing. That is, relatively large earth pressure, similar to the active earth pres-sure that develops in the unreinforced backfill, is ac-tivated to the FHR facing, which results in high con-fining pressure in the active zone, thus high stiffness and strength of the active zone. Then, high perfor-mance of the wall can be ensured.

A conventional type RW is a cantilever structure that resists against the active earth pressure from the unreinforced backfill (Fig. 5a). Therefore, large in-ternal moment and shear force is mobilized inside the facing while large overturning moment and lat-eral thrust force develops at the base of the facing. Thus, a pile foundation is usually used. These disad-vantages become more serious at an increasing rate with an increase in the wall height. Relatively large earth pressure, similar to the one activated on the conventional type RW, may be activated on the back

Unstable active zone Very stable active zone

Reinforcement

Connected

No connection strength High connection strength

High tensile forceLow tensile force

High confining

pressureLow confining

pressure

Unstable active zone Very stable active zone

Reinforcement

Connected

No connection strength High connection strength

High tensile forceLow tensile force

High confining

pressureLow confining

pressure

of the FHR facing of the new type GRS RW. De-spite the above, as the FHR facing behaves as a con-tinuous beam supported by reinforcement layers at many elevations with a small span, typically 30 cm, small forces are mobilised inside the facing structure (Fig. 5b). Hence, the facing structure becomes much simpler and lighter than conventional cantilever RWs. Besides, the overturning moment and lateral thrust force activated at the facing base becomes small, which makes unnecessary the use of a pile foundation in usual cases. Consequently, the GRS RW having a stage-constructed FHR facing becomes much more cost-effective (i.e., much lower construc-tion cost, much speedy construction using much lighter construction machines) than the conventional type cantilever RC RW.

a) b)

Figure 5. a) conventional type RW; and b) GRS RW with

FHR facing (Tatsuoka, 1992, Tatsuoka et a., 1997).

(2) The use of relevant reinforcement type: A poly-

mer geogrid is used for cohesionless soil to ensure

good interlocking and a composite of non-woven

and woven geotextiles for high-water content cohe-

sive soils to facilitate both drainage and tensile rein-

forcement. Then, low-quality on-site soil can be

used as the backfill if necessary.

(3) The use of relatively short reinforcement: The

length of geosynthetic reinforcement necessary for a

sufficient stability of GRS RW having staged con-

structed FHR facing is relatively short when com-

pared to metal strip reinforcement used for walls

having discrete panel facing. This is because: 1) the

anchorage length of planar geosynthetic reinforce-

ment to resist against the tensile load similar to the

tensile rupture strength of reinforcement is much

shorter; 2) a FHR facing prevents the occurrence of

local failure in the reinforced backfill zone by not al-

lowing the development of failure planes passing

through the wall face at an intermediate height; and

3) local failure in the facing, which may take place

with discrete panel or block facing and may result in

the failure of the whole wall, is difficult to take place

with FHR facing. Factors 2) and 3) become more

important when concentrated load is applied on the

top of, or immediately behind, the facing.

Figure 6. A GRS RW having FHR facing at Tanata, Kobe city;

a) immediately after construction; and b) one week after the

1995 Kobe Earthquake (Tatsuoka et al., 1997, 1998).

Figure 7. Railway embankment that collapsed during the

2004 Niigata-ken Chuetsu Earthquake and its reconstruction to

a GRW RW having FHR facing (Morishima et al., 2005).

2.2 Collapse of RWs and embankments by earthquakes and their reconstruction

Numerous embankments and conventional type RWs collapsed by earthquakes in the past. On the other hand, high performance of a GRS RW having stage-constructed FHR facing during the 1995 Kobe Earthquake (Fig. 6) validated its high-seismic stabil-ity. Many gently sloped embankments and conven-tional type RWs that collapsed by that and subse-quent earthquakes were reconstructed to GRS RWs of this type (Tatsuoka et al., 1977, 1998; 2007; Koseki et al., 2006, 2008; Koseki, 2012). Typically, three railway embankments supported by gravity type RWs on slope collapsed during the 2004 Nii-gata-ken Chuetsu Earthquake and they were recon-structed to GRS RWs of this type (Fig. 7). The adop-tion of this technology was due to not only much lower construction cost and much higher stability (in

Earth

pressure

Stress concentration

Earth

pressure

Reinforcement

8 July 1992a)

24 Jan. 1995b) 24 Jan. 1995b)

Silt rock

Sand rock

1:2.

0

13

.18

m

1:4 (V:H)

After remedy

work

Railway

(Jo-etsu line)

Shinano riiver

Gravel-filled steel wire

mesh basket

Rock bolt

After remedy work:

GRS-RW with a FHR facing; slope: 1:0.3 (V:H);

height= 13.2 m, vertical spacing of geogrid= 30 cm

Before failure: sand backfill including

round-shaped gravel on sedimentary soft

rock (weathered, more at shallow places)

After

failureFailed

gravity RW

Silt rock

Sand rock

1:2.

0

13

.18

m

1:4 (V:H)

After remedy

work

Railway

(Jo-etsu line)

Shinano riiver

Gravel-filled steel wire

mesh basket

Rock bolt

After remedy work:

GRS-RW with a FHR facing; slope: 1:0.3 (V:H);

height= 13.2 m, vertical spacing of geogrid= 30 cm

Before failure: sand backfill including

round-shaped gravel on sedimentary soft

rock (weathered, more at shallow places)

After

failureFailed

gravity RW

particular for these soil structures on steep slopes) but also much faster construction resulting from a significant reduction of earthwork when compared to the original gently sloped embankment with a gravity type RW.

Based on vast and very serious damage to railway structures during the 1995 Kobe Earthquake, the seismic design codes for railway structures were substantially revised (Tatsuoka et al., 2010a). The revised codes for soil structures have following sev-eral new concepts and procedures: 1) Introduction of very high design seismic loads

(i.e., level 2) and three ranks of required seismic performance.

2) Recommendation for the use of geosynthetic-reinforced soil structures.

3) Evaluation of seismic performance based on re-sidual displacement.

4) Use of peak and residual shear strengths with well compacted backfill, while ignoring apparent cohesion due to suction assuming that it fully disappears during heavy rains.

5) Design based on the limit equilibrium stability analysis.

6) An emphasis of good backfill compaction and good drainage.

7) No creep reduction factor applied to obtain the rupture strength of geosynthetic reinforcement in seismic design.

Figure 8. Locations and numbers of GRS RWs having FHR

facing for railways constructed before the 2011 Great East Ja-

pan Earthquake Disaster (Tateyama, 2012).

The 2011 Great East Japan Earthquake Disaster,

which took place 11th

March, is the most disastrous earthquake in Japan after the World War II. The damage was a combination of those from earthquake motions and the accompanying great tsunami. A great number of old embankments and RWs that

were not designed and constructed following the current seismic design standard collapsed. In com-parison, a number of GRS RWs of this type that had been designed based on the revised seismic design code described above and constructed in the affected areas of this earthquake performed very well (Fig. 8),

a)

b)

c)

Figure 9. a) Collapse of a wing RW (with a masonry facing

of a bridge abutment at site No. 11 in Fig. 8 (Nagamachi, Sen-

dai for Tohoku Freight line); and b) & c) its reconstruction to a

GRS RW (by the courtesy of the East Japan Railway Co.).

Several conventional type RWs and embankments that collapsed were reconstructed to GRS RWs of this type (Fig. 9). A very fast construction was one of the important advantages of this technology also in this case. In particular, the railway was re-opened at a restricted speed before constructing a FHR fac-

Morioka

Hananomaki

Ichino-

seki

Sendai

Epicenter

Fukushima

Tohoku Shinkansen

(high speed train)

Tsukuba Express (rapid

commuters’ train)

Tohoku Shinkansen

Tohoku Line

Tohoku Line

Tazawako Line

Senseki Line

Yard for Tsukuba Express

Number of GRS walls

Aomori: 58

Iwate: 23

Miyagi: 10

Akita: 1

Yamagata: 2

Fukushima: 1

Ibaraki 1

(total) (96)

Locations of GRS RWs

investigated in details

Tokyo

←Nagamatchi Higashi Sendai→

Irrigation channel

ing (Fig. 9b). Fig. 10 shows one of the three em-bankments that collapsed during an earthquake in-duced one day after the Great East Japan Earthquake Disaster and reconstructed to GRS RWs of this type.

Figure 10. One of the three embankments between Yokokura

and Morinomiya stations, Iiyama Line, that collapsed during

the Nagano-Niigata Border Earthquake and reconstructed to

GRS RWs (by the courtesy of the East Japan Railway Co.)

2.3 Collapse of RWs and embankments by heavy rains, floods and storms and reconstruction

Numerous embankments and conventional type RWs collapsed also by heavy rains, floods and storms in the past. GRS RWs (Fig. 2) were also con-structed to replace a number of embankments that were eroded and washed away by over-flowing flood water (Tatsuoka et al., 1997, 2007). Numerous embankments for roads and railways retained by gravity-type RWs along rivers and seashores col-lapsed by floods and storms, usually triggered by over-turning failure of the RWs caused by scouring in the supporting ground (Fig. 11a). The backfill is largely eroded resulting in the stop of the function of a railway or a road. This type of collapse is easy to take place, because the conventional type RW is a cantilever structure of which the stability is fully controlled by the bearing capacity at the bottom of the RW.

On the other hand, GRS-RWs with a FHR facing is not such a cantilever structure as above, therefore, much more stable against scouring in the supporting ground (Fig. 11b). It is particularly important that the major part of the backfill can survive even if the supporting ground is scoured. A high embankment retained by a masonry gravity-type RW at the lower part on the left bank of Agano river, Niigata Prefec-ture, for West Ban-Etsu line (a railway of East Japan Railway) was collapsed by a flood 30th July 2011 by the mechanism illustrated in Fig. 11a. It was recon-structed to an about 9.4 m-high and 50 m-long GRS RW with a FHR facing (Fig. 12).

a)

b)

Figure 11. a) Collapse of conventional type RW by scouring in

the subsoil (the numbers show the sequence of events); and b)

improved performance of GRS-RW with FHR facing.

