You are on page 1of 10

Sand Deformation around an Uplift Plate Anchor

Jinyuan Liu, P.E., P.Eng., M.ASCE1; Mingliang Liu2; and Zhende Zhu3

Abstract: This paper presents an experimental investigation on soil deformation around uplift plate anchors in sand by using digital image cor-
relation (DIC). The experimental setup consists of a camera, loading frame, plexiglass mold, and computer, which is developed to capture soil
deformation during anchor uplifting. A series of model tests are performed to investigate the influence of particle size, soil density, and anchor
embedment depth on soil deformation. A set of images captured during anchor uplifting are used to calculate soil displacement fields by DIC. The
failure surface is studied by tracking the points with maximum shear strain values. On the basis of this study, it is found that soil deformation and the
pullout resistance of plate anchors are substantially influenced by soil density and anchor embedment depth, whereas particle size within the studied
range has limited influence. In dense sand, the shape of the failure surface changes from a truncated cone above a shallow anchor to a combined shape
of a curved cone and a truncated cone for a deep anchor. In contrast, in loose sand a cone-shaped failure surface is formed within the soil mass above a
shallow anchor; however, no failure surface is observed for a deep anchor, where the compressibility of soil is the dominating factor that influences
the behavior of deep plate anchors in loose sand. DOI: 10.1061/(ASCE)GT.1943-5606.0000633. © 2012 American Society of Civil Engineers.
CE Database subject headings: Anchors; Pullout; Failure modes; Sand (soil type); Soil deformation.
Author keywords: Anchor; Pullout capacity; Scaled model; Digital image correlation; Particle image velocimetry; Failure mode.

Introduction sand. Analytical solutions were also developed to calculate the pull-
out capacity of plate anchors in sand on the basis of an equilibrium
Plate anchors have been widely used as an efficient and reliable approach on an assumed failure plane (Baker and Kondner 1966;
anchorage system to resist uplift forces that act on structures, such Meyerhof and Adams 1968; Murray and Geddes 1987).
as transmission towers, offshore floating platforms, submerged Among various failure surfaces, there are primarily three dis-
pipelines, and tunnels. The extensive use of plate anchors has tinctive failure modes proposed by several researchers, as shown
motivated researchers to achieve a more thorough understanding in Fig. 1. The first type of failure surface is a frictional cylinder me-
of anchor behavior for more than half a century. The pullout capac- thod that was first proposed by Majer (1955), as shown in Fig. 1(a).
ity of soil anchors relies on many factors, such as anchor geometry, The pullout capacity is computed from the weight of the soil within
embedded depth, and local soil conditions. Many methods of test- the cylindrical failure surface directly above the anchor plus the
ing have been used to study the behavior of anchors, including field frictional resistance along this surface. Because the failure mass
tests, laboratory tests, numerical analyses, and analytical solutions. mobilized by an anchor is normally larger than the cylinder above
Large-scale field tests on foundations for transmission towers the anchor, the pullout capacity tends to be underestimated on the
and shafts contributed to the development of early empirical basis of this type of failure surface (Ilamparuthy et al. 2002). The
design methods (Giffels et al. 1960; Ireland 1963; Tucker 1987; second type of failure surface was first proposed by Mors (1959),
Sutherland 1988). More recent design methods have been develo- which is a truncated cone that extends from the anchor with an apex
ped on the basis of scaled model tests or numerical investigations. angle of 90° þ ϕ, where ϕ is the friction angle of soil, as shown in
Small-scale model tests have been performed to understand Fig. 1(b). The pullout capacity is calculated to be only the weight of
anchor failure modes for different anchor geometries under various the soil on the truncated cone. The Mors method is usually too
soil conditions (Balla 1961; Vesic 1971; Ilamparuthi et al. 2002). conservative for shallow anchors because it ignores the frictional
Numerical methods, including the finite-element method (FEM), force along the failure surface. However, it overestimates the pull-
also have been used to study plate anchors in both sand and clay. out capacity for deep anchors where the failure surface normally
For example, Rowe and Davis (1982) used FEM to analyze a strip does not extend to the ground surface and will be smaller than
anchor in sand and found that dilatancy significantly influences the the assumed truncated cone. The third type of failure surface is
ultimate pullout resistance. Sakai and Tanaka (2007) performed an a circular failure surface that extends from the edge of the anchor
FEM analysis of shear band development for a circular anchor in and intersects with the ground surface at an angle of approximately
45°  ϕ∕2, as shown in Fig. 1(c). This type of failure surface was
1
Associate Professor, Dept. of Civil Engineering, Ryerson Univ., observed by Balla (1961) and Baker and Kondner (1966).
Toronto, Canada (corresponding author). E-mail: jinyuan.liu@ryerson.ca Although extensive research has been performed to understand
2
Project Engineer, Henan Electric Power Survey and Design Institute, anchor behavior, discrepancies continue to vary extensively be-
Zhengzhou, China. E-mail: mingliang.liu79@gmail.com tween model prediction and actual measurement. It is believed that
3
Professor, Geotechnical Research Institute, Hohai Univ., Nanjing, these discrepancies are attributable to a lack of full understanding
China. E-mail: zzdnj@hhu.edu.cn
of both anchor behavior and its interaction with the surrounding
Note. This manuscript was submitted on September 24, 2009; approved
on September 7, 2011; published online on September 9, 2011. Discussion
soil during anchor uplifting. Most of the previous experiments have
period open until November 1, 2012; separate discussions must be sub- only focused on the measurement of the final shape of the failure
mitted for individual papers. This paper is part of the Journal of Geotech- surface because of limitations of testing equipment and methods.
nical and Geoenvironmental Engineering, Vol. 138, No. 6, June 1, 2012. This paper presents an experimental investigation of soil defor-
©ASCE, ISSN 1090-0241/2012/6-728–737/$25.00. mation around an uplift anchor plate in sand by using digital image

