Download as pdf or txt
Download as pdf or txt
You are on page 1of 13

Model Tests and Analyses of Bearing Capacity

of Strip Footing on Stiff Ground with Voids


Makoto Kiyosumi1; Osamu Kusakabe, M.ASCE2; and Masatoshi Ohuchi3

Abstract: A series of 1G loading tests under the plane-strain condition were conducted on stiff ground with continuous square voids with the
view of shallow foundation on calcareous sediment rocks, which contain voids because of their susceptibility to water dissolution. Detailed
experimental observation revealed three types of failure modes for a single void: bearing failure without void failure, bearing failure with void
failure, and void failure without bearing failure, depending on the location of the void as well as the size of the void. Upper-bound calculations
were presented to interpret the changes of bearing capacity observed because of the existence of a void. DOI: 10.1061/(ASCE)GT.1943-
5606.0000440. © 2011 American Society of Civil Engineers.
CE Database subject headings: Footings; Limit analysis; Model tests; Voids; Load bearing capacity; Calcareous soils.
Author keywords: Bearing capacity; Calcareous soils; Footings; Limit analysis; Model tests; Voids.

Introduction may be a shallow foundation on stiff ground with voids. There


are two geotechnical issues involved: (1) how to estimate
Calcareous sediment rocks are widely spread throughout tropical the bearing capacity of a footing on stiff ground with voids; and
and subtropical regions. These sediments often pose difficult foun- (2) how to identify the size and location of these voids. This paper
dation design problems because of their unique geotechnical prop- focuses on the former issue.
erties, in particular, the existence of voids of various sizes and There may be two approaches to look into this problem: (1) from
shapes. Voids in the calcareous sediment rock are considered to the view of bearing capacity of footing above the void; and (2) from
be formed by dissolution and are related to the flow direction of the view of stability of voids under surface loading.
underground water. Field observations suggest that voids continu- Along with the view of bearing capacity of footing above the
ously extend horizontally and that isolated voids are very rare. void, the effect of the presence of a single void on the bearing
Therefore two-dimensional (2D) idealizations may be feasible capacity of shallow footings has been studied. Baus and Wang
for modeling purposes. Fig. 1 summarizes the dimensions of voids (1983) performed small-scale model tests of shallow footings on
in Japanese calcareous sediments. These dimensions, obtained a compacted silty clay in the plane-strain condition, in which
from the available literature, show that the size of the voids range the footing was centered above continuous rectangular voids.
from less than 1 m to as large as approximately 10 m, and the shape Al-Tabbaa et al. (1989) presented the results of a small-scale model
of the voids varies from square to rectangular with the longer axis in test of strip footing on a gypsum-sandy soil mixture with a continu-
the horizontal direction. ous circular void. Baus and Wang (1983) carried out 2D finite-
An unconfined compressive strength between 5 and element model (FEM) analyses with an elastic perfectly plastic
100 MN=m2 is reported for Ryukyu calcareous sediment rock in material. Badie and Wang (1984) extended their FEM program
Japan (Shinjo and Nakamura 1975). Considering such high com- to three-dimensional (3D), and Wang and Badie (1985) further
pressive strength, calcareous sediment rocks can be regarded to be extended their investigation to the situations where voids are con-
sufficiently competent as a bearing stratum for shallow founda- tinuous and cubic and the continuous voids are either parallel with
tions. However, because a number of voids exist in the calcareous or perpendicular to the existence of the strip footing. Wang and
sediment rocks, foundation design is not always straightforward. Hsieh (1987) used the upper-bound theorem of limit analysis to
The conventional current design option is a pile foundation pen- analyze the collapse load of strip footing centered above a continu-
etrating into a deeper bearing layer. An economical alternative ous circular void. They considered three failure mechanisms and
presented the results for a special case of a circular void having
1
Technical Division I, ITC Corporation, 1-7-18, Okubo, Shinjuku-ku, a diameter equal to 2.4 times the footing width, comparing the
Tokyo, 169-0072, Japan (corresponding author). E-mail: kiyosumi@ previous results by Badie and Wang (1984) and Baus and Wang
itc-consul.com (1983). Azam et al. (1997) conducted 2D plane-strain FEM analy-
2
Professor, Dept. of Civil Engineering, Tokyo Institute of Technol- ses to study the void effect on the stability of strip footing in two-
ogy, Ookayama 2-12-1, Meguro-ku, Tokyo 152-8552, Japan. E-mail: layer soils where the bottom layer includes a continuous circular
kusakabe@cv.titech.ac.jp void. Wang et al. (2001) extended the work by Wang and Hsieh
3
General Manager, Construction and Technical Development Division, (1987) for a wide range of footing sizes, void sizes, and locations
Oriental Shiraishi Corporation, 2-1-1, Hirakawa-cho, Chiyoda-ku, Tokyo relative to the footing and soil properties, considering 10 failure
102-0093, Japan. E-mail: masatoshi.ouchi@orsc.co.jp
mechanisms. They devised the correction equations and provided
Note. This manuscript was submitted on April 5, 2009; approved on
September 15, 2010; published online on March 15, 2011. Discussion per- figures for determining the influence zone, the critical depth to
iod open until September 1, 2011; separate discussions must be submitted void, and the ultimate bearing capacity of a strip footing underlined
for individual papers. This paper is part of the Journal of Geotechnical and by a circular void. Their results seem relevant for cases in which the
Geoenvironmental Engineering, Vol. 137, No. 4, April 1, 2011. ©ASCE, diameter of voids is larger than footing width. Sreng Sokkheang
ISSN 1090-0241/2011/4-363–375/$25.00. et al. (2002) reported 14 loading tests of a small-scale strip footing