Figure 12. Reconstruction of RW on the left bank of Agano

river, Niigata Prefecture, for West Ban-Etsu line to a GRS-RW

(by the courtesy of the East Japan Railway Co.).

Fig. 13 shows another case of this type of failure of a gravity-type RW for a length of about 1.5 km along seashore facing the Pacific Ocean, triggered by scouring in the supporting ground by strong ocean waves during a typhoon No. 9, 29

th Aug.

2007. The wall was reconstructed to a GRS RW with FHR facing (Figs. 13b & c). The FHR facing has a strong resistance against sea wave actions dur-ing storms, while the wall could be stable even if the supporting ground is scoured to some extent. On the other hand, discrete panel facing is not relevant in such a case as this, because the loss of the stability of a single or several panel(s) by scouring in the subsoil and/or erosion of the backfill from joints be-tween adjacent panels may easily lead to the failure of the whole wall.

≒6.0

5,0

2.0

1.0

After failure (before

reconstruction)

Mattress basket

Gravel

Stabilized soil

Welded

steel fabric

Slope before failure

Railway track

(Iiyama line) [All unit in m]

≒14.0

1. Scouring

3. Collapse of

embankment

2. Over-turning of RW

River bed

Flood

Flood

Scouring

GRS-RWs with a FHR facing has a

high resistance against scouring

(All units in m)

8.5722.808

0.3

Crusher run

(C-40)

Basal geogrid

(Td= 60 kN/m)

Short geogrid

(Td= 30 kN/m)

3.49

Ground improvement by

cement-mixing 5.0

3.0

2.0

Level pad

(concrete in

steel-mesh

cage

9.3

6

Design high water level

(38.14 m)

0.7

2.0

Ordinary low

water level

(26.08)

Long geogrid

(Td= 30 kN/m)

Height from the

base (m) =

13.1

12.9

11.5

8.9

4.51.0

a)

b)

c)

Figure 13. Seawall for Seisho by-pass of National Road No.

1 in Kanagawa Prefecture, southwest of Tokyo: a) collapse for

a length of about 1.5 km by Typhoon No. 9, 29th Aug. 2007; b)

a typical cross-section of GRS RW; and c) GRS RW under

construction (a & b: by the courtesy of the Ministry of Land,

Infrastructure, Transport and Tourism).

3 NEW BRIDGE TYPES

3.1 Problems with conventional type bridges

A conventional type bridge usually comprises a single simple-supported girder supported by a pair of abutments via fixed (or hinged) and moveable bearings, or multiple simple-supported girders supported by a pair of abutments and a single or multiple pier(s) via multiple sets of bearings. The approach fill is usually unrein-forced backfill. The abutment may be a gravity structure (unreinforced concrete or masonry) or a RC structure. The conventional type bridge has a number of drawbacks as described below (Fig. 14a).

a)

b) Figure 14. a) Several major problems with conventional type

bridges; and b) development of new bridge types for new con-

struction and reinforcement (Tatsuoka et al., 2008a, b; 2009).

Firstly, as the abutment is a cantilever structure

retaining unreinforced backfill, the earth pressure

induces large internal forces as well as large

thrust forces and overturning moment at its bot-

tom. This problem becomes more serious when

designed against severe earthquake loads. There-

fore, usually the abutment becomes massive and a

pile foundation becomes necessary more at an in-

creasing rate with an increase in the abutment

height. Secondly, although only very small

movement is allowed with the abutments, the

backfill are constructed after the abutments have

been completed. Hence, when constructed on

thick soft ground, many long piles may become

necessary to prevent any small movement of the

abutments caused by the earth pressure as well as

settlement and lateral flow in the subsoil. Large

negative friction may develop along the piles.

Thirdly, the construction and long-term mainte-

nance of the bearings and the connections be-

tween simple-supported girders are generally

costly. Fourthly, the bearings and the backfill are

the weakest elements against seismic loads. The

girder may dislodge at a moveable bearing by se-

vere seismic loads. A significant bump may be

formed behind the abutment by long-term settle-

Soil bags

Full-height

rigid facing

Geogrid Concrete cover

Scour protection

Selected

backfill (RC40)

Crushed gravel

(RC40)

Tie rod

Head

Steel

sheet pile

10 March 2010

Pacific

Ocean

Casting-in-place of

concrete for FHR

facing

(7) Long-term service: 7a. settlement by self-weight & traffic load; &

7b. large deformation by seismic load

(1) Piles

Supporting

ground

(3) Backfill

(4a) Earth pressure

(static & dynamic)

(4a) Displacement due to

earth pressure

(4b) Ground settlement and

lateral flow due to the weight

of backfill, and associated

negative effects on the piles

(i.e., negative friction &

bending).

(2)

abutment

(6) Girder (deck)

(5) Shoes (or bearing)

(fixed or movable);

5a. Long-term maintenance

5b. Low seismic stability

Numbers (1) – (7):

event sequence.

(7) Long-term service: 7a. settlement by self-weight & traffic load; &

7b. large deformation by seismic load

(1) Piles

Supporting

ground

(3) Backfill

(4a) Earth pressure

(static & dynamic)

(4a) Displacement due to

earth pressure

(4b) Ground settlement and

lateral flow due to the weight

of backfill, and associated

negative effects on the piles

(i.e., negative friction &

bending).

(2)

abutment

(6) Girder (deck)

(5) Shoes (or bearing)

(fixed or movable);

5a. Long-term maintenance

5b. Low seismic stability

Numbers (1) – (7):

event sequence.

Conventional

GRS RW Integral

GRS integral NRS integrated

Solving problems with backfill

Solving problems with girder and facing

Rehabilitation of existing conventional type bridges

Structural integration

Nailing

Combined

ment of the backfill due to its self weight, traffic

loads and seismic loads.

3.2 Integral bridge and GRS RW bridge

To alleviate the problems with the conventional type bridge (Fig. 14a), several new bridge types have been proposed (Fig. 14b), which are either for new construction or for reinforcement of existing old bridges. For new construction, the integral bridge comprises the girder structurally integrated to the facings without using bearings. This bridge type was developed to alleviate problems with the structural part (i.e., the girder and facings) of the conventional type bridge. This bridge type is now widely used for roads in the UK, the USA and Canada due mainly to a lower cost for construction and maintenance result-ing from no use of bearings and the use of a contin-uous girder.

However, as the backfill is not reinforced, thus not integrated to the abutments (or facings), the backfill and the structural part do not help each oth-er. Hence, this bridge type cannot alleviate some old problems with unreinforced backfill (described above). Moreover, as the girder is integrated to the abutments, seasonal thermal expansion and contrac-tion of the girder results in cyclic lateral displace-ments at the top of the abutments, which may result in: i) development of high passive earth pressure on the back of the facing; and ii) large settlements due to active failure in the backfill (England et al., 2000). The test results showing the above are pre-sented later.

A number of bridges comprising a pair of GRS RWs with FHR facing that supports a simple-supported girder via bearings placed on sill beams on the crest of the backfill immediately behind the facing were constructed (i.e., GRS RW bridges; Tatsuoka et al., 1997, 2005). Although this bridge type is more cost-effective than the conventional type, the length of the girder is restricted due to low stiffness of the backfill supporting the sill beams and a low seismic stability of the sill beams because of their small mass. Then, this type was modified by placing a girder on the top of the FHR facings via bearings (Fig. 13a; Tatsuoka et al., 2005, 2009). The first prototype was completed in 2003 for a new bul-let train line in Kyushu (Fig. 15b). The abutment was constructed by the staged construction proce-dure, following the GRS RW technology (Fig. 2). That is, the geosynthetic-reinforced backfill is first constructed, followed by the construction of the fac-ing by casting-in-place fresh concrete. Then a girder was placed on the top of a thin RC facing via bear-ings. In this case, a trapezoidal-shaped zone of back-fill back of the facing was a well-compacted cement-mixed well-graded gravelly soil to minimize the long-term residual deformation and to ensure a very high seismic stability. The conventional type RC

abutment laterally supports the unreinforced backfill and the backfill activates static and dynamic earth pressures on the abutment. In contrast, with this new type abutment, the reinforced backfill laterally sup-ports a thin RC facing, therefore, the backfill does not activate large earth pressure on the facing. After this project, nearly sixty similar bridge abutments were designed or constructed until today. Despite the above, this type of abutment is not free from several problems due to the use of bearings.

a)

b)

Figure 15. a) GRS RW bridge with the girder placed on the

top of the facing via bearings (n.b., the numbers denote the

construction sequence); and b) first prototype at Takada, Kyu-

shu, for a new bullet train line (Tatsuoka et al., 2005, 2009).

3.3 GRS integral bridge

To alleviate all these serious drawbacks with con-

ventional type bridge and also those with integral

bridges and GRS RW bridges, summarized above,

the authors proposed a new bridge type for new con-

struction, called the Geosynthetic-Reinforced Soil

(GRS) integral bridge (Fig. 14a; Tatsuoka et al.,

2008a, b, 2009, 2012a). This bridge type comprises

a girder integrated to a pair of FHR facings (without

using bearings) and backfill reinforced with

geosynthetic layers connected to the facings. This

type of bridge is staged-constructed as follows, re-

ferring to Fig. 16a:

0) When the supporting ground is soft and weak,

the subsoil below the abutments may be im-

proved by, for example, cement-mixing in-place.

1) A pair of GRS walls with the wall face wrapped-

around with geogrid reinforcement is constructed.

