SPECIAL ISSUE OF SOILS AND FOUNDATIONS 219-234, Jan.

1996
Japanese Geotechnical Society

BUILDING DAMAGE ASSOCIATED WITH GEOTECHNICAL PROBLEMS

KOHJI TOKIMATSUi),HATSUKAZUMIZUNOii)and MASAAKI KAKURAIiii)

ABSTRACT
An overview of the geotechnical aspects of the building damage in the 1995 Hyogoken-Nambu earthquake is
presented. It is shown that (1) the damage to buildings was caused by either strong ground shaking or ground prob-
lems, each of which combined with inadequate design of buildings, and (2) the pattern, extent, and distribution of the
damage were significantly affected by local soil conditions as well as ground failures including soil liquefaction.
Typical features of the ground problems are settlement and/or tilting of buildings. Possible causes of the foundation
damage include: (1) horizontal forces and overturning moments imposed on the foundation from the superstruc-
tures, (2) kinematic forces acting on deep foundations due to shear deformation of soils, (3) reduction in bearing
capacity due to ground failures including liquefaction, and (4) lateral spreading. The damage to pile foundations
caused by Categories (1) and (4) has posed particularly serious problems. Despite extensive soil liquefaction, the engi-
neered buildings on Port and Rokko Islands performed relatively well, probably due to (1) ground surface motions
attenuated by soil liquefaction, and (2) indirect effects of foundation designs related to consolidation settlement. The
buildings for which remedial measures were undertaken to mitigate liquefaction hazards, were found to have
performed well.

Key words: building, earthquake damage, liquefaction, pile, settlement, spread foundation (IGC: E8)

make a completely thorough investigation, important

INTRODUCTION
The Hyogoken-Nambu earthquake (M = 7.2) that oc-
curred on January 17, 1995, with its epicenter located
north of Awaji Island, struck the southern (Nambu)
Hyogo prefecture, causing extensive damage. More than
70,000 wooden private houses and about 3,000 commer-
cial and residential buildings totally collapsed. A loss of
over 5,500 lives resulted.
The primary cause of the building damage was the
strongest ground shaking in the recorded history in this
region. The "focusing phenomena" of seismic waves
and/ or "local site effects" appear to have been involved
since the damage was concentrated in a belt stretching
east and west in Hyogo prefecture. Ground failures in-
cluding soil liquefaction were also involved because a
large number of buildings settled and/ or tilted, with
their superstructures remaining intact. Although few
fatalities resulted from the foundation distress, most of
those buildings require extensive restoration including un-
derpinning. In this paper, the building damage associ-
ated with geotechnical problems during the 1995 event is
described, with emphasis placed on the damage caused
by soil liquefaction. Since the area affected is too large to
aspects of the problems may still be missed.
EFFECTS OF SOIL CONDITIONS ON BUILDING
DAMAGE
Geological Setting and Subsurface Soil Conditions
Figure 1 shows the geomorphological map of the
hardest hit area. Several known active faults run parallel
to the Rokko Mountains where granite outcrops are
found. Many small rivers originating in the mountains
have formed alluvial fans and flood-plains of Holocene
deposits in the south. Expanding further south are
reclaimed fills that overlie the sedimentary Holocene clay
layer. The Holocene deposit in turn overlies the upper
Pleistocene deposit, the layers of the Osaka and Infra-
Osaka Groups, and granite at a depth of the order of 1
km, as shown in Fig. 2.
The thicknesses of reclaimed fills are about 15-25
meters below Port and Rokko Islands, and less than
about 15 meters in other reclaimed land areas. The top of
the upper Pleistocene deposit appears at depths of about
5 m at the foot of the mountains, about 10 m in the cen-
ter of the alluvial fans, and 30-45 m in the reclaimed is-

i)
Professor. Tokyo Institute of Technology, O-okayama 2-12-1, Meguro-ku, Tokyo 152.
ii)
Program Director, Building Research Institute.
iii)
Senior Chief Researcher, Takenaka Corporation.
Manuscript was received for review on August 17, 1995.
Written discussions on this paper should be submitted before August 1, 1996, to the Japanese Geotechnical Society, Sugayama Bldg. 4F,
Kanda Awaji-cho 2-23, Chiyoda-ku, Tokyo 101, Japan. Upon request the closing date may be extended one month.

219
220 TOKIMATSU ET AL.

Fig. 1. Geomorphological map of affected area

lands, as shown in Fig. 2. The upper Pleistocene deposit

has been considered as the bearing stratum for multistory
buildings in this region.
Figure 3 shows a map indicating the area with ground-
water tables less than 2.5 meters. The zoning is based on
the boring logs complied mainly before 1980 (Kobe City,
1980), and thus may not correspond to the groundwater
conditions during the 1995 event precisely. In any case,
the zones with shallow groundwater not only lie in most
of the floodplains and reclaimed land areas, but extend
along the foot of the alluvial fans. The groundwater
depths in the central city zones of Port and Rokko
Islands range mostly from 3 to 6 meters, because of the
high ground levels (Oh-oka, 1995).