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / APRIL 2011 / 363

Downloaded 11 Jun 2012 to 180.211.192.67. Redistribution subject to ASCE license or copyright. Visit http://www.ascelibrary.org
102 Table 1. Properties of Silica Sand
1:1
Properties Value
Specific gravity 2.645
101
Maximum grain size (mm) 2.00
Height (m) 60% diameter of soil particle (mm) 0.552
50% diameter of soil particle (mm) 0.401
100 30% diameter of soil particle (mm) 0.218
10% diameter of soil particle (mm) 0.078
Coefficient of uniformity 7.074
Coefficient of curvature 1.098
10-1
10-1 100 101 102
Width (m)
with an unconfined compressive strength of 0:1 MN=m2 , whereas
Fig. 1. Dimensions of voids in Japanese calcareous sediments Al-Tabba et al. (1989) adopted a gypsum-sandy soil mixture with
an unconfined compressive strength of 0:1 MN=m2 . The typical
calcareous sediment rocks are much stronger than the gypsum-soil
on a kaolin soil artificially mixed with gypsum for various combi- mixtures by one order of magnitude. A cement-soil mixture was
nations of size and depth of rectangular void, together with FEM used in this experiment.
results. More recently, Kiyosumi et al. (2007) investigated the A number of unconfined compression tests were conducted to
yielding pressure of a strip footing above multiple voids using a achieve the target strength and stiffness after curing for seven days,
2D plane-strain finite-element method analysis (PLAXIS 1998). and the appropriate mixture was determined by trial and error.
The results indicated that the failure zone with the multiple voids Referring to the relationship between the tangent modulus at a
developed mainly toward the void nearest to the footing and the 50% unconfined compressive strength (Et;50 ) and the unconfined
yielding pressure with the multiple voids was equal to the yielding compressive strength (qu ) for the calcareous sediment rocks
pressure of the ground where only the nearest void exists. obtained from the Okinawa area (Shinjo and Nakamura 1975), the
From the view of the stability of the void, the problem has been model ground was made by mixing silica sand, high early strength
extensively studied in the field of tunnel construction. A classical portland cement, additives, and water. The properties of the silica
work by Davis et al. (1980) presented a detailed study on the sand are presented in Table 1. The unconfined compressive strength
stability of a tunnel in the form of stability number (N) using the (qu ) and Et;50 =qu of the model ground was measured. The results
theory of plasticity, together with centrifuge model test results for were qu ¼ 4:8 MN=m2 and Et;50 =qu ¼ 600 after curing for seven
verification. days, which is within the range of the properties of the target
This paper presents results of 1G loading tests of rigid strip foot- calcareous sediment rocks.
ing on stiff ground with square voids. Upper-bound calculations are Three unconsolidated-undrained triaxial compression tests were
also presented to interpret the changes of bearing capacity observed also conducted on the specimens of the model ground under the
because of the existence of a void. cell pressures of 2.5, 6.5, and 12 MN=m2 from which the internal
friction angle (ϕ) and the cohesion (c) were obtained to be 23.6° and
0:95 MN=m2 , respectively.
Model Test Caution is necessary when forming the model ground with
void. The cement-soil mixture was fully mixed for 10 min. A rec-
A series of 1G model tests was performed in a steel container under tangular bar of Styrofoam with a given size was horizontally placed
the plane-strain condition. The steel container had internal dimen- before pouring the mixture into the container for curing. Three
sions of 500 mm high, 700 mm wide, and 120 mm thick. The face specimens for unconfined compressive tests were also prepared
plate of the container was made of a 48-mm-thick acrylic plate using the same batch. Both the model ground and the specimens
with a reinforced steel grid to observe the development of the were cured in the same environment of 20°C and 50% humidity.
failure mechanism. A loading system of 20-ton capacity originally The specimens were subjected to unconfined compression tests
designed for high confining pressure triaxial tests was used, and the after seven days to obtain the reference strength of the model
container was placed directly on a turntable of the loading system. ground.
A rectangular model footing made of steel had a height of 50 mm, After curing the model ground for six days, acetone was intro-
a width of 40 mm (B), and a length of 120 mm. With these arrange- duced to the bar of Styrofoam to be chemically melted away. This
ments, only 10% of the load intensity of the footing propagates to method ensured the minimum disturbance to the surrounding
the base of the container, based on the elastic calculation, and the model ground and is considered to be superior to the previous
Terzaghi’s plastic failure mechanism remains in the container even methods, such as drilling after curing (Al-Tabbaa et al. 1989), of
for ϕ ¼ 40°. The boundary effect is therefore considered to have a removing a mold embedded in the model ground (Baus and Wang
limited influence on the test results. 1983; Badie and Wang 1984; Sreng Sokkheang et al. 2002). The
The model ground was selected to be stiff ground, the mechani- surface of the cured model ground was trimmed to form the model
cal properties of which are similar to those of the typical calcareous ground of 320 mm thick and the surface markers were drawn on the
sediments found in the Okinawa prefecture. Previous experimental side of the model ground before the loading test commenced.
studies of the bearing capacity of footing above void used a com- A thin mortar layer, of which the unconfined compressive
pacted clay soil and a gypsum-soil mixture. Basu and Wang (1983) strength was approximately 22 MN=m2, was glued beneath the
used a compacted silty clay soil. Badie and Wang (1984) used com- base of the model footing to achieve good contact with the model
mercially available kaolinite. Those more clayey soils indicated ground surface, simulating a rough base condition. Horizontal and
an unconfined compressive strength of approximately 0:2 MN=m2. rotational movements of the model footing were allowed by intro-
Sreng Sokkheang et al.(2002) used gypsum-clayey soil mixture ducing an arrangement of a roller and a spherical bearing between

364 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / APRIL 2011

Downloaded 11 Jun 2012 to 180.211.192.67. Redistribution subject to ASCE license or copyright. Visit http://www.ascelibrary.org
Fig. 2. Loading arrangements and specification of void relative to footing

the footing and a load cell, as shown in Fig. 2. Three potensiom- the footing width. The test program is shown in Table 2, along with
eters were arranged to measure the vertical displacement (SV ) and the test code. The present study focuses on the voids that are
the rotation of the footing. A dial gauge was set to measure the smaller than the footing width.
horizontal displacement of the footing (SH ). A load cell mea-
sured the loading pressure. A vertical load was applied with a con-
stant displacement rate of 2:5 mm=h up to the vertical displacement Results and Discussions
of 25% of the footing width. The loading ratio was selected to
complete the loading test within several hours to avoid any sig- Bearing Capacity of Footing on Stiff Ground without
nificant possible aging effect of the cemented material. The Void
development of the failure mechanism was monitored by taking
photographs with a digital camera. A loading test of strip footing on sandy ground typically produces a
Our previous FEM analysis indicated that the bearing capacity failure mechanism consisting of the active wedge beneath the foot-
of a footing above a square void is slightly smaller than that of ing, the radial fan zone, and the passive zone. Kusakabe et al.
circular void which has the cross-sectional area identical to that (1992) observed a similar failure mechanism for the stiff naturally
of a square void, probably owing to stress concentration at the cor- deposited sandy ground. However, the failure mechanisms of foot-
ners of the square void. The shape of the void was therefore ing on stiff ground, such as rock, is somewhat different from those
selected to be square for all the tests. observed in granular ground.
To represent the location of the void relative to the footing and The failure mechanism observed in the test case NV is shown
the size of the void, three notations are introduced, as shown in both from the face and back sides in Fig. 3, together with a photo-
Fig. 2. The symbols α and β indicate the ratio of the vertical dis- graph at SV =B ¼ 0:25. Two clear slip lines develop almost verti-
tance from the ground surface to the center of the void to the footing cally from both edges of the footing without forming an active
width (B) and the ratio of the horizontal distance from the void wedge. A close examination reveals that a few slip lines are begin-
center to the centerline of the model footing to the footing width, ning to develop from the edge of the footing, forming an incipient
respectively. The symbol n indicates the ratio of the void width to fan zone. In the photograph, the whiter areas appeared in the front

Table 2. Test Program


No. Test code Center single void (β ¼ 0) Off-center single void (β ¼ 1) Two voids
1 NV (no-void) — —
2 C1 n ¼ 0:1, α ¼ 0:75 —
3 C2 n ¼ 0:25, α ¼ 0:75 —
4 C3 n ¼ 0:25, α ¼ 1:75 —
5 C4 (load-unload-reload test) n ¼ 0:5, α ¼ 0:75 —
6 C5 n ¼ 0:5, α ¼ 1:75 —
7 E1 — n ¼ 0:25, α ¼ 0:75
8 E2 — n ¼ 0:5, α ¼ 0:75
9 M1 — — Serial configuration
10 M2 — — Parallel configuration