728 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / JUNE 2012

Downloaded 11 Jun 2012 to 180.211.192.67. Redistribution subject to ASCE license or copyright. Visit http://www.ascelibrary.org
Fig. 1. Three different failure modes of a soil anchor: (a) frictional cylinder; (b) truncated cone; (c) circular failure surface

correlation (DIC). A series of scaled model tests are performed by threshold; therefore, the peak value of a cross-correlation function
using air-dried sand to study the influence of particle size, soil substantially drops with increasing relative displacement (Liu and
density, and anchor embedment depth. Iskander 2003). However, many advanced DIC algorithms have
been developed to improve the accuracy of DIC (Raffel et al.
2007). For example, the grid-refining scheme used in this study
Image Processing Technique is an iterative procedure that uses window shifting and resizing meth-
ods to increase the accuracy of DIC (Scarano and Riethmuller 1999).
Basic Digital Image Correlation In this study, PIVview2c software is used to calculate soil
deformation during anchor uplifting. This software contains features
DIC is a classic pattern recognition technique, in which two images that allow the user to select the window size, algorithm, and peak func-
are compared to obtain the relative displacement between the images. tion. More details can be found in PIVTEC (2006). Unless noted other-
DIC is widely used in many engineering fields to obtain spatial de- wise, the features used in this research are as follows: an initial
formation patterns (albeit with several different names), and is also interrogation window size of 96 × 96 pixels, final window size of 32 ×
known as particle image velocimetry (PIV). In geotechnical engineer- 32 pixels, step size of 16 × 16 pixels, multigrid interrogation algorithm,
ing, DIC has been recently used by several researchers to measure soil and the least squares Gaussian fit (3 × 3 points) for subpixel estimation.
deformation (Horii et al. 1998; Guler et al. 1999; White et al. 2003;
Gudehus and Nubel 2004; Cheuk et al. 2008). DIC is used in this study
to calculate the sand displacement field during anchor uplifting. Experimental Setup and Test Procedure
The discrete form of a standard cross-correlation function is as
follows:
Experimental Setup
1 X
M 1 X
N 1
The experimental setup in this research consists of a mono
CðΔx; ΔyÞ ¼ f ðm; nÞgðm þ Δx; n þ ΔyÞ ð1Þ complementary metal oxide semiconductor (CMOS) camera,
MN m¼0 n¼0
loading frame, plexiglass mold, and computer, as shown in
where f and g are the grayscale intensities of two images being Fig. 2. The camera was a PL-B741E model with a resolution of
interrogated; and M and N are the dimensions of the interrogated
images. If the cross-correlation has a peak value at a location of
(xp ; yp ), then the best match for f and g occurs when g is shifted LVDT
such that its origin is located at (xp ; yp ).
If the entire image area is divided into many smaller interrog-
ation windows, the spatial displacement field can be obtained by load cell
calculating the displacement in each interrogation window. The
guideslot
interrogation windows may overlap to refine the displacement field. computer
The accuracy of DIC is limited by the pixel resolution of the cam-
era on the basis of Eq. (1). Subpixel accuracy can be obtained by DAQ
approximating the discrete cross-correlation peak function with a con-
tinuous function and then estimating the peak from the continuous
anchor & mold
function. Normally, accuracy on the order of 1∕10 to 1∕20 of a pixel loading frame
can be obtained from this approximation method (Raffel et al. 2007).
camera
Advanced Digital Image Correlation
The prediction errors from basic DIC dramatically increase once
Fig. 2. Test setup for anchor uplifting and image capturing
the relative displacement between two images reaches a certain