Effect of Soil Conditions on Damage to Superstructures

Fig. 2. Schematic diagram showing geologic cross section Figure 4 shows the distribution of building damage

Fig. 3. Map showing area with shallow groundwater table

 DAMAGE TO BUILDING FOUNDATION 221

Fig. 4. Map showing area with heavy damage to superstructure of both wooden houses and engineered buildings

(Chuo-Kaihatsu, 1995). In the heavily damaged area, a

large number of multistory engineered buildings sus-
tained either partial or complete collapse typically of a
single story. Such failures were concentrated in the build-
ings constructed before 1971 and 1981 when the old
Japanese building code was used. A large number of
traditional wooden-frame houses collapsed, crushing
many people to death.
The heavy damage was concentrated in a 1-km wide
belt that lies east and west in between the Rokko moun-
tains and the coastline. Also shown in Fig. 4 is the ap-
proximate area where field manifestation of soil liquefac-
tion was evident after the quake. The damaged belt is
bounded on the south approximately by the border be-
tween liquefied and non-liquefied zones. This border cor-
responds roughly to the old coastline, indicating that
most of the fill materials liquefied. The numbers shown
in Fig. 4 are the peak horizontal ground accelerations
registered during the main shock. The peak value ranges
from 0.3-0.6 G in the Rokko mountains and reclaimed
lands, and from 0.7-0.8 G in the heavily damaged area.
The ground motions in the damaged area were thus am-
plified by a factor of about 1.5-2 within the deep sedimen-
tary layer; however, those in the reclaimed lands were
almost the same as or even less than those on the rock.
This is probably due to "isolation effects" of the liq-
uefied soils.
Figure 5 shows the time histories of accelerograms ob-
tained at the downhole strong motion station in Port
Island (CEORKA, 1995). The upward traveling motions
particularly in the high frequency range were attenuated
in the liquefied fill at depths between 0-16 m. Thus, local
site effects including those resulting from soil liquefac-
tion appears to be major factors having reduced the
damage to superstructures along the coastline. In addi- Fig. 5. Time histories of accelerations and relative displacements be-
tion, the high ground motions in the heavily damaged tween layers at strong motion downhole array station on Port
area may have resulted from "focusing phenomena" of Island
222 TOKIMATSU ET AL.

Fig. 6. Map showing alluvial fans and floodplains with shallow groundwater tables

seismic waves propagating from the dipping rocks under-
neath and on the north edge of the sedimentary deposit
(Fig. 2).
Within the damaged belt stretching east and west,
some areas exist where structural damage was slight, e.g.,
west and east of Sannomiya. A comparison of Figs. 1, 3,
and 4 suggests that heavy structural damage occurred in
the area with shallow groundwater and vice versa, as
shown in Fig. 6. This indicates that soil conditions near
the surface had significant effects on the distribution of
damage.
Effect of Soil Conditions on Damage to Building Founda-
tions
Figure 7 shows the distribution of buildings with foun-
dation problems in Hyogo prefecture, prepared based on
our reconnaissance surveys as well as others (e.g.,
Tamura, 1995). The circles in the figure correspond to
pile foundations, and triangles denote spread founda-
tions. It should be noted that the data in this figure are
only a fraction of the entire number, and many have been
missed in Fig. 4. Most of the foundation problems were
found to be concentrated in the reclaimed lands between
Kobe and Nishinomiya cities, and in the Muko River del-
ta. This indicates that soil liquefaction was the primary
cause of the foundation distress. Typically, the tradition-
al Japanese wooden houses with low foundation rigidity
almost collapsed due to differential settlement. The en-
gineered buildings and modern wooden houses with rigid
foundation beams, by contrast, only settled and tilted
with their superstructures remaining intact or with little
damage. A considerable number of buildings located in
the alluvial fan areas with shallow groundwater also ex-
perienced foundation problems.

Fig. 7. Distribution of buildings with foundation problems

 DAMAGE TO BUILDING FOUNDATION 223

ported on point bearing piles tilted, probably because of

FOUNDATION DAMAGE DUE TO SOIL
LIQUEFACTION
Areas Other Than Port and Rokko Islands
In Kobe city, liquefaction-induced damage to building
foundations was visible throughout its coastline. These
include such districts as Wadamisaki, Hyogo-ku;
Shinko-cho (reclaimed before 1922) and Onohama-cho
(1935-47), Chuo-ku; Maya Wharf and Nadahama-
higashi-cho (1953-67), Nada-ku; Mikagehama and
Sumiyoshihama (1960-68), Uosakihama (1961-68) and
Fukaehama (1964-1970), Higashinada-ku. Most of these
areas were reclaimed either before or just after World
War II, using decomposed granite from the Rokko
Mountains, and were developed later.
In these areas, many old buildings suffered serious
ground problems regardless of their foundation type, as
schematically shown in Fig. 8(a)-(f). The buildings sup-
(a) Failure of pile head (b) Failure of pile due to
due to lateral force overturning moment
the damage to pile heads due to horizontal forces (Fig.
8(a)) and/ or overturning moment imposed on the foun-
dation from the superstructure (Fig. 8(b)). The buildings
supported on friction pile foundations or spread founda-
tions settled, often accompanied by severe tilt, due to
bearing capacity failure induced by soil liquefaction (Fig.
8(c), (e)) and/ or overturning moment (Fig. 8(d), (f)).
The damage patterns for these buildings are very similar
to those observed during past earthquakes.
It is conceivable that neither the effects of soil liquefac-
tion nor the effects of horizontal forces were considered
in the aseismic design of those buildings. Typical exam-
ples include a three-story building in Wadamiya-dori,
Hyogo-ku (A in Fig. 7), that settled about 1.1 meters and
tilted considerably; and a two-story building in Uosakiha-
ma (B in Fig. 7) that subsided more than 1.5 meters.
Soil liquefaction extended somewhat inland but did
contribute to the damage to some building foundations.
Typical examples include a building in Tamon-dori,
Chuo-ku (C in Fig. 7), that suffered severe settlement and
tilting; and four-story buildings of Uosaki Junior High
School, in Uosaki, Higashinada-ku (D in Fig. 7), that set-
tled and tilted up to 3 degrees.
In the center of Fukaehama, Higashinada-ku (E in Fig.
7), several old reinforced concrete buildings two to five
stories high settled and/ or tilted considerably (Photo. 1).
A newly-constructed five story building founded on
precast concrete piles also sustained severe tilt. An exca-
vation survey of the upper portion of its foundation re-
vealed that shear failure occurred near the pile heads.
Soil liquefaction in Fukaehama induced large ground
surface settlement up to about 50 cm, exposing the heads
of steel pipe piles of some buildings above ground.
Although some piles appeared to have inclined a little,
and minor damage occurred at the connections between
the pile cap and pile head, the piles themselves appeared
sound. This suggests that steel pipe piles might have per-
formed better than others, because of their ductility.
Ashiyahama Seaside Town, Ashiya city, was reclaimed
from 1969-1975, using sands from both Awaji Island
and the sea of Okayama prefecture. The thickness of the