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / APRIL 2011 / 365

Downloaded 11 Jun 2012 to 180.211.192.67. Redistribution subject to ASCE license or copyright. Visit http://www.ascelibrary.org
Fig. 3. Failure mechanism: (a) photograph at SV =B ¼ 0:25 for NV; (b) discontinuity lines from front face for NV; (c) discontinuity lines from back
face for NV; (d) discontinuity lines from front face for C1

face of the container as the result of lateral high pressures exerted Bearing Capacity of Footing on Stiff Ground with
by the footing pressure, pushing the zone on the acrylic plate. Voids
In fact, the whiter areas extended in a stepwise manner toward
Center Single Void
a deeper and wider location as the loading pressure increases.
Fig. 4 shows the normalized load intensity (q=qu )—the normalized
The direction, which is perpendicular to the acrylic plate of the settlement (SV =B) curves. Both axes are respectively normalized
container, is considered the intermediate strain direction because with the unconfined compressive strength (qu ) and the model foot-
the present study modeled the plane-strain condition. The term ing width (B). In Fig. 4(a), the yielding pressure is defined as a
intermediate strain direction is used to describe the direction deflection point on log-log plot of q=qu and SV =B. Fig. 4(b) shows
hereafter. The gap in the intermediate direction did not appear when the normalized load intensity (q=qu )—the normalized settlement
SV =B ¼ 0:25. When dismantling the container after the loading (SV =B) curves to see the effect of void width (n ¼ 0:1, 0.25, 0.5)
test, the gap became visible. Other discontinuity lines observed for the value of α ¼ 0:75 and β ¼ 0. As shown in the figure, the
were tension cracks, typically developing from the edge of the yielding pressure, except for the value α ¼ 1:75, as well as
footing toward the radial direction. In addition, in the failure ultimate bearing capacity defined at SV =B ¼ 0:25 increases with
mechanism for test case C1 as well as test case NV, three types decreasing n value. The effect of void size seems more marked
of discontinuity lines developed. when the n value is more than 0.25. When the value of n becomes
The literature review reveals that the bearing capacity of footing 0.5, the bearing characteristics after yielding pressure are different
(qb ) on stiff ground may be expressed by a ratio of the bearing pres- from those of NV, C1, and C2. Fig. 5 plots the normalized yielding
sure to unconfined compressive strength, such as Tomlinson (1994) pressure and the bearing capacity at SV =B ¼ 0:25 against the value
(qb =qu ¼ 4–12) and Poulos and Davis (1980) (qb =qu ¼ 4–11). of n, which clearly demonstrates that the effect of void size is more
Centrifuge tests of a rock socket pile by Leung et al. (1991) sug- marked when the n value is more than 0.25 at SV =B ¼ 0:25.
gested qb =qu ¼ 2:4–3. Ladanyi and Roy (1971) used a lower- The effect of void depth is shown in Fig. 6, where the compar-
bound solution to compare the in situ loading test on rock. For test isons are made for the value of α being 0.75 and 1.75 for n ¼ 0:25
case NV, the values of qb =qu are 2.5 at SV =B ¼ 0:1 and 3.6 at and 0.5. The increase in value of α results in an increase in yielding
SV =B ¼ 0:25, which are consistent with the centrifuge data and pressure and bearing capacity, and the effect of void depth is more
the lower-bound solution using the triaxial test results. marked for n ¼ 0:5.

(a) (b)
4
Yield point α = 0.75, β = 0
Normalized load intensity, q/qu
Normalized load intensity, q/qu

NV (no-void) NV C2
C4 ( n = 0.5,α = 0.75, β = 0) C1 C4
100 3 NV

2
10-1 C1 (n = 0.1)
C2 (n = 0.25)
CL
1 B

C4 (n = 0.5) nB 0.75B
10-2
0
10-2 10-1 100 101 0 5 10 15 20 25
Normalized settlement, SV /B (%) Normalized settlement, SV /B (%)

Fig. 4. Normalized load intensity (q=qu ); normalized settlement (SV =B) curves: (a) on log-log plot; (b) effect of void width

366 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / APRIL 2011

Downloaded 11 Jun 2012 to 180.211.192.67. Redistribution subject to ASCE license or copyright. Visit http://www.ascelibrary.org
(a) 2.0 (b) 3.5

Normalized yielding pressure, qy/qu


CL

q/qu at SV/B = 0.25


Normalized load intensity,
B
nB 3.0
αB α =1.75
1.5

2.5 α =1.75

1.0 CL
B
0.75 2.0
nB
αB
0.75
0.5 1.5
0.1 0.2 0.3 0.4 0.5 0.1 0.2 0.3 0.4 0.5
Ratio of void width to model footing width, n Ratio of void width to model footing width, n

Fig. 5. Effect of void width: (a) at yielding pressure; (b) at SV =B ¼ 0:25

4 Vector: ratio of total incremental displacement to incremental SV, scale: 1


CL Yield point
Slip line Gap in intermediate strain direction
Normalized load intensity, q/qu

B C2 C3
C3 ( n = 0.25, α =1.75) C4 C5 Tension crack Pushing zone on acrylic plate
nB
αB
3 C2 ( n = 0.25, α = 0.75)
Incremental SV = 9.2 mm Incremental SV = 9.2 mm
for SV /B = 0.02 ~ 0.25 for SV /B = 0.02 ~ 0.25
Footing (B = 40 mm) Footing

2 0
C5 (n = 0.5, α =1.75)

1 Depth (mm)
C4 ( n = 0.5, α = 0.75)
B

0
0 5 10 15 20 25
Normalized settlement, SV /B (%) 2B

Fig. 6. Effect of void depth 0 B 0 B


Horizontal distance (mm) Horizontal distance (mm)
(a) C2 : n = 0.25, = 0.75 (b) C3 : n = 0.25, = 1.75
Fig. 7 shows the failure mechanism for C2, C3, C4, and C5.
The shaded areas indicate the whitish areas that appeared in the Incremental SV = 9.6 mm Incremental SV = 9.2 mm
for SV /B = 0.01~0.25 for SV /B = 2 ∼ 25 %
acrylic plate. Because the footing was allowed to rotate and move Footing Footing

horizontally, the shaded areas do not appear symmetrically about 0


the footing centerline. Failure mechanisms among five tests (C1
through C5) may be classified into three groups. (1) The cases
Depth (mm)

of small and shallow void (C1 and C2) are bearing failure without B
void failure. The two slip lines vertically develop from both edges
of the footing and never extend to the upper corners of the void,
which is similar to the case of the NV failure mechanism. The
shaded zone extends even beneath the void. (2) The case of large 2B
and shallow void (C4) is bearing failure with roof failure. The two
slip lines extend from the edges of the footing to the upper corners 0 B 0 B
Horizontal distance (mm) Horizontal distance (mm)
of the void, and only the roof of the void fails. (3) The cases of (c) C4 : n = 0.5, = 0.75 (d) C5 : n = 0.5, = 1.75
small and deep void (C3) and large and deep void (C5) are void
failure without bearing failure. For C3, no clear slip line appeared Fig. 7. Four different failure mechanisms: (a) C2; (b) C3; (c) C4;
toward the upper corner of the void. By contrast, in C5, slip (d) C5
lines develop from the upper corners of the void and never
extend up to the footing base or edges, and only the roof of the
void fails. Off-Center Single Void
After scrutiny of the visual data on the development of the fail- For the off-center single void, our FEM analysis suggested that the
ure mechanism, Fig. 8 was obtained about the information of when yielding pressure increases with increased distance between the
each slip line had been formed in terms of normalized footing set- center of the void and the footing and that the failure mechanism
tlement (SV =B). There are three types: (1) slip lines only develop develops from both edges of the footing toward the upper right and
downward from both edges of the footing; (2) slip lines develop lower left corners of the void. In this series of tests, two cases of off-
downward from the footing edge as well as upward from the upper center single void tests were carried out (E1 and E2), which may be
corners of the void; and (3) slip lines only propagate upward from compared with C2 and C4, respectively, which has the single void
the upper corners of the void. with the same size on the center line.