2) After major deformation of the supporting

ground and backfill has taken place, thin RC

abutments (i.e., FHR facings) are constructed by

4. Girder

1 1. GRS RW2. FHR facing

3. Movable & fixed

shoes

0. Ground improvement

(when necessary)

casting-in-place fresh concrete on the wall face

wrapped-around with geogrid reinforcement, in

the same way as GRS retaining walls with FHR

facing (Fig. 2a: Tatsuoka et al., 1997).

3) A continuous girder is constructed structurally

integrated to the top of the facings.

Figure 16 GRS integral bridge: a) construction sequences

denoted by the numbers; and b) two-span bridge.

With conventional type bridges and GRS RW

bridges, the length of a single girder is restricted to avoid excessive lateral seismic loads to be activated on the abutment on which a fixed bearing supports the girder. With integral bridges, the girder length is restricted additionally to avoid excessive large cyclic lateral displacements at the top of the facings by sea-sonal thermal deformation of the girder. The girder length of conventional type integral bridge is pres-ently specified to be 50 - 60 m in the USA to restrict the maximum thermal deformation of the girder to about 10 cm. With the GRS integral bridge, such re-strictions as above are looser and its girder length limit would be larger than the value for the conven-tional type integral bridge. Fig. 16b illustrates a GRS integral bridge having two spans.

3.4 NRS integrated bridge

There exist a great number of old conventional type bridges with the girder placed on the abutments via a pair of bearings, that are judged not to satisfy the seismic stability requirement according to the cur-rent design standard. Due to deterioration with time, the number of such bridges as above is increasing every year. It is extremely time-consuming and cost-ly to replace such a bridge with a new one without closing a road or a railway, because the following complicated procedure is necessary: 1) a temporary bridge is constructed adjacent to the existing bridge together with constructing a new approach railway or road; 2) the existing bridge is removed; and 3) a new bridge is constructed at the place of the old bridge; and 4) the temporary bridge is removed.

Figure 17. A new method to reinforce old conventional type

bridges (n. b., the numbers denote the construction sequence).

The authors proposed to reinforce such existing

old conventional type bridges as described above by taking advantage of structural characteristics of GRS integral bridges (Shiranita et al., 2010; Tatsuoka et al., 2012b). This reinforcement method is illustrated in Fig. 17: 1. The girder is initially supported with a steel el-

bow member. 2. The backfill is reinforced with a large-diameter

nail (typically 40 cm in diameter), which com-prises a tie rod (steel or FRP) surrounded with the backfill mixed-in-place with cement-slurry provided from a central hollow rod containing the tie rod (Tateyama et al., 1996). The nails are connected to the top and bottom of the abutment.

3. The girder is integrated to the abutments by plac-ing casting-in-place concrete.

The bridge reinforced by this technology is called the nail-reinforced soil (NRS) integrated bridge. This technology has two major advantages. Firstly, the existing bridge is reinforced while in service, thus this method is much faster and much cheaper than the replacement with a new bridge. Secondly, the NRS integrated bridge is much more stable than the conventional type bridge because of no use of bearings and reinforcement of the backfill.

To validate the advantages of GRS integral bridge and NRS integrated bridge, a series of static cyclic loading tests and shaking table tests were performed on small models in 1g, as described below. In 2009, full-scale models of GRS integral bridge and NRS integrated bridge were constructed. A high construc-tability of these bridge types was confirmed. In the beginning of 2011, full-scale loading tests of these two bridge models were performed.

4. STATIC CYCLIC LOADING TESTS


Fig. 18 shows the setup of small model tests in the laboratory performed to simulate lateral cyclic dis-

3. GirderStructurally integrated

Firmly connected

1. GRS wall

2. FHR facing

a) 0. Ground improvement

(when necessary)

Gravel

bags

b)

ソイルセメント

セメントミルク

引張芯材

ソイルセメント

セメントミルク

引張芯材Steel or FRP rod

Soil mixed-in-place

with cement

Cement slurry

2. Reinforcing of the backfill

with a large diameter nails

3. Integration of the girder to

the abutment by casting-

in-place concrete

1. Initial support

of the girder

placements at the top of the abutment (or the facing) by seasonal thermal deformation of the girder.

Figure 18. Model tests to evaluate negative effects of lateral

cyclic displacements at the top of the abutment (i.e., facing):

this figure is when the backfill is reinforced (Tatsuoka et al.,

2009).

Figure 19. a) Loading method; and measured time histories

of b) horizontal displacement at the facing top, c) backfill set-

tlement; and d) total earth pressure, unreinforced dense

Toyoura sand (Dr= 90 %) (Tatsuoka et al., 2009). .

Fig. 19 shows the result of a typical test when the backfill is not reinforced and the facing is hinged at the bottom. The top of the facing is displaced cycli-cally on the active side. Despite a fixed relatively small amplitude of displacement, with cyclic load-ing, the earth pressure that increases when the facing is displaced in the passive direction continues in-creasing and the settlement in the backfill that takes place when the facing is displaced in the active di-rection continues increasing, both without showing a sign of stopping. Eventually, active failure takes place in the backfill, while the earth pressure be-comes close to the large passive pressure.

Tatsuoka et al. (2009, 2010b) showed that these trends of behavior are due to the dual ratchet mecha-nism in the backfill (Fig. 20a). That is, only the ac-tive mode of deformation develops when the facing moves in the active direction and it accumulates with cyclic loading, while only the passive mode of deformation develops when the facing moves in the

passive direction and it accumulates with cyclic loading. Fig. 20b illustrates the stress-strain behavior on the active failure plane and passive failure plane during cyclic displacements of the facing. Due to the dual ratchet mechanism, although the facing displac-es with the same amplitude, the shear strain consist-ently increases on both AFP and PFP, which eventu-ally result into active failure and high passive earth pressure.

a)

b)

Figure 20. a) Backfill behind an abutment with vertical

smooth face and a horizontal backfill crest; and b) stress-strain

behaviors on AFP and PFP (Tatsuoka et al., 2010b).

Figure 21. Effects of reinforcing of the backfill with

geosynthetic reinforcement connected to the facing on the set-

tlement in the backfill by lateral cyclic loading of the facing

(Tatsuoka et al., 2009).

Displacement-controlled

loading system

Lateral

displacement

Load cell

Hinge support

129.5

8.5

　5 10 20 35

50

.5

Settlement of the crest

Box width: 40

Air-dried Toyoura sand

(Dr = about 90%)

9 separated load cells: LCs

[unit : cm]

8 polyester

reinforcement

layers

Distance (cm) from

the back of the facing


loading system

Lateral

displacement

Load cell

Hinge support

129.5

8.5

　5 10 20 35

50

.5


Box width: 40


(Dr = about 90%)


[unit : cm]


loading system

Lateral

displacement

Load cell

Hinge support

129.5

8.5

　5 10 20 35

50

.5


Box width: 40


(Dr = about 90%)


[unit : cm]

8 polyester

reinforcement

layers

Distance (cm) from

the back of the facing

11.5 cm

L

50.5 cm

L

50.5 cm

0.0

0.1

0.2

0.3

0.4

0.5

0.6

La

tera

l d

isp

lacem

en

t o

f

facin

g, d

/H (

%)

0 5000 10000 150002.0

1.5

1.0

0.5

0.0L= 35 cm

L= 20 cm

L= 10 cm

Se

ttle

me

nt

at th

e c

rest

of

ba

ckfill,

Sg/H

(%

)

Elapsed time, t (sec)

Distance from

the back of facing, L= 5 cm

0 5000 10000 15000

0

1

2

3

4

Initial state

(K0=0.5)

To

tal e

art

h p

ressu

re c

oe

ffic

ien

t, K

Elasped time (second)Elapsed time, t (sec) Elapsed time, t (sec)

Se

ttle

me

nt a

t th

e b

ackfill c

rest,

Sg/H

(%

)

La

tera

l d

isp

lace

me

nt

at th

e fa

cin

g to

p,

d/H

(%

)

Ea

rth

pre

ssu

re c

oe

ffic

ien

t, K

15000

Sg

H= 50.5 cm

a) b)

c) d)

H

Abutment

Lateral displacement, δ

45 / 2o

P

45 / 2o

A

Shear

deformation, us

/ tan mob

Stress ratio,

Peak strength： ( / ) tanpeak

Active direction

3A

1A

2A

0A

5A

4A

7A

6A

3P1P

2P

0P

4P

5P

6P

Passive direction

/ 0 (when K= 1)

Peak strength：

( / ) tanpeak

Relation

on AFP

Relation

on PFP

Passive direction

Active direction

AFP

PFP

R&C: reinforced;

and connection

0 50 100 150 2003.0

2.5

2.0

1.5

1.0

0.5

0.0

-0.5

S5

5 cm

S5

5 cm

Reinforced & No Connected

(R & NoC): D/H= 0.2 %

NR (no reinforcement):

D/H= 0.2 %

Reinforced & Connected (R & C):

D/H= 0.2 %

NR:

D/H= 0.6 %R & NoC:

D/H= 0.6 %

R & C: D/H= 0.6 %

Resid

ual sett

lem

ent

of

the b

ackfill

at

5 c

m

back o

f th

e f

acin

g,

Sg/H

(%

)

Number of loading cycles, N (cycles)

NR: not reinforced R&NoC: reinforced,

but no connection

The result when the backfill is unreinforced

and the facing displacement D/H is 0.6 % pre-

sented in Fig. 21, together with a similar test re-

sult when D/H= 0.2 %, are summarized in Fig. 21.