(c) Settlement due to (d) Tilting due to over-

bearing capacity turning moment
failure

(e) Settlement due to (f) Tilting due to over-

bearing capacity turning moment
failure
Photo. 1. Liquefaction-induced settlement of building in Fukaehama
Fig. 8. Schematic figures showing foundation damage
224 TOKIMATSU ET AL.

fill ranges from 12-16 meters, with the ground surface 5 from 1966-1980 and from 1971-1985, using soils from
meters above sea level. Extensive liquefaction occurred the Rokko Mountains. Figures 9 and 10 show the boring
over large portions of the area. Many wooden private logs at Sites I and A in Port Island, and Sites G and B in
houses in Shiomi-cho and Midori-cho (F in Fig. 7), and Rokko Island. Figure 11 shows the range of the grain size
Hamakaze-cho (G in Fig. 7), experienced severe tilt with distribution curves for the fill materials obtained with an
little damage to the superstructures. These houses were SPT sampler. It should be noted that the actual fill
typically two stories with continuous footings or mat materials often contain gravel particles up to 150 mm,
foundations. A building of Hamakaze Elementary which could not be recovered with the SPT sampler. The
School supported by prestressed concrete piles about 30 locations of the boring sites are shown in Fig. 12. Site A
meters long also leaned considerably. Loss of bearing in Port Island is founded on fills improved by sand
capacity appeared to be the major cause of this damage, drains.
since no damage to piles was observed. At Daito-cho (H The fill materials in Port Island and on northern
in Fig. 7) northeast of Ashiyahama, several residential Rokko Island consist mainly of decomposed granite.
buildings four to five stories high tilted considerably.
In Nishinomiya city, liquefaction-induced damage was
obvious not only in the reclaimed land areas along the
coastline but also in the Muko River delta. Soil liquefac-
tion extended inland about 5 km from the mouth of the
Muko River (I in Fig. 7), where a five-story building
〓uffered ground settlement. In Uedanaka-cho, about 1
km from the mouth of the Muko River (J in Fig. 7),
many wooden houses and several multi-story engineered
buildings experienced severe settlement and/ or tilt. In
Osaka prefecture, east of Hyogo prefecture and not
shown in Fig. 7, soil liquefaction sporadically occurred
along the Yodo River and its tributaries, damaging over
200 wooden private houses. The damaged areas include
Tsukuda, Nishiyodo-ku; and Torishima, Konohana-ku.
Typically, the foundations in these houses consist of
floor post footings. Only a few engineered buildings ex-
perienced foundation problems in Osaka. These included
several multi-story buildings in Torishima, Konohana- (a) (b)
ku, and a three-story building in Tosabori, Nishi-ku.

Port and Rokko Islands

Fig. 10. Typical boring logs on Rokko Island: (a) Site G; (b) Site B
Port and Rokko Islands were reclaimed, respectively,

(a) Port Island

(b) Rokko Island

(a) (b)

Fig. 11. Grain size distribution curves for fill materials at (a) Site A in
Port Island and (b) Site C in Rokko Island (after Takenaka
Fig. 9. Typical boring logs on Port Island: (a) Site I; (b) Site A Corporation, 1995)