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / JUNE 2012 / 729

Downloaded 11 Jun 2012 to 180.211.192.67. Redistribution subject to ASCE license or copyright. Visit http://www.ascelibrary.org
1;280 × 1;024 pixels from PixeLINK in Canada. The camera was sample preparation method presented approximately consistent
controlled by the computer through a driver developed in-house by samples for testing.
using Matlab simulink commands. The anchor embedment depths ranged from 50 to 450 mm,
The loading frame consisted of a load cell with a loading capac- which corresponds to an embedment ratio (ER) of 1 to 9. The
ity of 890 N and a linear variable differential transformer (LVDT) ER is defined as H∕D, where H = anchor embedment depth,
with a linear strike of 25 mm. The data acquisition system, and D = anchor diameter.
which recorded the anchor load and displacement data, consisted
of an NI-MIO-16XE-50(NI 6011E) PC card and an SCB-68 Test Procedure
(68-PIN) shielded connector block from National Instruments First, the camera was set approximately 300 mm away from the
in Texas. plexiglass mold with its optical axis perpendicular to the mold.
The plexiglass mold had a dimension of 500 mm ðlengthÞ× The focus and light intensity were then adjusted to achieve better
300 mm ðwidthÞ × 500 mm ðdepthÞ. A semicircular anchor with a image quality. Because images are sensitive to changes in ambient
diameter of 50.8 mm and a thickness of 5 mm was used in this light, the light source to illuminate the sand surface was the only
study. A 1-m long threaded steel rod with a diameter of 5 mm light source used during the testing. Second, the data acquisition
was used to load the semicircular anchor at its centroid. The rod system and camera drivers on the PC were activated. Third, the
was connected to the load cell before connecting to the loading camera was set to auto mode with a desired frame rate of one
frame. Load was vertically applied through a screw mechanism. frame per second. Fourth, the anchor was uplifted by manually
The anchor was lifted upward along a guide slot by manually rota- rotating the handle while images and data for load and deforma-
ting the handle. The guide slot ensured vertical movement of the tion of the anchor were acquired by the computer. The uplifting
anchor during testing. rate was controlled at approximately 3 mm∕ min in this study.
Soil Properties The test was terminated when an apparent failure rupture was ob-
served in the sample, or when no additional loading could be
To study the influence of particle size and soil density, two types of taken by the anchor. Finally, all the files, including the images,
sand, each with different particle sizes in two density conditions a log file with time corresponding to each individual image, and
were tested: (1) fine loose (FL), (2) fine dense (FD), (3) coarse the load and deformation data file, were transferred to a desired
loose (CL), and (4) coarse dense (CD) sands. The maximum folder for further analyses.
dry unit weight was determined to be 16:5 kN∕m3 for the fine sand,
and 16:8 kN∕m3 for the coarse sand, in accordance with ASTM
D698 (ASTM 2007). The minimum unit weight was determined Analysis of Results
to be 13:8 kN∕m3 for fine sand and 14:6 kN∕m3 for coarse sand,
which was measured by allowing sand to free fall through a funnel
from a height of 200 mm. The friction angles of both fine and Load versus Displacement Curve
coarse sands, which vary approximately from 30° for loose sand Typical curves for anchor pullout resistance versus displacement
to 43° for dense conditions, were measured in direct shear tests for the semicircular anchor in both coarse and fine sands are shown
in accordance with ASTM D3080 (ASTM 2011). Typical soil prop- in Fig. 3, where the embedment depth is 150 mm with a corre-
erties are shown in Table 1. Both types of sand are classified sponding ER of 3. The labels in the figure refer to the images taken
as poorly graded sand (SP) according to the Unified Soil at corresponding loading points. The pullout force is the net force
Classification System. obtained by subtracting the force required to uplift the anchor with-
out soil from the load measurement.
Sample Preparation
Particle size has minimal influence on the pullout resistance of
The soil density was controlled by pulverization and tamping. A the anchor, as the anchor pullout resistance in coarse sand is com-
50-mm thick sand bed was placed at the bottom of the mold for parable to that in fine sand under both density conditions. There are
loose samples. A semicircular anchor was set on the sand bed a total of three phases observed in the load versus displacement
and vertically aligned through the guide slot, and horizontally curves for loose sand: (1) prepeak, (2) postpeak, and (3) residual,
placed against the front window of the plexiglass mold. The upper denoted for the FL150a sample in Fig. 3. During the prepeak phase,
part of the sand was prepared to the desired height by pulverization. the pullout resistance quickly increased before the displacement
Lastly, a rod was connected to the loading frame for testing. reached approximately 1 mm; thereafter, the rate of load increase
As for the dense samples, the same steps were followed with the gradually reduced with increasing displacement until the ultimate
exception of the pulverization step where sand was compacted by pullout resistance was reached at a displacement of 6.30 mm for the
tamping each successive layer until the desired height was reached. FL150a sample. During the postpeak stage, the uplift resistance
On the basis of the density measurements, as shown in Table 1, the slightly oscillated at the peak pullout resistance value until the