The results when the backfill is reinforced with

and without connected to the facing are also pre-

sented in Fig. 21. It may be seen that the active

failure in the backfill can be effectively prevented

by reinforcing the backfill with the geogrid layers

connected to the facing. When the geogrid layers

are not connected to the facing, the settlement is

still very large. The passive earth pressure in-

creases even when the backfill is reinforced with

geogrid connected to the facing. Even so, the sta-

bility of the facing is kept, as the facing is sup-

ported by a number of geogrid layers.


A series of similar tests as those presented above for the IGS integral bridge were performed on small models of the abutment of NRS integrated bridge (Fig. 22; Tatsuoka et al,, 2012b). To highlight the ef-fects of nailing, the abutment model was embedded only to a depth of 3 m in the supporting ground, simulating prototype abutments not supported with any pile foundation.

Figure 22. Static lateral cyclic loading tests on an abutment

model with nails in the backfill and supporting ground

(Tatsuoka et al., 2012b).

Fig. 23a compares the settlement ratio, S/H, at 5 cm from the abutment, where H is the height of the abutment model (48 cm), plotted against the number of loading cycle, N, when the backfill and support-ing ground was unreinforced or nailed. Fig. 21b compares the outward lateral displacement ratio at the abutment bottom, d/H. Without nailing, distinct active failure took place in the backfill with very large settlements even when D/H was as small as 0.2 %. Besides, due to the development of large pas-sive earth pressure, the abutment bottom was largely pushed out, which restrained the development of high passive earth pressure. These trends became more significant with an increase in the cyclic lateral displacement at the abutment top, D/H. With nails

connected to the top and bottom of the abutment, the active displacement at the abutment footing became essentially zero and the active failure in the backfill was effectively restrained, then the settlement in the backfill became much smaller. Yet, the settlement is not very small due to separated arrangements of nails in the backfill. Some additional measure is necessary to minimize the backfill settlement. Rela-tively high passive pressure developed because the dual ratchet mechanism was still somehow active in spite of nailing. However, the abutment supported by the nails was kept stable.

a)

b)

Figure 23. a) Settlement ratio; and b) lateral displacement ra-

tio at the abutment bottom, plotted against the number of cycle

in the cases with and without nailing (Tatsuoka et al., 2012b).

In summary, the abutment and backfill of a con-ventional type bridge become very stable against cy-clic lateral displacements caused by seasonal ther-mal expansion and contraction of the girder by reinforcing the backfill and supporting ground with nails connected to the top and bottom of the abut-ment. When further the girder is integrated to the abutments, the seismic stability increases substan-tially, as shown below.

Displacement-controlled load

system

Load cell 45

20

180

Units: cm

5 10 20 35

Laser sensor at different distances from the wall

Air-dried Toyoura sand Dr=90 %）

Laser sensor at the top of the wall

Laser sensor at the bottom of the wall

40

Nail reinforcement

Air-dried Toyoura sand (Dr= 90 %)

Displacement transducers Displacement

Transducer

Load cell

Displacement

Transducer

Precise gear

system to control

displacement

1,800

All units (mm)

50 200 (distance from the back of the facing)

100 350

45

02

00

Nail (40 mm in diameter

inclined at 10 degrees

from the horizontal)

Nine local

load cells

0 50 100 150 2004

3

2

1

0

S: measured at the center

between two nails

Nailed backfill

D/H= 0.2%

D/H= 0.4%

D/H= 0.6%

Number of cycles, N

Se

ttle

me

nt

ratio

, S

/H (

%)

(S a

t 5

cm

fro

m t

he a

bu

tme

nt;

& H

= 4

8 c

m)

D/H= 0.2%

D/H= 0.4%

D/H= 0.6%

Unreinforced

backfill

0 50 100 150 200-1

0

1

2

3

4

5

Unreinforced backfill

Range over

Range over

D/H=0.2%

D/H=0.2%

D/H=0.4%

D/H=0.4%

D/H=0.6%

D/H=0.6%

Passive

Active

Late

ral d

isp

lacem

ent

ratio

at

abu

tme

nt

bott

om

, d

/H (

%)

Number of cycles, N

Nailed backfill

Figure 24. Models for shaking table tests (n. b., the whole

model is shown only with GRS-IB) (Muñoz et al., 2012).

5. SHAKING TABLE TESTS


The following four models were tested (Fig. 22): CB (conventional type bridge): A pair of gravity

type abutments (without a pile foundation) sup-ported the girder via fixed and movable bearings (a hinge and a roller). The backfill was unreinforced air-dried Toyoura sand (Dr= 90 %). Due to a space limitation in the sand box, an additional mass was attached to the girder to simulate a 2 m-long girder. The length scale of the model relative to the as-sumed prototype was 1/10.

GRS-RW (geosynthetic-reinforced soil retaining wall bridge): A pair of sill beams supporting the girder via fixed and movable bearings was placed on the crest of GRS RWs having a FHR facing.

IB (integral bridge): The model was the same as GRS-IB except that the backfill was not reinforced.

GRS-IB (GRS integral bridge): The girder is con-nected to the facings with a pair of L-shaped metal fixtures (20 cm-long, 5 cm-wide and 3 mm-thick),

which did not exhibit the major resisting moment. The backfill was reinforced with grid reinforcement layers connected to the facing. Two other models with a cement-mixed backfill zone immediately be-hind the facing (GRS-IB-C and GRS-IB-C-T) were also prepared. The details are reported in Tatsuoka et al. (2009).

In the first series of shaking tests (on these models described in Fig. 24), twenty sinusoidal waves at a frequency fi of 5 Hz was exerted to the shaking table at each stage increasing the acceleration level incre-mentally by 100 gals (i.e., cm/sec

2) until the models

collapsed. In the second series (on models IB and GRS-IB), input motions at various frequencies were used to examine whether the conclusions from the first series is relevant for a wide range of fi.

Figure 25. Damped SDOF system.

The observed dynamic behaviours of the bridge

models were analyzed by assuming a damped single-degree-of-freedom (SDOF) system (Fig. 25). From the amplification ratio M (= the ratio of the response acceleration at the girder (αt) to the input accelera-tion (αb)) and the phase difference φ between αt and αb observed in every cycle, the tuning ratio β (= the ratio of the input frequency (fi) to the transient natu-ral frequency (f0)) and the transient damping ratio ξ were back-calculated. With each model, the value of β decreased and the value of ξ increased with an in-crease in the number of loading cycle N at the same input acceleration (αb) and with an increase in αb, as typically seen from Figs. 26 - 28. These trends were different among the different models due to their dif-ferent nonlinear load-deformation properties (Munoz et al., 2012). In the first series, all the models started failing when the resonant state was reached. Then, the β value increased exceeding the unity and even-tually collapse took place.

The test results showed that the response accelera-tion for a given input motion decreases and the pos-sibility to reach the resonant state decreases: 1) with an decrease in the initial value of β (i.e., with an in-crease in the initial value of fo, representing the ini-tial stiffness of the structure); and 2) with a decrease

surcharge: 1 kPa

[units：cm]

205.8

60.8

20

Toyoura sand (Dr=90%)

Additional mass (180kg)

Accelerometer

Accelerometer

AbutmentAbutment

Movablesupport

Fixedsupport

Movable

bearingFixed

bearing

Accelerometer

Gravity-type

abutmentLoad cells

CB

[units：cm]

205.8

60.8

surcharge: 1 kPa


Additional Mass (180kg)

Accelerometer

Accelerometer

35 Reinforcements made of a

phosphor bronze grid

1

3

4

5

6

7

89

2Movablesupport

Fixedsupport

Accelerometer

bearingbearing

GRS RW

60.8

20


surcharge: 1 kPa

[units：cm]

205.8

Additional mass (180kg)L-shapemetal fixture

Accelerometer

AccelerometerE

mb

ed

de

d a

cce

lero

me

ters

Accelerometer

20

20

43.5

3020105 5

A15 A6 A7 A8

A17

Acceleration extrapolated

A4

A10

A0

Accelerometer

L-shaped

metal

connector

IB

1

2

3

4

5

6

7

8

9

10

60.8

20

Toyoura sand Dr=90%)

surcharge: 1 kPa

[units：cm]

205.8

weight: 200kg

L-shape metal fixture

Model reinforcement made of

a phosphore bronze gridAccelerometer

Accelerometer

35

GRS-IB

AccelerometerL-shaped metal fixtureSurcharge of 1 kPa

Accelerometer Model geogrid

205.8[all units: cm]

85

35

51

Additional

mass: 200 kg60.8

20 35

6

Fixed (x= 0)

Input dis.: ub

Total displacement: ut

Mass: m

Sinusoidal input motion

Inertia = m･d2ut/dt2

Spring: kDashpot: c

Relative dis.: u Acceleration

amplification ratio,

M =

Phase difference, φ

/t b

Tuning ratio:

β=

Measured Unkowns

Damping ratio:

ξ=

= 2πf0

fi=

input freq.(Hz)

f0=

natural freq. (Hz)

in the increasing rate of β (i.e., the decreasing rate of f0) associated with an increase in N and αb. Further-more, 3) for a given input motion, the response ac-celeration decreases with an increase in the damping ratio. Finally, 4) the maximum response acceleration at which the model starts failing increases with an increase in the structural strength. It is shown below that the GRS integral bridge has advantages in all of these factors, therefore, it exhibits a very high seis-mic stability.

Figure 26. Response of model CB (fi= 5 Hz): I, II …. denote

loading stages; and the numerals in ( ) denote the number of

cycle at each stage (Muñoz et al., 2012).

Figure 27. Response of model IB (Muñoz et al., 2012).