DAMAGE TO BUILDING
(a) Port Island
(b) Rokko Island
Fig. 12. Map showing approximate contour lines of ground
Those in southern Rokko Island, by contrast, consist
mainly of crushed tuff and sandstone of the Kobe and
Osaka Groups from the Rokko Mountains. Figure 11
shows that both soils are well-graded and contain a sig-
nificant amount of fines and clay as well as large gravel
particles. The soils of southern Rokko Island, however,
includes more fines and clay, i.e., about 20% and 10%,
compared with about 15% and 5% of the decomposed
granite (Fig. 11(a)). In addition, those fines and clay
show higher plasticity.
The materials in the reclaimed fills for these islands
were often considered to be resistant to liquefaction. Rel-
atively low seismic activities in this region for the last
several decades might have facilitated that judgment.
The consolidation settlements of the underlying Holo-
225
FOUNDATION
settlement in Port and Rokko Islands (in cm)
cene and upper Pleistocene clays after reclamation, on
the other hand, were expected to be over 3 meters. There-
fore, the primary engineering concern in these islands
was the remedial measures to be used for protection
against consolidation settlement.
Unexpectedly, extensive liquefaction occurred in much
of Port Island and on the northern part of Rokko Island,
as shown in Fig. 4. Figure 12 shows approximate con-
tours of liquefaction-induced settlement, estimated
mainly from the vertical gaps between the ground surface
and the base of the building founded on point bearing
piles. The data were based on our reconnaissance surveys
in addition to others (Tanimoto, 1995; Shimizu Corpora-
tion, 1995). Figure 12 indicates that the ground surface in
many of the liquefied areas settled over 30 cm, which is
226
equivalent to that observed in the 1964 Niigata earth-
quake.
Figure 13 shows the time histories of the ground settle-
ments in Port and Rokko Islands, measured with differen-
tial settlement gauges since the middle of 1980s. Site PA
is situated on a road near Site A in Port Island (Fig. 12(a)),
and Site RB near Site B in Rokko Island (Fig. 12(b)).
The accumulated ground settlement at Site RB before the
1995 earthquake is significantly larger than that at Site
PA. This is due simply to the effects of sand drains that
had accelerated the settlement at Site PA by the middle
of 1980s. The earthquake-induced ground settlement of
4.5 cm at Site RB, however, is significantly smaller than
about 24 cm at Site PA. Most of the earthquake-induced
settlements at both sites (2.2 cm and 22 cm) occurred in
the fills. The better performance of the fill materials in
southern Rokko Island is probably due to their higher
plasticity. In any case, the earthquake induced settle-
ments in Fig. 13 are consistent with the values shown in
Fig. 12. Figure 12 therefore appears to outline the
(a) Port Island
(b) Rokko Island
Fig. 13. Time history of ground settlement measured
settlement gauges in Port and Rokko Islands
1995; Mine et al., 1995)
TOKIMATSU ET AL.
Fig. 14. Effects of soil improvement on ground settlement in Port
Island (after Hirai et al., 1995)
ground settlements caused by the earthquake in these is-
lands.
Figure 12 also shows that the settlements in the central
city zones of both islands are considerably smaller than
those in the surrounding industrial zones. This is primari-
ly due to the extensive soil improvements taken in the city
zones and partly due to the higher ground levels, i.e.,
deeper groundwater tables. Previous studies have shown
that the liquefaction susceptibility decreases with increas-
ing groundwater depth.
Figure 14 shows the effects of soil improvement on the
earthquake-induced ground settlements in Port Island,
which have been estimated from the vertical gaps around
the buildings founded on point bearing piles (Hirai et al.,
1995). Although still significant, the settlements of the im-
proved deposits decreased to half, compared with those
of the unimproved ones. Harada et al. (1995) presented a
similar result.
As described earlier, there were few buildings for
which remedial measures were positively taken to
mitigate liquefaction hazards in these islands. Most of
the engineered buildings, nevertheless, had neither tilted
nor settled, with their superstructures remaining intact,
as schematically shown in Fig. 15(a)-(d). In the buildings
founded on point bearing piles, the foundation beams,
pile caps, and even pile heads had projected over the
ground surface, due to large ground settlements. As a
result, large vertical gaps occurred at the bases of many
buildings, as shown in Fig. 15(a) and Photo. 2.
Many low-rise buildings with spread or friction pile
foundations appear to have settled with the ground sur-
face, without appreciable tilting. Small or no vertical
gaps occurred at the bases of these buildings, as shown in
Fig. 15(b)-(c). The floating foundations adopted for
many low-rise buildings as remedial measures against con-
at differential
solidation settlements may have contributed to their
"good performance ." The wooden residential houses, lo-
(after Hirai et al.,
cated on the soil improved by preloading in central
 DAMAGE TO BUILDING FOUNDATION 227

Rokko Island, did not exhibit any serious foundation ticularly concerning lateral forces acting on piles (Recom-
problems. mendations for Design of Building Foundations, Ar-
The performance of these buildings was completely chitectural Institute of Japan, revised in 1974 and 1988;
different from that in the old reclaimed land areas during and Recommendations for Design of Building Founda-
the 1995 earthquake and that during past earthquakes. tions during Earthquakes in 1984); (2) the unexpected
The reasons for this difference are probably due to: (1) positive effects of the foundation designs for consolida-
the revisions of the design codes made in 1970s-80s, par- tion settlement; and (3) unexpectedly high stiffness of the
liquefied fills consisting of decomposed granite. Concern-
ing Item (3), the analysis of the strong motion records
shown in Fig. 5 has indicated that the S-wave velocity
and shear modulus of the liquefied fills are about 50 m/ s
and 5 MPa. These values are, respectively, 1/ 4-1/3 and
1/ 20-1/10 of the unliquefied ones, and are significantly
higher than those of sands liquefied during past earth-
quakes.
(b) Small vertical gap
The lower three rows in Fig. 5 show the time histories
(a) Projection of pile due
to ground settlement at building base of the relative displacements between the depths where
the downhole strong motion records are available. The
relative displacement between the top and the bottom of
the fill exceeded 30 cm which corresponds to a shear
strain of about 2%. There is a possibility that such a
large shear deformation of the fill might have caused the
piles to fail at some depths, as schematically shown in
Fig. 15(d). To assess the integrity of such problematic
(c) Small vertical gap at (d) Damage due to shear defor-
building base mation during earthquake piles, sonic testing based on stress wave theory (Photo. 3)
as well as television observation was performed at vari-
Fig. 15. Schematic figures showing foundation damage due to soil
ous locations (Mizuno and Hirade, 1995; Oh-oka et al.,
liquefaction in Port and Rokko Islands
1996). In Photo. 3, an impact is applied with a hammer
to the pile cap, while the stress wave is monitored at the
pile head. A comparison between the results of television
observation and sonic testing, and their applicability and
limitations will be described elsewhere (Oh-oka et al.,
1996).
Such tests have shown that some of the concrete piles
were damaged near the interface between two layers with
different stiffnesses, e.g., the interface between reclaimed
fill and natural deposit. Of particular interest was the
damage to piles not carrying full design loads, such as
those of buildings under construction. Probably, The
shear deformation of the fill together with the low axial
compressional stress in piles probably caused the piles to
fail in a flexural-tensile mode.