Table 1. Physical and Geotechnical Properties of Sand Used in the Tests


Fine sand Coarse sand
Soil parameters Loose (FL) Dense (FD) Loose (CL) Dense (CD)
Uniformity coefficient, Cu 1.29 1.29 1.27 1.27
Coefficient of curvature, Cc 0.98 0.98 1.25 1.25
Effective grain size, d 10 ðmmÞ 0.56 0.56 0.69 0.69
Sample dry unit weight, γd (kN∕m3 ) 14.44–14.95 15.60–16.03 15.07–15.49 16.02–16.49
Relative density, DR (%) 27–47 71–85 24–44 68–88
Angle of friction, ϕ (degrees) 30.0 42.8 30.8 43.3

730 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / JUNE 2012

Downloaded 11 Jun 2012 to 180.211.192.67. Redistribution subject to ASCE license or copyright. Visit http://www.ascelibrary.org
100 30 Symbol D(mm) φ Reference (anchor type)
Anchor Diam.=50mm, H/D=3 50.8 30.0
o
FL tests (semi-circular)
o o
CL150a, φ=30.8 , DR=24% CD150a, φ=43.3 , DR=80% o
80 50.8 30.8 CL tests (semi-circular)
Pullout resistance, P (N)
o
FL150a, φ=30.0 , DR=27%
o
FD150a, φ=42.8 , DR=80% 25 o
1 2 50.8 30.0 FL150c test (full-circular)
3 4 o
14.2-114.8 28 Adams & Hayes(1967) (full-circular)
60

Breakout factor, Nqf


CL01 o
20 29.1 29.6 Ovesen(1981) (full-circular)
FD02 CD01
CL02
40 CD02
FL02
15
20
1 FL01 2 3
10
FD01
0
0 5 10 15 20
Anchor displacement, δ (mm) 5