Figure 28. Response of model GRS-IB (fi= 5 Hz) (Muñoz et

al., 2012)

a)

b)

Figure 29. Responses of four bridge models (fi= 5 Hz)

(Muñoz et al., 2012)

Fig. 29a shows the relationship between the input

and response accelerations, αt and αb. With CB, M=

αt/αb started increasing at the smallest αb, because of

the largest initial value of β and the largest increas-

0.8

1.0

1.2

1.4

1.6

1.8

2.0

0.0 0.5 1.0 1.5

Resp

on

se r

ati

o o

f accele

rati

on

, M

Tuning ratio, b (b= fi/f0)

I

II

IV

V

VI

VII (1)

VII (20)

1.0

0.8

0.6

0.5

0.4

0.3

x= 0.2

Damping

ratio:

III (2)

III (5)

III (20)

Tuning ratio, b= fi/f0

-2.0

-1.5

-1.0

-0.5

0.0

0.5

0.0 0.5 1.0 1.5

Ph

ase d

iffe

ren

ce,

f(r

ad

)


0.2

0.3

0.4

0.5

0.6

0.8

1

Series101

Series102

Series103

Series104

Series105

Series106

Series107

I

0.5

0.6

0.8

x= 1.0

Damping

ratio:

0.4

0.30.2

II

III (20)

VI

V

-/4


Solid curves:

theoretical

First resonant state

Tuning ration, β＝ fi/fo

Re

sp

on

se

ra

tio

of

acce

lera

tio

n, M

Ph

ase

diffe

ren

ce

, φ

(in

rad

ian

) Loading stage (No. of

cycle at each stage)

-2.0

-1.5

-1.0

-0.5

0.0

0.5

0.0 0.5 1.0 1.5

Ph

ase d

iffe

ren

ce,

f(r

ad

)


I II III

IV

V (1) VI (20)

VII (1)

VI (1)

V (20)

VII (20)

0.5

0.6

0.8

x= 1.0

Damping

ratio:

0.4

0.30.2


0.8

1.0

1.2

1.4

1.6

1.8

2.0

0.0 0.5 1.0 1.5

Resp

on

se r

ati

o o

f accele

rati

on

, M


III III

IV

V (1)

VI (20)

VII (1)

VI (1)

V (20)VII (20)

1.0

0.8

0.6

0.5

0.4

0.3

x= 0.2

Damping

ratio:


Solid curves: theoretical

Resonant state


Response r

atio o

f

accele

ration,

MP

hase d

iffe

rence, φ

(in r

adia

n)

Loading stage

(No. of cycle at

each stage)

0.8

1.0

1.2

1.4

1.6

1.8

2.0

0.0 0.5 1.0 1.5

Resp

on

se r

ati

o o

f accele

rati

on

, M


I

IIIVI VIII

VII IX

XXI (1)

XI (7)

1.0

0.8

0.6

0.5

0.4

0.3

x= 0.2

Damping

ratio:

XII


-2.0

-1.5

-1.0

-0.5

0.0

0.5

0.0 0.5 1.0 1.5

Ph

ase d

iffe

ren

ce,

f(r

ad

)


I IIIVI

VII

XIX

XI (20)

VIII

XI (1)

XII

0.5

0.6

0.8

x= 1.0

Damping

ratio:

0.4

0.30.2



Resonant state


Response r

atio o

f

accele

ration,

MP

hase d

iffe

rence, φ

(in r

adia

n)

Loading stage

(No. of cycle at

each stage)

0 500 1000 15000

500

1000

1500

GRS IB-T

GRS IB-C

IB

Response a

ccele

ration a

t girder

(gal)

Base acceleration at table (gal)

CB橋桁加速度

GRS RW

GRS IB

CB

1:1

; Moment when failure startes

at resonance

Re

sp

onse a

cce

lera

tio

n a

t th

e g

ird

er,

αt(g

al)

Input acceleration,αb (gal)

Resonance state (start

of failure)

Cement-mixed

backfill part

behind the

facing

0.0 0.5 1.0 1.50

500

1000

1500

CB (stages IV -VII)

GRS-IB-C-T

Response a

ccele

ration a

t th

e g

irder

(gal)

Tuning ratio, bfi/f

0

IB

CB (stages I - III)

GRS-RW

GRS-IB

GRS-IB-C

*


Re

sp

onse a

cce

lera

tio

n a

t th

e g

ird

er,

αt (g

al)

*Response

after failure

*

ing rate of β (i.e., the largest decreasing rate of f0)

with dynamic loading, as seen from Fig. 30. In Figs.

26 - 20, the multiple data points at the same input

acceleration αb with each model denote an increase

in β due to to a decrease in the f0 value with an in-

crease in the number of loading cycles (up to 20).

An increase in β resulted in an increase in M, there-

fore an increase in αt (Fig. 29b). Besides, with CB,

the strength, defined as the response acceleration at

which failure started, was smallest while the damp-

ing ratio at failure was smallest (Fig. 31). Conse-

quently, CB reached the resonant state fastest and

the collapse took place at the smallest αb.

a)

b)

Figure 30. Decrease in fo with an increase in the input accel-

eration of four bridge models (fi= 5 Hz) (Muñoz et al., 2012).

On the other hand, with GRS IB, the initial value

of β was smallest (i.e., the initial value of f0 was

largest) and the increasing rate of β (i.e., the decreas-

ing rate of f0) was smallest (Fig. 30), while the

damping ratio at failure was largest and the strength

was largest (Fig. 31). Consequently, GRS IB

reached the resonant state most slowly and the col-

lapse took place at the largest αb (Fig. 31). This very

high performance of GRS IB is due all to integration

of the girder, the facing and the backfill. The per-

formance of models GRS-RW bridge and integral

bridge (IB) was intermediate due to their intermedi-

ate levels of structural integration. As seen from

Figs. 29 - 31, the dynamic performance of GRS IB

was improved slightly by arranging a cement-mixed

backfill zone immediately behind the facing.

Figure 31. Accelerations and damping ratios at the start of

failure (i.e., at resonance) of six bridge models (fi= 5 Hz)

(Muñoz et al., 2012).

As seen from Fig. 31, the damping ratio, ξ, at the

start of failure increases with an increase in the dy-namic strength and the ξ value of model GRS IB is much higher than model CB. This is because, with an increase in the structural integration among the girder, facing and backfill, the material damping at the start of failure of: i) the material damping of the backfill and supporting ground as part of the bridge system; and ii) the dynamic energy of the girder and abutments dissipated via wave propagation toward the outside of the bridge system increase.

a)

b)

Figure 32. Response of model IB (fi= 2 ~ 20 Hz).

I

IIIII IV

V VI

VII

VIIIIX

X

XIXII

0 200 400 600 800 1000 1200 14000

5

10

15

20

25

30

35

40

Na

tura

l fr

eq

uen

cy o

f b

ridg

e,

f0 (

Hz)

Base acceleration amplitude, Amp[üb] (gal)

GRS-RW GRS-IB

CB

: Resonance state

(fi= 5Hz)


0 200 400 600 800 1000 1200 14000

5

10

15

20

25

30

35

40


Na

tura

l fr

eq

uen

cy o

f b

ridg

e,

f0 (

Hz)

I

IIIII

IV

V

VI

VII

VIIIIX

X

XIXII

GRS-IB-C

GRS-IB

IB

: Resonance state

(fi= 5Hz)

GRS-IB-C-T

XIII


0 500 1000 1500 20000.0

0.2

0.4

0.6

0.8

1.0 ..

..

GRS-IB-T

GRS-IB-C

GRS-IBIB

GRS-RW

damping ratio

Dam

pin

g r

atio x

at

failu

re

Acceleration at resonance (gal)

CB

Response acceleration at resonance, Amp[ut]resonance

(strength)

Base acceleration at resonance, Amp[ub]resonance

A single value of damping ratio

is presented for the respective models.

Response acceleration at resonance, (αt)resonance

(dynamic strength)

Input acceleration at resonance, (αｂ)resonance

GRS-IB-C-T

0 500 1000 1500

0

500

1000

1500

2 Hz

5 Hz

10 Hz

20 Hz

1 : 2

2 Hz

20 Hz

10 Hz

5 Hz

Resp

onse

Acc

ele

ration , A

r [g

al]

Base Acceleration , Ab [gal]

1 : 1

IB


Response a

ccele

ration,

αt(g

al)

10 Hz

20 Hz

2 Hz

Input frequency,

fi= 5 Hz

A Start of failure

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4

0

500

1000

1500

IB-f

Tuning ratio , b

Resp

onse

Acc

ele

ration ,

t [ga

l]

20 Hz

10 Hz5 Hz

2 Hz


Response a

ccele

ration,

αt(g

al)

Input frequency,

fi= 5 Hz

20 Hz10 Hz2 Hz

A

a)

b)

Figure 33. Response of model GRS IB (fi= 2 ~ 20 Hz).

Figs. 32 and 33 show the results from the second

series on models IB and GRS IB, in which the input acceleration αb was increased changing the input frequency fi at respective stages. In each figure, the data plotted in a broken circle denote the responses at different fi values for the nearly same αb. For the same f0 value, the β value increases with an increase in fi, which results in an increase in the M value be-fore the resonant state is reached and a decrease in M after the resonant state has been passed (i.e., the case denoted by A in Fig. 32). Therefore, with an in-crease in fi, the resonant state is reached at a lower αb and it becomes more likely to pass the resonant state without collapse taking place. It may also be seen from Figs. 32 and 33 that, not only at fi= 5 Hz but al-so at other fi values, the M value increases with an increase in β associated with an increase in the num-ber of loading cycle and αb; and GRS IB was much more dynamically stable than IB. Fig. 34 shows acceleration spectra for a wide range of damping ratio ξ of a typical very severe ground earthquake motion. The aseismic design of RC and steel structures is usually based on a re-sponse spectrum for ξ= 5 %, which is considered representative of the value at failure of these types of super-structure. On the other hand, the ξ value of model GRS IB reached as high as 60 % at the start of failure (Fig. 31). Therefore, the response spectra for ξ up to 70 % are also depicted in Fig. 34. The ini-tial value (before the start of seismic loading) of the

natural period T0 of ordinary prototype bridges, in-cluding ordinary GRS integral bridges, is shorter than the predominant period Tp of severe earthquake motions, which is about 0.35 second in the case of Fig. 34: i.e., the initial value of the natural frequency f0 is higher than the predominant frequency fp=1/Tp (= about 3 Hz). The implications of the shaking table test results are discussed below based on Fig. 34 by assuming that similar response spectrum are kept during a given major earthquake motion.