Photo. 2. Large gap induced by liquefaction-induced ground settle-

ment in Port Island Photo. 3. Integrity sonic testing for piles (Courtesy of Mr. M. Sumi)
228 TOKIMATSU ET AL.

steel pipe piles were fixed tightly in the pile caps, the
FOUNDATION DAMAGE DUE TO failure in either the connection or the pile cap preceded.
LIQUEFACTION-INDUCED LATERAL If they were connected loosely to each other, the pile
SPREADING heads rotated, or even became detached from the pile
Many of the quay walls in the reclaimed land areas be- caps and were displaced toward the sea. Sonic tests have
tween Kobe and Amagasaki cities moved several meters suggested that many of those concrete piles cracked and/
towards the sea. The major cause is either the inertia or failed at some depths. The damage to the piles in turn
force acting on them or soil liquefaction of their founda-
tion soils and/or back-fills. This induced lateral spread-
ing of fills which resulted in large horizontal ground
movements as well as differential ground settlements
along a line perpendicular to the shoreline. Such ground
distress sometimes extended inland by more than a hun-
dred meters (Shimizu, 1995; Ishihara, 1995).
Figure 16 shows schematic figures illustrating damage
patterns associated with lateral spreading. The buildings
with pile foundations tilted as their piles were damaged
due mainly to the kinematic forces arising from horizon-
tal ground displacement (Fig. 16(a)-(b)). The inertia
forces from the superstructures, i.e., horizontal forces
and overturning moments, may have contributed to
some extent. The buildings with friction pile foundations
or spread foundations settled and tilted with the adjacent Photo. 4. Damage to ferry terminal structures due to liquefaction-in-
duced lateral spreading
ground surface (Fig. 16(c)-(d)).

Kobe City Including Port and Rokko Islands

On the north shoreline of Rokko Island (H in Fig.
12(b)), a ferry terminal supported on spread foundations
settled and tilted slightly toward the sea. The boarding
bridge collapsed due to large differential settlement
caused by extensive ground subsidence in the backfill just
behind the quay wall (Photo. 4).
The large ground subsidence near the waterfront often
exposed pile heads in many of the warehouses and build-
ings with small penetration depths (Photo. 5). If the con-
crete piles were well embedded in the pile caps, shear or
flexural cracks occurred near the pile heads, often leading
to complete failures of the pile heads (Photo. 6). If the

Photo. 5. Large ground subsidence near the waterfront at the Maya

Wharf district (Courtesy of Mr. Y. Suzuki)

(a) Failure of pile due to (b) Failure of pile due to

lateral spreading lateral spreading

(c) Differential settlement (d) Differential settlement

due to lateral spreading due to lateral spreading

Fig. 16. Schematic figures showing foundation damage due to lateral Photo. 6. Damage to piles due to lateral spreading (Courtesy of Mr.
spreading Y. Suzuki)
 DAMAGE TO BUILDING FOUNDATION 229

caused damage to foundation beams and superstructure,

if the foundation beams were not rigid enough. Other-
wise, the building simply settled and/ or titled with little
damage to the superstructure.
At the foot of the Kobe O-hashi Bridge on Port Island
(I in Fig. 12(a)), a four-story firehouse, supported on
prestressed concrete piles 40 cm in diameter and 30
meters long, moved and tilted towards the sea. An excava-
tion survey of the pile heads shows that compressional
shear failure occurred on the sea side, with minor flexural
cracks on the opposite side.
At the campus of Kobe University of Mercantile Ma-
rine (K in Fig. 7), located on a reclaimed land in
Higaishinada-ku, soil liquefaction induced large lateral
movements of the quay walls. Consequently, a three-sto-
ry building facing the waterfront tilted by 3 degrees and
shifted by at least 30 cm towards the sea. Supporting this (a) (b)
building were prestressed concrete piles 17 meters long,
which penetrated through a reclaimed fill about 9 meters
thick. Television observation detected shear failure at a Fig. 17. Boring logs in Ashiyahama: (a) Shiomi Junior High School;
depth of 4.5 meters on the seaside pile as well as horizon- (b) Waste disposal facilities (after liba and Mizuno, 1995)
tal cracks at a depth of 9 meters on both sides of the piles
(Oh-oka et al., 1996).

Ashiya City
At the southeast corner of Ashiyahama (L in Fig. 7), a
four-story building with basement settled up to 1 meter
and tilted towards the sea by 1/ 40. The quay walls
together with the soil around the building appear to have
slid toward the sea by 1.5 meters. The large settlement
simply resulted from the fact that this building was sup-
ported on a spread foundation. At the northeast corner
of Ashiyahama (M in Fig. 7), two of three 11-story
residential buildings founded on precast concrete piles,
tilted by 1/ 60-1/40 towards the sea due to lateral ground
movements.
Shiomi Junior High School (N in Fig. 7) is situated at
the south end of Ashiya-hama, and faces the shore.
Figure 17(a) shows a boring log at the site. Several
fissures occurred with many sand boils parallel to the
shoreline, as shown in Fig. 18(a). Of particular interest (a) Plan
were the large fissures that run through the north side of
Building B. The quay walls immediately south that shift-
ed towards the sea by 1.5 meters, might have contributed
to the ground distress.
The school buildings supported on prestressed con-
crete piles 30 meters long, particularly the extensions of
Buildings B and C, tilted to the southeast. The passage-
ways connecting school buildings tilted to the south (Fig. (b) Section
18(b)). Consequently, large horizontal openings and verti-
Fig. 18. Displacement of structures and ground fissures at Shiomi
cal gaps occurred at the connections between school Junior High School: (a) Plan (after Tanaka et al., 1995); (b)
buildings and passageways. Section
The horizontal gap between Buildings A and B was the
largest among others, being consistent with the ground
distress. The sum of the horizontal opening amounting tests suggested that the piles had been damaged at depths
to about 1 meter, which is also consistent with the from 6 to 15 meters. Therefore, the school buildings in-
differential ground displacements detected by an aerial cluding pile foundations might have shifted laterally with
the liquefied fills, without showing any significant
photographic survey (Hamada et al., 1995). Although
the excavated pile-heads appeared sound, integrity sonic resistance.
230 TOKIMATSU ET AL.