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / APRIL 2011 / 367

Downloaded 11 Jun 2012 to 180.211.192.67. Redistribution subject to ASCE license or copyright. Visit http://www.ascelibrary.org
Footing Footing Footing

SV /B = 0.13 ~ 0.25 0.05 ~ 0.25 SV /B = 0.06 ~ 0.15 0.05 ~ 0.06

0.11~ 0.15 0.04 ~ 0.06

SV /B = 0.11~ 0.16 0.11~ 0.18

(a) C2 : n = 0.25, = 0.75 (b) C3 : n = 0.25, = 1.75 (c) C4 : n = 0.5, = 0.75

Fig. 8. Propagation of slip lines: (a) C2; (b) C3; (c) C4

4 toward the edge of the right footing. The void collapses with roof
CL α = 0.75 Yield point
and sidewall failure.
Normalized load intensity, q/qu

B C2 E1
0.75B nB C4 E2 Associated with the formation of slip lines toward the off-center
3 βB void, the footing tends to move toward the direction of the void.
E1 (n = 0.25, β =1) Fig. 11 shows the normalized load intensity (q=qu ) versus the
horizontal displacement of the footing. Probably because of the
2 characteristics of the turntable of the loading apparatus, initially
E2 (n = 0.5, β = 1)
even the case of NV moved horizontally by 1% of SH =B, then
C2 ( n = 0.25, β = 0) in the cases of off-center void (E1 and E2), the footing suddenly
1 C4 ( n = 0.5, β = 0) sways toward the direction of void, This is associated with the
tilting of the footing up to 0.3°.
0 Two Voids
0 5 10 15 20 25
Normalized settlement, SV /B (%) In this series of experiments, two configurations were examined:
serial and parallel configuration. Normalized load intensity
Fig. 9. Normalized load intensity (q=qu ); normalized settlement (q=qu )—settlement (SV =B) curves for the two cases are plotted
(SV =B) for off-center void in Fig. 12, together with C4. The first obvious observation is that
the curves for C4 and M1 are very close to each other, supporting
our FEM results, although the bearing capacity of M1 is slightly
Load intensity settlement curves are shown in Fig. 9. The loca- smaller initially than that of C4 and the difference between the
tion of the void has a significant influence on bearing capacity, two curves is gradually increasing until the end of the loading.
particularly the larger void case by comparing C4 and E2. The bear- On the contrary, the behavior of M2 is different after the yielding
ing capacity increases by 70% when the void is off-center by the point. The load intensity starts dropping at around 4% of SV =B until
footing width. Fig. 10 presents the failure mechanism of E1 and E2. about 10% of SV =B and then gently recovers.
The slip lines initially develop almost vertically from both edges of The failure mechanism observed explains this behavior. Fig. 13
the footing without aiming at any corner of the void, and then the compares the failure mechanism of C4 and M1, which are of bear-
slip line on the right extends to the lower left corner of the void. ing capacity problems and exhibit a similar roof failure mechanism.
Another slip line is seen from the upper right corner of the void After SV =B ¼ 0:14, the visual data indicate that the tension cracks

Vector: ratio of total incremental displacement to incremental SV, scale: 1


Incremental SV = 9.2 mm Incremental SV = 9.2 mm
for SV /B = 0.02 ~ 0.25 for SV /B = 0.02 ~ 0.25
Footing (B = 40 mm) Footing
0

B
Depth (mm)

Slip line
Tension crack
Gap in
2B intermediate strain direction
Pushing zone on acrylic plate

3B
0 B 0 B
Horizontal distance (mm) Horizontal distance (mm)
(a) E1: n = 0.25, = 0.75 (b) E2 : n = 0.5, = 0.75

Fig. 10. Failure mechanisms for off-center void: (a) E1; (b) E2

368 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / APRIL 2011

Downloaded 11 Jun 2012 to 180.211.192.67. Redistribution subject to ASCE license or copyright. Visit http://www.ascelibrary.org
4
NV E1 (n = 0.25, β =1) Yield point the sidewall failure mode, together with the extension of another

Normalized load intensity, q/qu


NV discontinuity line from the right edge of the footing toward the
E1
3 E2 upper right corner of the off-center void. The development of a
E2 ( n = 0.5, β = 1)
number of slip lines leads to the mobilization of the shear strength,
resulting in a slight increase in load intensity.
2
Closure of Void during Loading
CL
- + When the soil is incompressible, footing settlement leads to the
1
closure of a void, provided that there is no heave at the ground sur-
face. The measurement confirmed that there was no vertical heave
0
during loading. Taking the volume balance, we have a simple equa-
-1 0 1 2 3 4 tion for the condition of full closure of a void as BSV ¼ ðnBÞ2 , from
Normalized horizontal displacement, SH/B (%) which the ratio of current cross-sectional area to the initial cross-
sectional area RA can be expressed as
Fig. 11. Normalized load intensity (q=qu ) versus horizontal displace-
ment of footing
RA ¼ ðA0  SV BÞ=A0 ð1Þ

where B = footing width;, SV = footing settlement; and A0 = initial


4
Yield point CL
cross-sectional area. Putting A0 ¼ ðnBÞ2 into Eq. (1) yields the fol-
Normalized load intensity, q/qu

C4 B lowing expression for the ratio of the void width to the footing
M1
M2 CL 0.75B
0.5B
CL width:
3
B 1.75B B
0.5B
0.75B
0.5B
0.75B
0.5B 0.5B RA ¼ 1  ðSV =BÞ=n2 ð2Þ
1B