Figure 34. Acceleration response spectra for different damp-ing rations of NS component of the earthquake motion record-ed at JMA Kobe, 1995 Kobe Earthquake (by the courtesy of Dr Izawa, J., RTRI Japan).

With GRS integral bridges, when compared with

the other bridge types examined in the present study, the initial value of β= fp/f0 (<< 1.0) is lower, which means a lower initial value of the natural period T0 (=1/f0) (<< Tp), which results in a lower initial re-sponse acceleration. Besides, the increasing rate of β is lower, which means a lower increasing rate of T0. Then, the chance to reach the resonant state becomes lower. Furthermore, as the ξ value at failure is very high, even if having reached the resonant state, the response acceleration is kept very low. When ξ= 60 %, the M value at resonance is about 1.35 for the stationary sinusoidal input motions (Figs. 26a – 29a). In the case of Fig. 34, the magnification ratio M at resonance for an irregular input motion defined as the ratio of the response acceleration at T0 = Tp to the value at T0= 0.0 (at which M= 1.0) is also around 1.35. This observation indicates that the design re-sponse acceleration at the girder of a GRS integral bridge can be set to be substantially lower than ordi-nary RC and steel structure (e.g., a bridge girder supported by RC piers). Besides, it is reasonable to set the design maximum response acceleration at the girder of GRS integral brides to be lower than the other bridge types, despite a much higher strength. Based on the above, a design M value between 1.0 and 1.35 is reasonable for such a GRS integral bridge as the one examined in these shaking table tests. Yet, it is necessary to examine whether and how the damping ratio at the start of failure decreas-

0 500 1000 1500

0

500

1000

1500

2 Hz

5 Hz

10 Hz

20 HzResp

onse

Acc

ele

ration , A

r [g

al]


1 : 2

20 Hz

10 Hz

5 Hz

2 Hz

1 : 1桁での応答加速度

,αt(g

al)

台での入力加速度,α b (gal)

5 Hz

2 Hz

GRS IB


Response a

ccele

ration,

αt(g

al)

Input frequency,

fi= 10 Hz Start of failure

20 Hz

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4

0

500

1000

1500

after failure

GRS-IB-f

Tuning Ratio , b

20 Hz10 Hz5 Hz

Resp

onse

Acc

ele

ration ,

t [ga

l]

2 Hz

5 Hz

20 Hz

2 Hz


Response a

ccele

ration,

αt(g

al)

Input frequency,

fi= 10 Hz

0.1 0.5 1 50

1000

2000

3000

周期 (sec)応答加速度

(gal)

1995兵庫県南部地震神戸海洋気象台 NS成分

h=5% h=10% h=20% h=30% h=40% h=50% h=60% h=70%

Period T0 = 1/(frequency f0) (second)R

esponse a

ccele

ration (

gal) Damping ratio ξ (%)= 5

10

20

30

40, 50, 60 & 70

es with an increase in the girder weight under other-wise the same conditions.


The following five models (Fig. 35) were prepared

by reinforcing the conventional type bridge model

(CB) (Fig. 22a):

a)

b)

c)

Figure 35. Small models of conventional type bridge rein-

forced in different ways for shaking table tests, only visible

zones through windows (denoted by solid lines) presented (all

units are in mm: Tatsuoka et al., 2012b).

CB-L (integrated conventional type bridge): The girder was integrated to a pair of gravity-type abutments of model CB using a pair of L-shaped metal fixtures. The backfill was unreinforced air-dried Toyoura sand (Dr= 90 %).

CB-L-N1 (integrated CB with nailing): The backfill and part of supporting ground of model CB-L was reinforced with two layers of 40 cm-long nails on each side. Two nails of each top layer were con-nected to the top of the abutment and two of each bottom layer to the footing toe of the abutment via pin connectors. The model nails were a hollow cir-cular brass rod with a diameter of 4 cm, modeling a full-scale large-diameter nail with a diameter of 40 cm based on a length scale factor of 10. The sur-face of the model nail was made rough by gluing Toyoura sand particles.

CB-L-N2: The length of the bottom nails (40 cm) of model CB-L-N1 were made longer to 60 cm to in-crease the pull-out strength. This model exhibited the highest dynamic stability among those tested in the present study.

Although two other models similar to CB-L-N2

were tested (Tatsuoka et al., 2012b), the results

are not presented herein due to page limitations.

Two series of tests were performed. In the first se-

ries on all the models described above, twenty si-

nusoidal waves at a frequency fi of 5 Hz was exerted

to the shaking table at each stage increasing the ac-

celeration level incrementally by 100 gals. In the se-

cond series on model CB-L-N2, input motions at

various frequencies were used. The observed dy-

namic behaviours of the models were analyzed by

the theory of the SDOF system (Fig. 25). A typical

analysis is presented in Fig. 36. In this test, the shak-

ing was ceased before reaching the resonant state

due to the capacity of the shaking table.

Figure 36. Response of model CB-L-N2 (fi= 5 Hz).

From the results from series 1 (Figs. 36 – 39), it

may be seen that, by integrating the girder to the abutments and further by nailing the backfill and part of the subsoil, i) the initial β value decreased (i.e., the initial value of the natural frequency f0 in-creased); ii) the increasing rate of β (i.e., the de-creasing rate of f0) decreased; iii) the response accel-eration at resonance, at which failure started; increased; and iv) the damping ratio at the start of failure increased, all contributing to a substantial in-crease in the dynamic stability. It may also be seen that the dynamic stability increased by using longer nails. Tatsuoka et al. (2012b) reported that connect-ing the bottom nails to the foundation toe of the

L-shaped

metal fixture

CB-L Accelero-

meter

608 Accelerometer

CB-L-N1

608Accelerometer

Accelerometer

608Accelerometer

AccelerometerCB-L-N2

-2.0

-1.5

-1.0

-0.5

0.0

0.5

0.0 0.5 1.0 1.5

Ph

ase d

iffe

ren

ce, f

(rad

)

Tuning ratio, b = fi/f0

0.2

0.3

0.4

0.5

0.6

0.8

1

Series101

Series102

Series103

Series104

Series105

Series106

Series107

Series108

Series109

Series1

I

0.5

0.6

0.8

x= 1.0

Damping

ratio:

0.4

0.30.2

II

VIIV

III V

VII

VIII

IX

X (20)

0.8

1.0

1.2

1.4

1.6

1.8

2.0

0.0 0.5 1.0 1.5

Resp

on

se r

ati

o o

f accele

rati

on

s, M

Tuning ratio, b = fi/f0

III

IVV

VI

IX

1.0

0.8

0.6

0.5

0.4

0.3

x= 0.2

Damping

ratio:

III

VIII

VII

X (20)


End of shaking before

reaching the resonant state


Response r

atio o

f

accele

ration,

MP

hase d

iffe

rence, φ

(in r

adia

n)

Loading stage (No. of

cycle at each stage)

abutment (as models CB-L-N1 and N2) is essential to maintain the contact of the bottom of the footing with the subsoil, thereby to maintain a high dynamic stability.

a)

b)

Figure 37. Responses of five bridge models (fi= 5 Hz).

Figure 38. Decrease in the natural frequency with an increase

in the input acceleration of four bridge models (fi= 5 Hz).

The behavior of model CB-L-N2 for various input frequencies (series 2) is presented in Fig. 40. It may be seen that the magnification of acceleration chang-es by changes in the tuning ratio, β, in the course of dynamic loading in the same way as series 1. This means that the conclusions from series 1, described above, are relevant also for a wide range of fi. In summary, it can be concluded that the dynamic stability of conventional type bridge can be substan-tially improved by the reinforcing technology de-

picted in Fig. 17 (i.e.., the NRS integrated bridge) to the similar level as GRS integral bridge (Fig. 16).

Figure 39. Accelerations and damping ratios at the start of fail-

ure (i.e., at resonance) of four bridge models (fi= 5 Hz).

a)

b)

Figure 40. Response of modelCB-L-N2 (fi= 2 ~ 20 Hz).