Waste disposal facilities with a four-story building are

located at the southeast corner of Ashiyahama (0 in Fig.
7). This building (under construction) is supported on
cast-in-place concrete piles 1.0-1.2 meters in diameter
and 30 meters long with enlarged bases. Figure 17(b)
shows a boring log at the site. Ground fissures and settle-
ments of 20-100 cm occurred around the building, with
many sand boils on the north, as shown in Fig. 19.
A land survey showed that the foundation of a chim-
ney (C and D in Fig. 19) had shifted by 53 cm to the
south and by 21 cm to the east. While the direction of the
displacement is consistent with those from an aerial pho-
tographic survey (Hamada et al., 1995), the displacement
itself was considerably smaller.
Although the damage to the superstructure of this
building appeared slight, there was a serious concern that
the fill may have pushed this building towards the sea.
An excavation survey did show that cracks occurred in
some foundation beams and that concrete was spalled
from some pile heads. Four cast-in-place concrete piles

Photo. 7. Damage to pile detected by television observation

were then cored down to depths of 12-16 meters, and tele-

vision observation was made.
Figure 20 shows the result where several large horizon-
tal cracks appeared at a depth of about 10 meters (Photo.
7) as well as near the pile heads (Mizuno and Hirade,
1995). This suggests that the fill down to a depth of about
10 meters might have liquefied and slid towards the sea,
leading to the failure of piles. The smaller displacement
Fig. 19. Map showing horizontal ground displacement and fissures at
of the foundation suggests, however, that the cast-in-
waste disposal facilities in Ashiyahama
place concrete piles might have resisted lateral spreading
to some extent.

EFFECTS OF SOIL IMPROVEMENT ON

Fig. 20. Cracks in piles detected by television observation at waste
disposal facilities in Ashiyahama (after Mizuno and Hirade, 1995)
PERFORMANCE OF BUILDING FOUNDATION
As described earlier, there were few buildings for
which remedial measures were positively taken to
mitigate liquefaction hazards in the affected area. The
few cases include five-story residential buildings at
Koshienhama, Nishinomiya city; an eleven- to 21-story
residential building complex at Ashiyahama, Ashiya city;
and a 14-story building at the pier of American Park,
Chuo-ku, Kobe city. It is noteworthy that, on both Port
and Rokko Islands, the soil improvements which were
not undertaken for liquefaction mitigation, unexpectedly
reduced the extent of soil liquefaction and resulting foun-
dation damage. Such soil improvements include sand
drains and preloading, and sand compaction piles and
rod compaction. The former methods were used for ac-
 DAMAGE
celerating consolidation settlements, and the latter
methods for reducing differential settlements of spread
or friction pile foundations.
Figure 21 shows the effects of soil improvements on the
performance of several buildings on Port Island that had
spread foundations. Their locations are shown in Fig.
12(a). Buildings A-E are founded on soils improved by
either sand drains or pre-loading, but Buildings F-H are
supported on untreated soils. The number outside the
building outline is the ground settlement nearby, and the
fraction inside the building outline is the angle of tilt in
radians. Since small vertical gaps occurred around the
buildings, the absolute settlements of those buildings are
approximately equal to the ground settlement nearby. It
should be noted that the buildings founded on improved
soils exhibited smaller angles of tilt as well as smaller set-
tlements.
Figure 22 shows the effects of aspect ratio on the perfor-
mance of the buildings shown in Fig. 21. The aspect ratio
is the ratio between the height, H, and the square root of
the building area, B x L. There is a tendency in which the
angle of tilt increases with increasing aspect ratio. In addi-
tion, the buildings with individual footings and/ or a
small penetration depth appeared more prone to tilt.
Building A
Building
Buildings
Building C
Building F Building
Building H More than one hundred residential buildings five sto-
Fig. 21. Effects of soil improvements on building
Port Island (after Hirai et al., 1995)
TO BUILDING FOUNDATION 231
Fig. 22. Effects of aspect ratio of building on angle of tilt (after Hirai
et al., 1995)
Probably, the higher the aspect ratio, the larger the over-
turning moment becomes during earthquakes, which in-
creases the contact pressure near the edges of the founda-
tion. This together with liquefaction-related softening of
B the foundation soil might have led to local bearing capaci-
ty failure and resulting differential settlement. It is reason-
able that such local bearing capacity failure tends to oc-
cur in individual footings and/or foundations with small
penetration depths.
On Rokko Island, several buildings exist that are
founded on soils improved with sand compaction piles,
typically 70 cm in diameter and driven at equal intervals
D,E
of 1.8-2.5 meters. Figure 23 shows the performance of
such buildings with spread foundations. Their locations
are shown in Fig. 12(b). Buildings A-C are situated in an
area where signs of soil liquefaction were obscure, while
Buildings D-F in the area where signs of liquefaction
were apparent. The number outside the building outline
is the ground settlement nearby, and the number and frac-
tion inside the building outline are the settlement and an-
gle of tilt. It appears that the installation of sand compac-
tion piles considerably reduced the amount of settlement
G
and the tilt of the buildings. In addition, this effect gets
more pronounced, as the thickness of the untreated fill be-
low the sand compaction piles and/ or the pile spacing
become smaller. Figure 24 shows the SPT N-values mea-
sured before and after the installation of sand compaction
piles at the Building D location. On the average, the SPT
N-values increased by 8 from 12 to 20.
ries high that were constructed in the mid 1960s were lo-
cated at Koshienhama, Nishinomiya city (P in Fig. 7).
These buildings were founded directly on soils improved
by sand compaction piles down to a depth of about 10
meters. While only three of them tilted by up to 1/ 60, the
performance in
remainder survived without any damage.
Over fifty residential steel-frame buildings 14-29 sto-
232 TOKIMATSU ET AL.