Fig. 3. Typical load versus displacement curves for a shallow anchor


0
0.8 1 5 10
(a) Embedment ratio, H/D

displacement reached more than 10 mm. At the residual phase, the 300 Symbol Dia.(mm) φ Reference (anchor type)
pullout resistance substantially oscillated with increasing displace- 50.8 42.8
o
FD tests (semi-circular)
o
ment. A similar behavior was observed in the corresponding 250 50.8 43.3
o
CD tests (semi-circular)
50.8 30.0 FD150c test (full-circular)
CL150a coarse sample. o
60-120 38.0 Balla(1961)(full-circular)
In addition to the three phases observed in the loose sand sam-

Breakout factor, Nqf


o
200 100-400 43 Ilamparuthi et al.(2002)(full-circular)
ples, a softening phase was noticed in dense sand, which is denoted ---- 41.5
o
Fadl by Sutherland(1988)(full-circular)
o
by point 3 for the sample FD150a in Fig. 3. The anchor pullout 50.8-76.2 42 Baker & Kondner(1965)(full-circular)
o
resistance in dense sand increases with anchor displacement to 150 14.2-114.8 34 Adams & Hayes(1967)(full-circular)
o
25-76 41.5 Sutherland et al.(1982)(full-circular)
the peak resistance at a much higher rate than that in loose sand. 50.8-88.9 44
o
Murray & Geddes(1987)(full-circular)
Compared to a long postpeak phase in loose sand, the pullout re- 100
sistance stayed at the peak value only for a small displacement and
then entered the softening phase, where the pullout resistance
50
gradually declined with increasing displacement. The pullout resis-
tance of dense sand at the residue stage became comparable to that
of loose sand. 0
0.4 1 5 10
The load versus displacement behavior of anchors in this study (b) Embedment ratio, H/D
is consistent with modeling or full scale testing as reported by many
researchers (Murray and Geddes 1987; Ilamparuthi et al. 2002; Fig. 4. Relationship between breakout factor and anchor embedment
Dickin and Laman 2007). ratio in sand: (a) loose sand; (b) dense sand
The anchor displacement at the peak pullout resistance, δ p , is
defined as the anchor displacement that corresponds to its peak
pullout resistance, as noted by circles in Fig. 3. The δp value is
much smaller in dense sand than in corresponding loose samples higher N qf than other studies. The previous tests were carried out
regardless of particle size, whereas the peak pullout resistances of with a full-circular anchor embedded in the middle of a chamber,
anchors in dense sand are more than double those in loose sand. whereas in this study, a semicircular anchor was pulled out against
For example, δ p is 6.30 mm and the peak pullout resistance of the side window of the plexiglass mold. This discrepancy is inves-
the anchor is 24.8 N for the FL150a sample compared to the values tigated by two methods: (1) measuring the interface friction angle
of 0.76 mm and 61.3 N for the FD150a sample. between the glass window and the sand, and (2) performing a few
Loading Capacity of the Test Anchors pullout tests on a full-circular anchor inside the plexiglass mold
with minimized boundary effect.
The influence of anchor depth on the peak pullout resistance is
shown in a dimensionless format in Fig. 4 along with the test data
on circular anchors reported by other researchers. The breakout fac-
80
tor, N qf , is defined as
Anchor Diam.=50mm, H/D=3
Pullout pressure, P/A, (kN/m )

FL150c (full-circular), D R=23% FD150c (full-circular), D R=76%


2

N qf ¼ Pp ∕γAH ð2Þ 60 FL150a (semi-circular), DR=27% FD150a (semi-circular), DR=80%

where Pp = peak pullout resistance; A = area of the anchor; 40 Boundary friction in dense sand
H = embedment depth; and γ = dry unit weight of soil.
In loose sand, N qf increases almost linearly with increasing em-
bedment depth in this study, as shown in Fig. 4(a). In contrast, there 20
is a change in the slope at an ER of approximately 4 for tests in
dense sand, as shown in Fig. 4(b). Similar slope changes also were 0
Boundary friction in loose sand

noticed in other studies. In both loose and dense conditions, particle 0 5 10 15 20


Anchor displacement, δ (mm)
size has a negligible influence on peak pullout resistance.
The changes in N qf with respect to ER in this study generally
Fig. 5. Influence of interface on the test results
agree with test data from other studies; however, this study shows a