3.3 Dynamically brittle versus ductile behaviour

Tatsuoka et al. (1988) reported the performance of various types of retaining wall (RW) during the 1995 Kobe Earthquake. Based on the aboe and those from a series of shaking table tests, they concluded that the dynamic behaviour of gravity-type RWs with unreinforced backfill is generally dynamically very brittle in the sense that the failure starts at a rel-atively low ground acceleration and full collapse may quickly take place subsequently. On the other hand, the dynamic behaviour of GRS RWs with a FHR facing (Fig. 2) is dynamically ductile in the

0 500 1000 15000

500

1000

1500

CB-L-N2 (resonance not reached)

Re

sp

onse a

cce

lera

tion a

t g

irder,

t (

gal)

Base accleration at table, b (gal)

CB応答加速度

; Moment when failure startes

at resonance

CB-L

CB

CB-L-N1

1:1

0.0 0.5 1.0 1.50

500

1000

1500

CB-L-N2

CB (stages IV - VII)

Response a

ccele

ration a

t girder,

t

(gal)

Tuning ratio, b= fi/f

0

CB応答加速度

CB (stages I - III)

CB-L-N1

CB-L

0 500 10000

10

20

30

Na

tura

l fre

qu

en

cy o

f th

e b

rid

ge

, f 0

(H

z)

CB

CB-LCB-L-N1

CB-L-N2

Before

resonance

Base acceleration at respective stages, b (gal)

Resonance state

XIXVIIIVII

VI

V

IV

III

II

I

0 500 1000 15000.0

0.2

0.4

0.6

0.8

Da

mp

ing

ra

tio

x a

t re

so

na

nce

CB

CB-L

CB-L-N1

CB-L-N2

(before resonance)

Acceleration at resonance, (gal)

Base acceleration at resonance, (b)

resonance

Response acceleration at resonance, (t)

resonance

(strength)

0 500 1000 1500

0

500

1000

1500

Resp

onse

Acc

ele

ration , A

r [g

al]


2 Hz

5 Hz

10 Hz

20 Hz

5 Hz

20 Hz

10 Hz

2 Hz

1 : 2

1 : 1

2 Hz

CB-L-N2


Response a

cce

lera

tion,

αt(g

al) Input frequency,

fi= 10 Hz

Start of

failure

20 Hz

5 Hz

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4

0

500

1000

1500

CB-L-N2-f

Tuning ratio , b

20 Hz10 Hz5 Hz2 Hz

Resp

onse

Acc

ele

ration ,

t [ga

l]

5 Hz 20 Hz2 Hz


Response a

ccele

ration,

αt(g

al)

Input frequency,

fi= 10 Hz

sense that the failure starts at a relatively high ground acceleration and full collapse may take place only after a relatively large increase in the ground acceleration. Based on the above, they proposed to use a higher design seismic coefficient in the quasi-static limit equilibrium-based stability analysis with gravity type RWs than with GRS RWs under other-wise the same conditions.

However, they did not show the mechanism for brittle and ductile behaviours of RWs. The results from the shaking table tests shown above indicate that: 1) the dynamic behaviour of conventional type

bridge is indeed dynamically brittle while the one of GRS integral and NRS integrated bridges is ductile; and

2) the dynamically brittle or ductile behaviour of soil structures results from: a) a low or high ini-tial value of the natural frequency f0; b) a high or low decreasing rate of f0 in the course of dynam-ic loading; c) a low or high response acceleration that the soil structure can survive; and d) a low or high damping ratio of the structure.

a)

b)

Figure 41. A full-scale model of GRS integral bridge con-

structed at Railway Technical Research Institute: a) overall

structure; and b) the left-side abutment under construction.

6. FULL SCALE MODELS AND PROTOTYPES

6.1 Full-scale models

A full-scale model of GRS integral bridge was con-structed during a period of 2008 – 2009 at Railway Technical Research Institute (Fig. 41). The abut-ments were constructed by excavating the backfill of a pair of full-scale models of GRS-RW with FHR facing constructed in 1988 (Tatsuoka et al., 1997). A high constructability of GRS integral bridge was confirmed. The behaviour during construction was

observed and the long-term behaviour is now being observed.

At the end of 2009, a full-scale model of conven-tional type bridge was constructed and the backfill of both of the 6.05 m-high abutments was reinforced with 40 cm-diameter nails, then, the 13.31 m-long steel girder was integrated to a pair of abutments (Fig. 42; Suga et al., 2011). A high constructability of this bridge reinforcement method was confirmed.

In January and February of 2012, full-scale lateral loading tests of the GRS integral and NRS integrated bridge models were performed. The results will be reported in the near future.

a)

b) Figure 42. Full-scale model of conventional type bridge re-

inforced by nailing of the backfill and integrating the girder to

the abutments; a) overall structure; and b) a picture of the mod-

el immediately after completion (Suga et al., 2011).

6.2 Prototype GRS integral bridge

In 2011, the first prototype GRS integral bridge was constructed at the south end of Hokkaido for a new high-speed train line (Fig. 43). The design maximum train speed is 260 km/h. The estimated construction cost is about a half of the one for a box girder type bridge, which is the most conventional solution in this case (Watanabe, 2011). The bridge is heavily in-strumented to observe the behaviour during con-struction and after opening to service. Several GRS integral bridges for railways, including those recon-structed to restore bridges that collapsed by tsunami during the 2011 Great East Japan Earthquake Disas-ter, are now at the stage of planning and design.

Earth pressure cells Earth pressure cells

Cement-mixed backfill Well-graded gravelly soil

5.5

5 m

14.75 m

27 November 2008

13.3

Steel girder

Abutment

Steel girder

Abutment

Integration

Large diameter nails

2.9

Elevation

Plan

2.9

56

.1

[All units in m]

Large diameter nails

a)

b)

Figure 43. First GRS integral bridge, Kikonai at the south

end of Hokkaido, for a new bullet train line (11.7 m wide): a)

overall structure; and b) abutments under construction, summer

2011 (by the courtesy of Japan Railway Construction,

Transport and Technology Agency).

6.3 2011 Great East Japan Earthquake Disaster

The girders and approach fills behind the abutment of a great number of road and railway bridges (more than 300) were washed away by tsunami of the 2011 Great East Japan Earthquake Disaster (Kosa., 2012), as typically seen from Figs. 44 - 46. The above was due to the fact that a girder supported by bearings has a very low resistance against uplift and lateral forces of tsunami while the unreinforced backfill is very weak against erosion by over-flow of tsunami. Connectors and anchors that had been arranged to prevent dislodging of the girders from the abutments and piers by seismic loads could not prevent the flow away of the girders by tsunami forces. These cases showed that the bearings and unreinforced backfill are two weak points of the conventional type bridge not only for seismic loads but also for tsunami current.

Figure 44. A view from the seaside of Namiitagawa bridge,

between Namiitagawa and Kirikiri stations, Otsuchi-cho,

Yamada Line, Iwate Prefecture, East Japan Railway.

Figure 45. A view from the upstream of Tsuyagawa Bridge,

between Motoyoshi and Rikuzenkoizumi stations, Kesen-numa

line, East Japan Raiway.

a)

b)

Figure 46. a) A view from the seaside; and b) the back of the

right bank abutment of Yonedagawa bridge, between Rikuchu-

noda and Rikuchu-tamagawa stations, Noda village, Iwate Pre-

fecture, North-Rias Line, Sanriku Railway.

Figure 47. Collapse mechanisms of conventional type coastal

dyke by over-flowing tsunami current.

6.4 Proposal of tsunami-resistant bridges and dykes

The resistance against the uplift and lateral forces of tsunami of the girder of GRS integral bridge is sub-stantially higher than the conventional type bridge. This is because the girder of GRS integral bridge is integrated to the facings that are connected to the massive approach fill with geogrid layers, while the approach fill that is reinforced and has full-height rigid facings connected to the geogrid layers is very stable against tsunami current. In this case, the ap-proach fill should also have facings on both sides in parallel of the bridge axis to be stable against tsuna-mi current. Even with GRS integral bridges, neces-

GCM: Ground improvement by cement -mixing

5.04 2.21.0

5.42.21.0

0.7 0.7

[All units in m]

GCM

Road surface

12.0

GCMGL= 5.0

Original

ground

Backfill (uncemented)Backfill (cement-mixed

gravelly soil)

6.1

10.75

EastWest RC slab

0.6

0.6

Unreinforced

backfill

Concrete

panel facing

Drainage ditch

Concrete slab on the crest

Leveling pad

Sheet pile

FoundationFoot protection work

Tetrapod

Recurved parapet

Concrete panel facing

Seaside

S

U

sary provisions should be made for protecting the abutments from scouring in the supporting ground by tsunami current.

Most of conventional type coastal dykes comprise unreinforced backfill with concrete panel facings covering upstream and downstream slopes and crest. When subjected to deep over-flow of tsunami cur-rent caused by the 2011 Great East Japan Earth-quake Disaster, at many places, the coastal dykes fully collapsed by the following mechanisms (Fig. 45) and had totally lost their function when attached by subsequent tsunamis. Firstly, a strong tsunami current flowing down the down-stream slope scored in the ground in front of the toe of the downstream slope (as denoted by letter ‘S’ in Fig. 47), by which the concrete facing on the downstream slope was de-stabilized and washed away; and/or the concrete fac-ings at the crest and the upper part of the down-stream slope were peeled off by associated strong uplift forces (as denoted by the letter ‘U’ in Fig. 47) and washed away. Then, the unreinforced backfill was exposed to tsunami current and eroded quickly from the crest and downstream slope. Subsequently, the drawback of a tsunami damaged the seaside slope by the same mechanism as described above. Finally, the full section disappeared.

Figure 48. GRS coastal dykes as a tsunami barrier that can

survive deep over-flow of tsunami current.

On the other hand, coastal dykes that comprise the

geogrid-reinforced backfill with continuous lightly steel-reinforced concrete facings that are firmly con-nected to the reinforcement, like the GRS RWs illus-trated in Fig. 2, should have a much stronger re-sistance against over-flowing tsunami current. Fig. 48 shows several types that can be proposed at this moment. Yamaguchi et al. (2012) performed a series of small model tests and showed that the stability of these types of GRS coastal dykes (Fig. 48) have a very high resistance against flow away by over-flowing tsunami current, much higher than the ordi-nary embankment with concrete facing (Fig. 47).

Tatsuoka and Tateyama (2012) proposed to con-

struct GRS integral bridges (Fig. 16) and GRS em-

bankments/dykes (Fig. 48) to restore the conven-

tional type bridges of railways and roads that col-

lapsed by tsunamis of the 2011 Great East Japan

Earthquake Disaster. Fig 49 illustrates a proposal for

coastal railways. These technologies can also be

used to newly construct coastal railways and roads

that should survive earthquakes and tsunamis.

Figure 49. Coastal railway that can survive a great tsunami

(Railway Technical Research Institute, Japan, 2011).