Building A Building B

Building C Building D

Building E
Building F (a) (b)

Fig. 25. Boring logs: (a) Ashiya; (b) American Park

Fig. 23. Effects of sand compaction piles on building performance in

Rokko Island

(a) Section

(b) Foundation Plan

Fig. 26. Section and plan of a building founded on piles surrounded

by Deep Cement Mixing wall (after Takenaka Corporation, 1995)

Fig. 24. SPT N-values before and after installation of sand

compaction piles placements in excess of 2 meters occurred on the sea side
and along the Miya River running through the town.
Nevertheless, neither appreciable ground movement nor
ries high which were constructed in the mid 1970s are lo- foundation distress was visible around the buildings. In-
cated in the center of Ashiyahama (Q in Fig. 7) and fac- stead, serious brittle failures occurred in many of the
ing the sea. These buildings were supported on 1100 mm main steel columns of the superstructures.
diam. steel pipe piles about 30 meters long that had been Many sand boils and ground fissures were obvious at
driven through fills improved by sand compaction piles. American Park, Chuo-ku (R in Fig. 7), that faces the
Figure 25(a) shows a boring log before soil improvement. Port of Kobe. Figure 25(b) shows a boring log at the site.
On the average, the SPT N-value increased by 5 after On the pier at the park, a 14-story building under con-
compaction. In the 1995 earthquake, large ground dis- struction was supported on cast-in-place concrete piles
 DAMAGE TO BUILDING FOUNDATION 233

2.5 meters in diameter and 33 meters long. During the underpinning was often required. Some cases exist in
earthquake, the quay walls on the west, south, and east which overturning moments might have contributed to
of the building moved horizontally by 1 m, 2 m, and 0.5- the compression or tension failure, or loss of bearing
0.6 m; and settled by 0.4-0.6 m, 0.5-0.7 m, and 0.2-0.3 capacity of perimeter piles (Fig. 8(b), (d)). This damage
m; respectively. This building, nevertheless, survived pattern again appeared prominent in buildings with high
without damage to its pile foundations, probably due to
the deep cement mixing (DCM) walls that surrounded the
piles, as shown in Fig. 26. The DCM walls were adopted
to withstand lateral forces from both superstructures and
liquefied soils during earthquakes. The performance of
this foundation was better than those of nearby damaged
pile foundations for which no measures were taken to
mitigate liquefaction hazards.

FOUNDATION DAMAGE IN ALLUVIAL FANS

A large number of buildings tilted with their super-
structures remaining intact, even in the areas where
damage to superstructures was concentrated (Photo. 8).
In most of these buildings with pile foundations, the pile
heads were damaged due to lateral forces imposed on the
foundation, as illustrated in Fig. 8(a) and Photo. 9.
Figure 27 shows boring logs at Fukae and Motoyama,
Higashinada-ku (S and T in Fig. 7), where such founda-
tion damage was obvious. A thin soft surface layer 6-10
meters thick that overlies a stiff layer may have accelerat-
ed the shear failure of piles that penetrate through it. The
damage to piles supporting multistory residential build-
ings was so serious that extensive restoration including

Photo. 9. Damage to pile due to horizontal forces from superstruc-

ture

(a) (b)