2 C4 M1 M2
When SV =B ¼ 0:1 is at the ultimate conditions, then the value
(Serial configuration) (Parallel configuration)
of n is about 0.3, suggesting that a void larger than 0.3 times the
footing width never completely closes. In contrast, a smaller void
1
that is 10% of the footing width will close at a settlement of 1%
of the footing width. Experimental results shown in Fig. 4
0 suggest that the value of SV =B is about 2% of the footing width
0 5 10 15 20 25 at yielding. In other words, the void of that size will completely
Normalized settlement, SV /B (%) close before yielding, implying that the voids smaller than 10%
of footing width may not cause a significant drop in the bearing
Fig. 12. Normalized load intensity (q=qu ); normalized settlement capacity of footing.
(SV =B) for two voids Fig. 14 shows the change of the measured cross-sectional area of
void normalized by its original area plotted against the normalized
footing settlement with respect to the void size, the void depth,
start developing above and beneath the lower void, which may be the eccentricity, and the multiple voids. Fig. 14(a) shows that the
one of the causes of the difference between the two curves after void size considerably affects the change of void cross-sectional
yielding. The failure mechanism of M2 is rather complicated. Slip area. At SV =B ¼ 0:1, the void cross-sectional area for n ¼ 0:1 re-
lines develop at SV =B ¼ 0:04 in the central column between the duces by up to 70%, whereas it reduces by only 30% for n ¼ 0:5.
voids, connecting the slip lines between the upper-right corner When taking the value of SV =B ¼ 0:01, the value of the calculated
of the center void and the lower-left corner of the off-center void. cross-sectional area from Eq. (2) is 0 for n ¼ 0:1 and 0.96 for
Further settlement results in developing the roof failure mode and n ¼ 0:5. The experimental results also suggest that the value of

Vector: ratio of total incremental displacement to incremental SV, scale: 1


Incremental SV = 9.6 mm Incremental SV = 9.2 mm Incremental SV = 9.6 mm
for SV /B = 0.01 ~ 0.25 for SV /B = 0.02 ~ 0.25 for SV /B = 0.01 ~ 0.25
Footing (B = 40 mm) Footing Footing
0
Depth (mm)

B
Slip line
Tension crack
Gap in
2B intermediate strain direction
Pushing zone on acrylic plate
0 B 0 B 0 B
Horizontal distance (mm) Horizontal distance (mm) Horizontal distance (mm)
(a) C4 : n = 0.5, = 0.75 (b) M1: Serial Configuration (c) M2 : Parallel Configuration

Fig. 13. Failure mechanisms for two voids: (a) C4; (b) M1; (c) M2

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / APRIL 2011 / 369

Downloaded 11 Jun 2012 to 180.211.192.67. Redistribution subject to ASCE license or copyright. Visit http://www.ascelibrary.org
(a) 1.0 (b) 1.0
CL

Change of cross sectional area

Change of cross sectional area


B
nB CL
0.8 0.75B 0.8
B

0.6 C1 (n = 0.1) 0.6 1.75B


nB
C2 (n = 0.25) C4 (n = 0.5) C5 (n = 0.5)

0.4 0.4

0.2 0.2 C3 (n = 0.25)

0 0
0 5 10 15 20 25 0 5 10 15 20 25
Normalized settlement, SV /B (%) Normalized settlement, SV /B (%)

(c) 1.0 (d) 1.0


Change of cross sectional area

Change of cross sectional area


M2 (Void II)
C
0.8 0.8 L M1 (Void I)
C2 ( = 0) E1 ( = 1) B
0.5B 0.75B C5 (n = 0.5, = 1.75)
I
0.6 0.6 1.75B
0.5B
II

0.4 0.4
CL
B

0.2 0.75B 0.25B 0.2


βB
C4 (n = 0.5, = 0.75)
0 0
0 5 10 15 20 25 0 5 10 15 20 25
Normalized settlement, SV /B (%) Normalized settlement, SV /B (%)

Fig. 14. Reduction of cross-sectional area with settlement (SV =B): (a) effect of void size; (b) effect of void depth; (c) effect of void eccentricity;
(d) two voids in serial configuration

the cross-sectional area increases as n increases. It is less marked by the surface loading throughout the loading process and the
for deeper voids, as shown in Fig. 14(b). When α ¼ 1:75, there is change of cross-sectional area considerably differs from the corre-
virtually no void size effect on the change of void cross-sectional sponding single void case (C5). The trend of the cross-sectional
area. The void eccentricity also shows a significant influence on area reduction of the upper void I is very similar to the correspond-
the change of the cross-sectional area, as shown in Fig. 14(c). ing single void case (C4), which is consistent with the observation
Fig. 14(d) confirms that the lower void II is not greatly affected of the failure mechanism.
CL CL CL

V0 V0 V0

B B B
a b a b a b
VI VI VI
ζB φ I φ ζB φ I φ ζB φ I φ
η B(1 + 2ς tan φ ) η B (1 + 2ς tan φ ) η B(1 + 2ς tan φ )
d e d e d e
c θ1 θ2 θ3 θ4 c θ1 θ2 θ3 θ4 c θ1 θ2 θ3 θ4
VII VIII VII VIII VII VIII
φ φ φ φ φ φ
IV IV IV
αB II
VIV
III αB II
VIV
III αB II
VIV
III

θ5 θ5 g
θ5
g h
f g θ6 f i
θ6 θ7
nB VV V nB VI VVI
ρ nB
V φ φ
VV ξ ρ
f
θ7
φ
ξ

βB βB h j
βB k

(a) Roof failure (b) Side wall failure (c) Combined failure

Fig. 15. Three upper-bound mechanisms: (a) roof failure; (b) sidewall failure; (c) combined failure

370 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / APRIL 2011

Downloaded 11 Jun 2012 to 180.211.192.67. Redistribution subject to ASCE license or copyright. Visit http://www.ascelibrary.org
Upper-Bound Mechanism Analysis developed, i.e., the rigid block V is pushed into the sidewall of
the void. The combined failure mode is a combination of the roof
An upper-bound mechanism for the admissible velocity field for a failure mode and the sidewall failure mode. However, we did not
single void may be deduced on the basis of experimental observa- observe the combined failure mode in the experiment.
tions. Where the void size is smaller than the footing width, three The geometrical parameters used for minimization ðζ; η; ξ; ρÞ
types of failure mode were considered, depending upon the location are also shown in Fig. 15. Detailed derivations and the final expres-
of void relative to the footing: (1) roof failure mode, (2) sidewall sions are given in the Appendix. To obtain ultimate bearing capac-
failure mode, and (3) combined failure mode, which are illustrated ity (qub ) value, the parameters minimizing the qub value ðζ; η; ξ; ρÞ
in Fig. 15. varied incrementally to search for the minimum values. Each
For center void, except the cases of C1 and C2, the slip lines parameter varied by an increment of 0.005. Among the calculated
develop from both edges of the footing to the upper edges of the values, the minimum value was selected to be the qub value.
void, as shown in Fig. 7. The zone formed by the slip lines is Fig. 16 compares the critical upper-bound mechanisms with the
pushed into the roof of the void together with the settlement of experimental observations. The critical upper-bound mechanisms
the footing. The admissible velocity field of the roof failure mode were obtained on the basis of the soil parameters deduced from
refers to the observation that the rigid block IV moves into the roof UU test results. Although the agreement between the experiment
of the void. For the off-center cases, as seen in Fig. 10, the zone and the upper-bound solutions needs to be further improved, the
formed by the slip lines is pushed toward the sidewall of the plasticity solution provides the general trend of the effect of the
void. The admissible velocity of the sidewall failure mode is also void and may be used for parametric study.