7. CONCLUSIONS

Geosynthetic-reinforced soil retaining walls (GRS RWs) having staged-constructed full-height rigid (FHR) facing have been constructed as important permanent RWs for a total length of more than 130 km since 1989 until today in Japan, mainly for rail-ways and also for roads and other types of infra-structure. Its current popular use is due to a high cost-effectiveness with high long-term performance and high seismic stability. This success can be at-tributed to: i) the use of a proper type of reinforce-ment (i.e., geogrids for cohesionless soil and nonwoven/woven geotextile composites for high-water content cohesive soil); ii) construction of a FHR facing by such a staged construction procedure that, after the major deformation of the supporting ground and backfill has taken place, fresh concrete is cast-in-place on the wrapped-around wall face of completed geosynthetic-reinforced backfill so that reinforcement layers are firmly connected to the fac-ing; and iii) taking advantage of the rigidity of the facing in design.

A number of embankments and conventional type RWs that collapsed during recent severe earth-quakes, heavy rains, floods and storms were recon-structed to GRS RWs of this type. It was validated that this technology is also highly cost-effective in reconstructing collapsed old soil structures.

The GRS integral bridge was developed, which comprises an integral bridge and geosynthetic-reinforced backfill with staged-constructed FHR fac-ing. The results from model tests in the laboratory,

Planar reinforcement (e.g., geogrid) Planar reinforcement

(e.g., geogrid)

FHR facing (connected to

reinforcement layers)

Planar reinforcement (e.g., geogrid)

Concrete facing (connected to reinforcement,

not allowing the backfill to flow out from openings

Foot protection to

prevent scouring

GRS integral

bridge

GRS RW with FHR

facing

GRS coastal

dyke

Evacuation shelters

along elevated railway

full-scale model tests and construction of a prototype showed the following superior characteristic features over other types of bridge: a) lower cost for con-struction and maintenance; b) less detrimental ef-fects of seasonal thermal deformation of the girder; and c) a much higher seismic stability. Feature c) is due to: 1) a high initial natural frequency; 2) a low decreasing rate of the natural frequency during dy-namic loading; 3) a high energy dissipation capacity at failure; and 4) a high dynamic strength, all by full integration of the girder, the abutments and the rein-forced backfill.

It is argued that GRS integral bridges (with geogrid-reinforced approach fill having facings on its three sides) and GRS embankments/dykes with facings connected to the reinforcement also have a very high resistance against tsunami. It is proposed to reconstruct bridges and embankments of railways and roads that collapsed by tsunami of the 2011 Great East Japan Earthquake Disaster to these types of structures. It is also proposed to newly construct tsunami-resistant GRS integral bridges and GRS embankments/dykes for coastal railways and roads.

8. ACKNOWLEDGEMENTS

The authors would like to sincerely thank their pre-

vious and current colleagues at the University of

Tokyo, Tokyo University of Science and Railway

Technical Research Institute, Japan, for their great

help to a long-term research performed to develop

the GRS technologies described in this paper.

9. REFERENCES

England, G, L., Neil, C, M. and Bush, D, I. 2000. Integral

Bridges, A fundamental approach to the time-temperature loading problem. Thomas Telford.

Kosa, K. 2012. Damage analysis of bridges affected by tsuna-mi due to Great East Japan Earthquake. Proc. International Sym. on Engineering Lessons Learned from the 2011 Great East Japan Earthquake, March, Tokyo, Japan, 1386-1397.

Koseki, J., Bathurst, R.J., Guler, E., Kuwano, J. and Maugeri, M. 2006. Seismic stability of reinforced soil walls. Proc. 8th International Conference on Geosynthetics, Yokohama, 1: 51-77.

Koseki, J., Tateyama, M., Watanabe, K. and Nakajima, S. 2008. Stability of earth structures against high seismic loads.” Keynote Lecture, Proc. 13

th ARC on SMGE, Kolka-

ta, Vol. II. Koseki, J. 2012: Use of geosynthetics to improve seismic per-

formance of earth structures. Mercer Lecture 2011. Geotex-tiles and Geomembranes, 34 (2012): 51-68.

Morishima, H., Saruya, K. and Aizawa, F. 2005. Damage to soils structures of railway and their reconstruction.” Special Issue on Lessons from the 2004 Niigata-ken Chu-Etsu Earthquake and Reconstruction, Foundation Engineering and Equipment (Kiso-ko), Oct., 78-83 (in Japanese).

Munoz, H., Tatsuoka, F., Hirakawa, D., Nishikiori, H., Soma, R., Tateyama, M.

and Watanabe, K. 2012. Dynamic stabil-

ity of geosynthetic-reinforced soil integral bridge. Geosynthetics International, 19(1): 11-38.

Railway Technical Research Institute, Japan. 2011. Technical solutions proposed to restore railways damaged by earth-quakes (in Japanese).

Suga, M., Kuriyama, R., Tateyama, M., Kouda, M., Sugimoto, I. and Kobayashi, Y. 2011. Reinforcing method of bridge by integration of steel girder, abutment and backfill, Proc. 46

th Japan Conf. on Geotechnical Engineering, JGS, Kobe,

Paper No. H-06, pp.1499-1500 (in Japanese). Tateyama, M., Tarumi, H. and Fukuda, A. 1996. Development

of a large diameter short reinforced anchor by cement-mixing method, Grouting and Deep Mixing, Proc. cond Int. Conf. on Ground Improvement Geosystems, To-kyo, pp. 759-765.

Tateyama, M. 2012. Damage to railways by the 2011 Great East Japan Earthquake Geosynthetics Engineering Infor-mation, 28(1), March: 1-8 (in Japanese).

Tatsuoka, F. 1992. Roles of facing rigidity in soil reinforcing. Keynote Lecture, Proc. Earth Reinforcement Practice, IS-Kyushu ‘92 (Ochiai et al. eds.), 2: 831-870.

Tatsuoka, F., Tateyama, M, Uchimura, T. and Koseki, J. 1997. Geosynthetic-reinforced soil retaining walls as important permanent structures. Mercer Lecture, Geosynthetic Inter-national, Vol.4, No.2, pp.81-136.

Tatsuoka, F., Koseki, J., Tateyama, M., Munaf, Y. and Horii, N. 1998. Seismic stability against high seismic loads of geosynthetic-reinforced soil retaining structures.” Keynote Lecture, Proc. 6

th Int. Conf. on Geosynthetics, Atlanta, 1:

103-142. Tatsuoka, F., Tateyama, M., Aoki, H. and Watanabe, K. 2005.

Bridge abutment made of cement-mixed gravel backfill. Ground Improvement, Case Histories, Elsevier Geo-Engineering Book Series, Vol. 3 (Indradratna & Chu eds.), 829-873.

Tatsuoka, F., Tateyama, M., Mohri, Y. and Matsushima, K. 2007. Remedial treatment of soil structures using geosynthetic-reinforcing technology. Geotextiles and Geomembranes, 25 (4 & 5): 204-220.

Tatsuoka, F., Hirakawa, D., Nojiri, M., Aizawa, H., Tateyama, M. and Watanabe, K. 2008a. Integral bridge with geosynthetic-reinforced backfill. Proc. First Pan American Geosynthetics Conference & Exhibition, Cancun, Mexico, 1199-1208.

Tatsuoka, F., Hirakawa, D., Aizawa, H. Nishikiori, H. Soma, R and Sonoda, Y. 2008b. Importance of strong connection be-tween geosynthetic reinforcement and facing for GRS inte-gral bridge. Proc. 4

th GeoSyntheticsAsia (4

th Asian Regional

Conference on Geosynthetics), Shanghai. Tatsuoka, F., Hirakawa, D., Nojiri, M., Aizawa, H., Nishikiori,

H., Soma, R., Tateyama, M. and Watanabe, K. 2009. A new type integral bridge comprising geosynthetic-reinforced soil walls. Gesynthtetics International, IS Kyushu 2007 Special Issue, 16(4): 301-326.

Tatsuoka, F., Koseki, J. and Tateyama, M. 2010a. Introduction to Japanese codes for reinforced soil design, Panel Discus-sion on Reinforced Soil Design Standards, Proc. 9

th Inter-

national Conference on Geosynthetics, Brazil, 245-255. Tatsuoka, F., Hirakawa, D., Nojiri, M., Aizawa, H., Nishikiori,

H., Soma, R., Tateyama, M. and Watanabe, K. 2010b. Clo-sure to Discussion on “A new type of integral bridge com-prising geosynthetic-reinforced soil walls.” Gesynthtetics International, 17(4): 1-12.

Tatsuoka, F., Kuroda, T. and Tateyama, M. 2012a. Research and practice of GRS integral bridges, Proc. EuroGeo 5, Va-lencia, September (to appear).

Tatsuoka, F., Munoz, H., Kuroda, T., Nishikiori, H., Soma, R., Kiyota, T., Tateyama, M. and Watanabe, K. 2012b. Stabil-ity of existing bridges improved by structural integration and nailing, Soils and Foundations, 52(3) (to appear).

Tatsuoka, F. and Tateyama, M. 2012. Geosynthetic-reinforced soil structures for railways in Japan, Keynote Lecture, Proc. International Conference on Ground Improvement and Ground Control (ICGI 2012) (Indraratna et al., eds.)

Watanabe, K. 2011. Application of GRS integral bridge tech-nology to Hokkaido High-Speed Train Line (Shinkansen). Journal of Japan Railway Civil Engineering Association, 49(10): 83-86 (in Japanese).

Yamaguchi, S., Yanagisawa, M., Uematsu, Y., Kawabe, S., Tatsuoka, F. and Nihei, Y. 2012. Experimental evaluation of the stability of GRS coastal dyke against over-flowing tsunami. Proc. 47

th Japan Conf. on Geotechnical Engineer-

ing, JGS, Hachinohe (in Japanese).

grs structures recently developed and constructed for ... · fhr facing for yamanote line over chuo...

Documents