Fig. 27. Boring logs in heavily damaged area: (a) Fukae; (b)
Photo. 8. Tilt of building in the heavily damaged area Motoyama
234 TOKIMATSU
aspect ratios.
Settlement and tilting of buildings with spread founda-
tions were also evident in the areas with shallow ground-
water and the areas adjacent to fills which were con-
structed over irrigation ponds. This suggests that ground
failures including soil liquefaction may have been in-
volved with the damage (Fig. 8(e)). This damage pattern
also appeared prominent in buildings with high aspect ra-
tios, again suggesting the effects of overturning moment
(Fig. 8(f)).
CONCLUSIONS
A preliminary investigation of the building damage as-
sociated with geotechnical problems in the 1995 Hyo-
goken-Nambu earthquake tentatively leads to the follow-
ing conclusions:
(1) The damage to buildings was caused by either
strong ground shaking or ground problems, each of
〓hich combined with inadequate design of buildings.
addition, the distribution, extent, and pattern were sig-
nificantly affected by local site effects as well as ground
failures including soil liquefaction.
(2) The damage to superstructures was concentrated
in the alluvial fans and floodplains with shallow ground-
water, indicating significant effects of soil conditions
near the ground surface.
(3) Although the occurrence of soil liquefaction had
not been seriously considered, the buildings in the city
zones of Port and Rokko Islands sustained little damage.
This is probably due to (1) ground surface motions at-
tenuated by soil liquefaction and (2) unexpected effects
of the foundation design related to consolidation settle-
ment. The buildings with spread or friction pile founda-
tions appear to have performed well.
(4) Regardless of the type of foundation, many build-
ings settled and/ or tilted with their superstructures
remaining intact. Such damage was obvious not only in
〓he reclaimedlandareasbutalsointhealluvialfansand
floodplains. Possible causes of the foundation damage
dude: (1) horizontal forces and overturning moments
from the superstructures, (2) kinematic forces acting on
deep foundations due to shear deformation of soil
deposits, (3) reduction in bearing capacity due to soil liq-
uefaction, and (4) lateral spreading.
ACKNOWLEDGMENTS
The authors are grateful to Prof. M. Hamada, Waseda
University, for generously providing the results of the ae-
rial photographic survey for the preparation of this
report. Our gratitude is also extended to Mr. Y. Shamo-
to, Shimizu Corporation, and Mr. S. Mori, Tobishima
Corporation, for providing information concerning
ground settlements on Port and Rokko Islands; and to
Mr. Y. Suzuki, Kajima Technical Research Institute, and
Mr. M. Sumi, Geotop Corporation, for providing the
ET AL.
photographs. In addition, the authors extend their sin-
cere gratitude to Messrs. S. Mihara, Y. Taya, K. Tada,
and Y. Asaka, Tokyo Institute of Technology, for long
hours in preparing the figures of this report.
REFERENCES
1) Chuo-Kaihatsu Corporation (1995): Reconnaissance Report on the
1995 Hyogoken-Nambu Earthquake—Great Hanshin Disaster,
(Hyogoken-Nambu earthquake).
2) Committee of Earthq. Obs. and Res. in the Kansai Area (CEOR-
KA) (1995): Digitized Strong Motion Records in the Affected Area
during the 1995 Hyogoken-Nambu earthquake.
3) Hamada, M., Isoyama, R. and Wakamatsu, K. (1995): "The 1995
Hyogoken-Nambu (Kobe) earthquake, Liquefaction, ground dis-
placement and soil condition in Hanshin area," Association for De-
velopment of Earthquake Prediction, 194pp.
4) Harada, K., Yasuda, S. and Shinkawa, N. (1995): "Consideration,
on liquefaction-induced ground settlement and effects of soil im-
provements," Proc., 23rd JSCE Earthq. Engineering Symposium,
pp. 229-232 (in Japanese).
5) Hirai, Y., Kakurai, M., Maruoka, M., Yamashita, K., Aoki, M.
In
and Fukuyama, K. (1995): "An investigation of damage to build-
ing foundation in Kobe Port Island during the Hyogoken-Nambu
earthquake of 1995," Proc., Annual Meeting, AIJ, Vol. B-1, pp.
1143-1144 (in Japanese).
6) Iiba, M. and Mizuno, H. (1995): "Damage of pile foundations dur-
ing 1995 Hyogoken-Nambu earthquake: Part 3-Analytical study
on pile damage due to liquefaction lateral flow," Proc., Annual
Meeting, AIJ, Vol. B-1, pp. 1131-1132 (in Japanese).
7) Ishihara, K. (1995): "Ground displacement and soil characteris-
tics," Proc., Special Meeting on Great Hanshin Disaster,
JSSMFE, pp. 25-28 (in Japanese).
8) Kobe City (1980): Soil Conditions in Kobe (in Japanese).
9) Mine, T., Fukuyama, K., Mukai, H. and Matano, H. (1995): "Set-
tlement measurement of floating foundation in Rokko Island after
1995 Hyogoken-Nambu earthquake," Proc., Annual Meeting,
AIJ, Vol. B-1, pp. 1137-1138 (in Japanese).
10) Mizuno, H. and Hirade, T. (1995): "Damage of Pile foundations
during 1995 Hyogoken-Nambu earthquake: Part 1-Damage pat-
terns and damage cases," Proc., Annual Meeting, AIJ, Vol. B-1,
pp. 1127-1128 (in Japanese).
11) Oh-oka, H. (1995): "Characteristics of liquefying reclaimed
ground and damage features of buildings in the Hyogoken-Nambu
earthquake, Proceedings, Spring Meeting of Building Research In-
stitute," Ministry of Construction, pp. 81-84 (in Japanese).
in- 12) Oh-oka, H., Iiba, M., Abe, A. and Tokimatsu, K. (1996): "Investi-
gation of earthquake-induced damage to pile foundation of using
borehole televiewer observation and integrity sonic tests," Tsuchi-
to-Kiso, Vol. 458 (in Japanese).
13) Shimizu Corporation (1995): Reconnaissance Report on the 1995
Hyogoken-Nambu Earthquake, 226pp. (in Japanese).
14) Takenaka Corporation (1995): Reconnaissance Report on Great
Hanshin Disaster (Hyogoken-Nambu earthquake)-Part 2 (in
Japanese).
15) Tamura, M. (1995): "Damage to pile and spread foundations of
building in the Hyogoken-Nambu earthquake," Proceedings,
Spring Meeting of Building Research Institute, Ministry of Con-
struction, pp. 76-80 (in Japanese).
16) Tanaka, S., Mizuno, H. and Iiba, M. (1995): "Damage of pile
foundations during 1995 Hyogoken-Nambu earthquake: Part 2-
A case due to liquefaction lateral flow and applicability of nondes-
tructive test," Proc., Annual Meeting, AIJ, Vol. B-1, pp. 1129-
1130 (in Japanese).
17) Tanimoto, K. (1995): "Geotechnical hazards in the Hyogoken-
Nambu earthquake," Proc., Special Session, National Confer-
ence, Japanese Geotechnical Society, pp. 1-24 (in Japanese).