Vector: ratio of total incremental displacement to incremental SV, scale: 1


Incremental SV = 9.6 mm Incremental SV = 9.2 mm Incremental SV = 9.2 mm
for SV /B = 0.01 ~ 0.25 for SV /B = 0.02 ~ 0.25 for S V /B = 0.02 ~ 0.25
Footing (B = 40 mm) Footing Footing
0
Depth (mm)

Slip line
2B Admissible velocity field

0 B 0 B 0 B
Horizontal distance (mm) Horizontal distance (mm) Horizontal distance (mm)
(a) Roof failure and C4 (b) Combined failure and C5 (c) Side wall failure and E2

Fig. 16. Comparison between critical upper-bound mechanisms and experimental observations: (a) roof failure and C4; (b) combined failure and C5;
(c) sidewall failure and E2

(a) (b)
6.0 6.0
Upper bound (qub/qu) and model test (q/qu at SV/B = 0.25)

Upper bound (qub/qu) and model test (q/qu at SV /B = 0.25)

CL
B
n = 0.25 0.25 n = 0.25 0.5 0.25 CL
nB B
0.75B nB
0.75B βB
5.0 5.0
0.5
0.5
4.0 4.0
0.5

3.0 3.0
Model test (SV /B = 0.25)
n = 0.25 0.5

2.0 2.0 Model test (SV /B = 0.25)


Roof failure n = 0.25 0.5
Combined failure
Roof failure Side wall failure
Combined failure
1.0 1.0
0.7 0.9 1.1 1.3 1.5 1.7 0 0.2 0.4 0.6 0.8 1.0
Ratio of void depth to model footing width, Ratio of void eccentricity to model footing width,

Fig. 17. Comparison of effect of void between experimental and upper-bound calculation: (a) center void; (b) off-center void

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / APRIL 2011 / 371

Downloaded 11 Jun 2012 to 180.211.192.67. Redistribution subject to ASCE license or copyright. Visit http://www.ascelibrary.org
(a) (b)
1.00 1.00

Normalized upper bound, qub/qub,0

Normalized upper bound, qub/qub,0


CL C
L
0.95 B 0.95 B
0.5B 0.75B 0.5B
0.75B
1B

Roof failure
0.90 0.90
(B0 = 0.04 m, qub,0 = 5.51 MN/m2)
Combined failure Side wall failure
(B0 = 0.04 m, qub,0 = 5.14 MN/m2) (B0 = 0.04 m, qub,0 = 15.39 MN/m2)

0.85 0.85
10-1 100 101 102 103 10-1 100 101 10 2 103
Normalized footing width, B/B0 Normalized footing width, B/B0

Fig. 18. Relationship between normalized upper-bound (qub =qub;0 ) and normalized footing width (B=B0 ): (a) roof failure mode and combined
failure mode; (b) sidewall failure mode

Fig. 17 compares the ultimate bearing capacity (qub ) calculated of slip lines: (1) slip lines only develop downward from both
by three admissible velocity fields with the test results at edges of the footing; (2) slip lines develop downward from the
SV =B ¼ 0:25. The curves are envelopes of the minimum values footing edge as well as upward from the upper corners of the
among the three failure modes. As is seen in Fig. 17(a), the com- voids; and (3) slip lines only start from the upper corners of
bined failure mode gives the critical failure mode in the range of the void, propagating upward.
α ¼ 0:7–1:8. For the center-void cases, the combined failure mode 2. Three upper-bound mechanisms for a single void were
is the closest to the test result. For α ¼ 0:75, the values by the com- deduced from the experiments: roof failure, sidewall failure,
bined failure mode generally agree with the test results both for and combined failure, and the upper-bound solutions of bear-
n ¼ 0:25 and n ¼ 0:5. However, the discrepancy widens when ing capacity for strip footing were respectively derived and the
the void becomes deeper, which implies that the combined failure parametric study was presented for a wide range of void sizes,
mode is not applicable to the cases where α is larger than 0.75. depths, and locations.
A similar statement can be made for the roof failure mode. For 3. If two voids of serial configuration are at relatively shallow
the off-center void cases, shown in Fig. 17(b), the critical-failure depth, the lower void had virtually no influence on the failure
mode changes from the combined to the sidewall-failure mode. mode and consequently had very limited effect on the reduc-
The comparison between the test results and the calculation sug- tion of bearing capacity.
gests that the upper-bound calculations give larger bearing capacity
than the test results when increasing the value of β for off-
center void. Appendix
In some cases, the reasons for the large discrepancy between the
test results and the results of limit analysis may occur partly be- The ultimate bearing capacity (qub ) for the upper-bound solution is
cause of a possible side effect and the nonhomogeneity of the defined by the total rate of energy dissipation (D) and the total rate
model ground and partly because of the simply admissible velocity of work done (W) as
fields consisting of rigid blocks. Better solutions may be obtained qub ¼ ðD  WÞ=V 0 B
by adopting curved slip lines with shearing zones.
Fig. 18 presents the upper-bound calculation results that show The total rate of energy dissipation (D) and the total rate of work
the effect of soil weight on the stability of the void. The figure done (W) for the roof failure mode are
shows that the bearing capacity generally decreases with increasing
footing width. But up to the footing width of 4 m, i.e., 100 times D ¼ cV 0 fcos ϕf2lac þ ½lcf = sinðθ1  ϕÞ þ ½leg = sinðθ4  ϕÞg
larger than the model footing width, the decrease is negligible. For
þ lcd j1= tanðθ1  ϕÞj þ lde j1= tanðθ4  ϕÞj
larger footing, such as the footing width of 40 m, i.e., 1,000 times
larger than the model footing width, the bearing capacity may drop þ ½ldf = sinðθ1  ϕÞj sinðθ1 þ θ2  ϕÞ tanðθ5  θ2 Þ
by 5 to 10%.  cosðθ1 þ θ2  ϕÞj
þ ½ldg = sinðθ4  ϕÞj sinðθ3 þ θ4  ϕÞ tanðθ5  θ3 Þ
Concluding Remarks  cosðθ3 þ θ4  ϕÞjg ð3Þ
A series of 1G loading tests under the plane-strain condition were
conducted on a stiff ground with continuous square voids with W ¼ γV 0 fAI þ AII þ AIII þ AIV sin½ðθ1 þ θ2  ϕÞ cos θ5 =
the view of shallow foundation on calcareous sediment rocks con-
taining voids, the size of which is smaller than footing width. The ½sinðθ1  ϕÞ cosðθ5  θ2 Þg ð4Þ
following observations were made from this study:
1. There are three types of failure modes for a single void: where c and ϕ = strength parameters; γ = unit weight of the soil;
(1) bearing failure without void failure; (2) bearing failure with B = footing width; lac ∼ ldg = side lengths of the various zones;
void failure; and (3) void failure without bearing failure, θ1∼4 = angles of zones; θ5 = inclination of VIV ; and AI ∼ IV = areas
depending on the location of the void as well as the size of of zones. The expressions for the side lengths of the various zones
the void. Correspondingly, there are three types of propagation are obtained from

372 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / APRIL 2011

Downloaded 11 Jun 2012 to 180.211.192.67. Redistribution subject to ASCE license or copyright. Visit http://www.ascelibrary.org
lac ¼ ςB= cos ϕ; lcd ¼ ηBð2ς tan ϕ þ 1Þ; lde ¼ Bð1  ηÞð2ς tan ϕ þ 1Þ
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
lcf ¼ B ½β þ ς tan ϕ þ ð1=2Þð1  nÞ2 þ ðα  n=2  ςÞ2
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ldf ¼ B ½β þ ς tan ϕð1  2ηÞ  η þ ð1=2Þð1  nÞ2 þ ðα  n=2  ςÞ2
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
leg ¼ B ½β  ς tan ϕ  ð1=2Þð1  nÞ2 þ ðα  n=2  ςÞ2
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ldg ¼ B ½β þ ς tan ϕð1  2ηÞ  η þ ð1=2Þð1 þ nÞ2 þ ðα  n=2  ςÞ2 ð5Þ

The expressions for angles of zones and the inclination of VIV are The expression for the areas of zones are calculated by
written by
θ1 ¼ cos1 fB½β þ ς tan ϕ þ ð1=2Þð1  nÞ=lcf g > ϕ
AI ¼ ςB2 ðς tan ϕ þ 1Þ;
θ2 ¼ cos1 fB½β  ς tan ϕð1  2ηÞ þ η  ð1=2Þð1  nÞ=ldf g
AII ¼ ηB2 ð2α  2ς  nÞð2ς tan ϕ þ 1Þ=4
θ3 ¼ cos1 fB½β þ ς tan ϕð1  2ηÞ  η þ ð1=2Þð1 þ nÞ=ldg g ð8Þ
AIII ¼ B2 ð1  ηÞð2α  2ς  nÞð2ς tan ϕ þ 1Þ=4;
θ4 ¼ cos1 fB½β þ ς tan ϕ þ ð1=2Þð1  nÞ=leg g > ϕ ð6Þ
AIV ¼ nB2 ð2α  2ς  nÞ=4
θ5 ¼ tan1 ½sinðθ 1 þ θ2  ϕÞ sinðθ4  ϕÞ cos θ3
 sinðθ3 þ θ4  ϕÞ sinðθ1  ϕÞ cos θ2 = sinðθ1 þ θ2
 ϕÞ sinðθ4  ϕÞ sin θ3 The total rate of energy dissipation (D) and the total rate of work
done (W) for the sidewall failure mode are written by
þ sinðθ3 þ θ4  ϕÞ sinðθ1  ϕÞ sin θ2  ð7Þ

D ¼ cV 0 ðcos ϕf2lac þ ½lcf = sinðθ1  ϕÞ þ ½leg = sinðθ4  ϕÞ


þ lf h sinðθ1 þ θ2  ϕÞ cosðθ6  θ5  θ3 Þ= sinðθ1  ϕÞ sinðθ7  ϕÞ cosðθ5  θ2 Þg þ lcd j1= tanðθ1  ϕÞj þ lde j1= tanðθ4  ϕÞj
þ ldf = sinðθ1  ϕÞj sinðθ1 þ θ2  ϕÞ tanðθ5  θ2 Þ  cosðθ1 þ θ2  ϕÞj þ ldg = sinðθ4  ϕÞj sinðθ3 þ θ4  ϕÞ tanðθ5  θ3 Þ
 cosðθ3 þ θ4  ϕÞj þ lf g sinðθ1 þ θ2  ϕÞ= sinðθ1  ϕÞ cosðθ5  θ2 Þj  sinðθ6  θ5  θ3 Þ  cosðθ6  θ5  θ3 Þ= tanðθ7  ϕÞjÞ ð9Þ

W ¼ γV 0 fAI þ AII þ AIII þ sinðθ1 þ θ2  ϕÞ= sinðθ1  ϕÞ cosðθ5  θ2 Þ


× ½AIV cos θ5 þ AV cosðθ6  θ5  θ3 Þ cosðξ þ ϕÞ= sinðθ7  ϕÞg ð10Þ

where
lac ¼ ςB= cos ϕ; lcd ¼ ηBð2ς tan ϕ þ 1Þ; lde ¼ Bð1  ηÞð2ς tan ϕ þ 1Þ
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
lcf ¼ B fβ  n½ð1=2Þ þ tan ξ tan ρ= tan ξ þ tan ρÞ þ ς tan ϕ þ ð1=2Þ2 þ fα þ n½ð1=2Þ  tan ρ=ðtan ξ þ tan ρÞ  ςg2
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ldf ¼ B fβ  n½ð1=2Þ þ ½tan ξ tan ρ=ðtan ξ þ tan ρÞ þ ς tan ϕð1  2ηÞ  ½η þ ð1=2Þg2 þ fα þ n½ð1=2Þ  tan ρ=ðtan ξ þ tan ρÞ  ςg2
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ldg ¼ B ½β þ ς tan ϕð1  2ηÞ  η  ð1=2Þðn  1Þ2 þ ðα þ n=2  ςÞ2 leg ¼ B ½β  ς tan ϕ  ð1=2Þðn þ 1Þ2 þ ðα þ n=2  ςÞ2
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
lf g ¼ nB ½tan ξ tan ρ=ðtan ξ þ tan ρÞ2 þ ½tan ρ=ðtan ξ þ tan ρÞ  12 lf h ¼ nB tan ρ=ðnB tan ρ=ðtan ξ þ tan ρÞ tan2 ξ þ 1 ð11Þ

θ1 ¼ cos1 ½Bðfβ  n½ð1=2Þ þ tan ξ tan ρ=ðtan ξ þ tan ρÞ þ ½ς tan ϕ þ ð1=2Þ=lcf gÞ > ϕ
θ2 ¼ cos1 ðBfβ þ n½ð1=2Þ þ tan ξ tan ρ=ðtan ξ þ tan ρÞ þ ς tan ϕð2η  1Þ þ ½η  ð1=2Þ=ldf gÞ
θ3 ¼ cos1 fB½β þ ς tan ϕð1  2ηÞ  η þ ð1=2Þð1  nÞ=ldg g
θ4 ¼ cos1 fB½β þ ς tan ϕ þ ð1=2Þð1 þ nÞ=leg g > ϕ θ6 ¼ 90 þ θ3  ρ θ7 ¼ ð180  ρ  ξÞ > ϕ ð12Þ

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / APRIL 2011 / 373

Downloaded 11 Jun 2012 to 180.211.192.67. Redistribution subject to ASCE license or copyright. Visit http://www.ascelibrary.org
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
AII ¼ sII ðsII  lcd ÞðsII  lcf ÞðsII  ldf Þ; sII ¼ ðlcd þ lcf þ ldf Þ=2
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
AIII ¼ sIII ðsIII  lde ÞðsIII  leg ÞðsIII  ldg Þ; sIII ¼ ðlde þ leg þ ldg Þ=2
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
AIV ¼ sIV ðsIV  ldf ÞðsIV  ldg ÞðsIV  lf g Þ; sIV ¼ ðldf þ ldg þ lf g Þ=2
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
AV ¼ sV ðsV  lf g ÞðsV  lf h ÞðsV  n · BÞ; sV ¼ ðlf g þ lf h þ n · BÞ=2Þ ð13Þ

The total rate of energy dissipation (D) and the total rate of work done (W) for the combined failure mode are written by

D ¼ cV 0 fcos ϕ½2lac þ lcf = sinðθ1  ϕÞ þ lei = sinðθ4  ϕÞ þ sinðθ1 þ θ2  ϕÞ cos θ5 = sinðθ1  ϕÞ cosðθ5  θ2 Þ½lf j = sinðθ6  ϕÞ
þ ½lik = sinðθ7  ϕÞ þ lcd =j tanðθ1  ϕÞj þ lde =j tanðθ4  ϕÞj þ ldf = sinðθ1  ϕÞj sinðθ1 þ θ2  ϕÞ tanðθ5  θ2 Þ  cosðθ1 þ θ2  ϕÞj
þ ldi = sinðθ4  ϕÞj sinðθ3 þ θ4  ϕÞ tanðθ5  θ3 Þ  cosðθ3 þ θ4  ϕÞj
þ ½sinðθ1 þ θ2  ϕÞ= sinðθ1  ϕÞ cosðθ5  θ2 Þ½lf g j sin θ5  cos θ5 = tanðθ6  ϕÞj þ lhi j sin θ5 þ ½cos θ5 = tanðθ7  ϕÞjg ð14Þ

W ¼ γV 0 fAI þ AII þ AIII þ ½sinðθ1 þ θ2  ϕÞ cos θ5 =½sinðθ1  ϕÞ cosðθ5  θ2 ÞðAIV þ AV þ AVI Þg ð15Þ

where

lac ¼ ςB= cos ϕ; lcd ¼ ηBð2ς tan ϕ þ 1Þ; lde ¼ Bð1  ηÞð2ς tan ϕ þ 1Þ
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
lcf ¼ B fβ  n½ð1=2Þ þ tan ξ þ ς tan ϕ þ ð1=2Þg2 þ ðα  n=2  ςÞ2
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
lei ¼ B fβ  n½ð1=2Þ þ tan ρ þ ς tan ϕ þ ð1=2Þg2 þ ðα  n=2  ςÞ2
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ldf ¼ B fβ  n½ð1=2Þ þ tan ξ þ ς tan ϕð1  2ηÞ  η þ ð1=2Þg2 þ ðα  n=2  ςÞ2
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ldi ¼ B fβ þ n½ð1=2Þ þ tan ρ þ ς tan ϕð1  2ηÞ  η þ ð1=2Þg2 þ ðα  n=2  ςÞ2
lf g ¼ nB tan ξ; lhi ¼ nB tan ρ; lf j ¼ nB= cos ξ; lik ¼ nB= cos ρ ð16Þ

θ1 ¼ cos1 ðBfβ  n½ð1=2Þ þ tan ξ þ ς tan ϕ Acknowledgments

þ ð1=2Þg=lcf Þ > ϕ
The authors would like to thank Miss Y. Maedera, former graduate
θ2 ¼ cos1 ðBfβ þ n½ð1=2Þ þ tan ξ student of Tokyo Institute of Technology for her contribution to the
þ ς tan ϕ½ð2η  1Þ þ η  ð1=2Þg=ldf Þ model tests. The authors also would like to express our appreciation
to Dr. J. Izawa and Mr. S. Seki for their continuous support of the
θ3 ¼ cos1 ðBfβ þ n½ð1=2Þ þ tan ρ þ ς tan ϕð1  2ηÞ  η model tests.
þ ð1=2Þg=ldi Þ
References
θ4 ¼ cos1 ðBfβ  n½ð1=2Þ þ tan ρ þ ς tan ϕ
Al-Tabbaa, A., Lisa, R., and Michael, O. R. (1989). “Model tests of
þ ð1=2Þg=lei Þ > ϕ footings above shallow cavities.” Ground Eng., 22(7), 39–42.
θ6 ¼ ð90  ξÞ > ϕ; θ7 ¼ ð90  ρÞ > ϕ ð17Þ Azam, G., Jao, M., and Wang, M. C. (1997). “Cavity effect on sta-
bility of strip footing in two-layer soils.” J. Geotech. Eng., 28(2),
151–164.
Badie, A., and Wang, M. C. (1984). “Stability of spread footing above void
AIV ¼ nB2 ð2α  2ς  nÞðtan ξ þ tan ρ þ 1Þ=4 in clay.” J. Geotech. Eng., 110(11), 1591–1605.
ð18Þ Baus, R. L., and Wang, M. C. (1983). “Bearing capacity of strip footing
AV ¼ ðnBÞ2 tan ξ=2; AVI ¼ ðnBÞ2 tan ρ=2Þ above void.” J. Geotech. Eng., 109(1), 1–14.

374 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / APRIL 2011

Downloaded 11 Jun 2012 to 180.211.192.67. Redistribution subject to ASCE license or copyright. Visit http://www.ascelibrary.org
Davis, E. H., Gunn, M. J., Mair, R. J., and Seneviratne, H. N. (1980). “The Poulos, H. G., and Davis, E. H. (1980). Pile foundation analysis and
stability of shallow tunnels and underground openings in cohesive design. Wiley, New York, Chapter 3, 18–51.
material.” Geotechnique, 30(4), 397–416. Shinjo, T., and Nakamura, K. (1975). “On the compressive strength of
Kiyosumi, M., Kusakabe, O., Ohuchi, M., and Peng, F. L. (2007). “Yield- Ryukyu limestone.” Science Bulletin of the College of Agriculture,
ing pressure of spread footing above multiple voids.” J. Geotech. University of the Ryukyus, 22, 269–277 (in Japanese).
Geoenviron. Eng., 133(12), 1522–1531. Sreng Sokkheang, T. T., Ueno, K., and Mochizuki, A. (2002). “Bearing
Kusakabe, O., Maeda, Y., and Ohuchi, M. (1992). “Large-scale loading capacity of ground having a void.” 57th JSCE Annual Meeting,
tests of shallow footings in pneumatic caisson.” J. Geotech. Eng., 1221–1222 (in Japanese).
118(11), 1681–1695. Tomlinson, M. J. (1994). Pile design and construction practice, 4th Ed.,
Ladanyi, B., and Roy, A. (1971). “Some aspects of bearing capacity of rock
E&FN Spoon, New York, Chapter 4, 99–165.
mass.” Proc., of the 7th Canadian Symp. on Rock Mechanics, Edmon-
Wang, M. C., and Badie, A. (1985). “Effect of underground void on foun-
ton, Canada, 161–190.
dation stability.” J. Geotech. Eng., 111(8), 1008–1019.
Leung, C. F., Khan, A., and Ko., H.-Y. (1991). “Capacity of socketed
drilled piers.” Proc., Centrifuge 91, Boulder, CO, 269–275. Wang, M. C., and Hsieh, C. W. (1987). “Collapse load of strip footing
Plaxis, B. V. (1998). “PLAXIS professional version 7.2.” Geotechnical above circular void.” J. Geotech. Eng., 113(5), 511–515.
finite element code for soil and rock analyses (CD-ROM), Delft, Wang, M. C., Kim, Y. U., and Jun, J. T. (2001). “Cavity effect on collapse
Netherlands. load of strip footings.” Proc., 15th ICSMGE, 1, 317–319.

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / APRIL 2011 / 375

Downloaded 11 Jun 2012 to 180.211.192.67. Redistribution subject to ASCE license or copyright. Visit http://www.ascelibrary.org

You might also like