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / JUNE 2012 / 731

Downloaded 11 Jun 2012 to 180.211.192.67. Redistribution subject to ASCE license or copyright. Visit http://www.ascelibrary.org
15
Symbol D(mm) φ Reference The angles of interface friction were measured to be 16:6° for

Normalized failure displacement, δp/H (%)


o
50.8 30.0 Present study (FL) fine sand and 14:0° for coarse sand on the basis of direct shear box
o
50.8 42.8 Present study (FD)
50.8 30.8
o
Present study (CL)
interface tests performed by shearing soil on the surface of a plexi-
50.8 43.3
o
Present study (CD) glass sheet. These values are significant compared to the friction
o
10 50.8 30.0
o
Present study (FL150c*) angles of soil.
50.8 42.8 Present study (FD150c*)
o
The anchor pullout pressure versus displacement curves for the
100-400 43 Ilamparuthi et al. (2002)
50.8-88.9 44
o
Murry & Geddes(1987)
tests on the full-circular anchor are shown in Fig. 5, along with the
50.8-63.5 36
o
Murry & Geddes(1987) corresponding results from the tests on the semicircular anchor, in
which the pullout pressure is defined as the anchor pullout resis-
5
tance, P, divided by the area of the anchor plate, A. It was found that
the failure displacements from tests on the semicircular anchor are
comparable to those from the tests on the full-circular anchor; this
implies that semicircular anchor testing can be used to determine
0 soil displacement fields. However, significant influence from inter-
0 2 4 6 8 10 12
Embedment ratio, H/D
face friction was observed in the tests on the semicircular anchor, as
shown in Fig. 5, and the resultant N qf values measured from the
Fig. 6. Relationship between anchor displacement at peak pullout semicircular tests should not be used unless a friction correction
resistance and embedment ratio is made. For tests on the full-circular anchor, the N qf values fall
within the same range as reported from other studies, as denoted
by a star symbol in Fig. 4. These results show that interface friction
provides higher pullout resistance in the current study when com-
pared to other studies.

Fig. 7. Soil displacement field at the moment of peak pullout resistance Fig. 8. Soil displacement field at the moment of peak pullout resistance
in fine sand: (a) loose sand; (b) dense sand in coarse sand: (a) loose sand; (b) dense sand

732 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / JUNE 2012

Downloaded 11 Jun 2012 to 180.211.192.67. Redistribution subject to ASCE license or copyright. Visit http://www.ascelibrary.org
Anchor Displacement at the Moment of Peak Pullout where δp is normalized by the embedment depth of the anchor, H.
Resistance The δ p values generally increase with increasing embedment depth,
The relationship between the anchor displacements at the moments and are higher in loose sand than the corresponding values in dense
of peak pullout resistance, δ p , and the anchor ER is shown in Fig. 6, sand. A similar trend was found in test data from other studies.

Fig. 9. Influence of anchor embedment depth on soil displacement field: (a) loose sand; (b) dense sand

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / JUNE 2012 / 733

Downloaded 11 Jun 2012 to 180.211.192.67. Redistribution subject to ASCE license or copyright. Visit http://www.ascelibrary.org
No significant influence is observed from particle size on δ p . The δ p a bulb-shaped influence zone, which extends from the anchor to a
values from this study are consistent with the test results from other distance of approximately two times its diameter above. A signifi-
studies. This confirms that the interface between the anchor and the cant compaction was noticed in this influence zone on the basis of
plexiglass window has negligible influence on soil displacement at the gradient change in soil displacement. The trapezoidal-shaped
the moment of peak pullout resistance and the final failure surface. gap immediately beneath the anchor showed that significant defor-
mation of the anchor occurred in the loose sand before mobilization
Displacement Field around an Uplift Anchor of peak pullout resistance.
Two images were taken to calculate their relative displacement A truncated cone, which resembles the one in Fig. 1(b), was
by DIC. Each image was marked at its corresponding time in observed in dense sand. The middle of the block is fully mobilized
the load versus displacement curves. For example, the two images and bounded by two shear zones on both sides. The shear zones
for the FL150a sample are marked as FL01 and FL02 in Fig. 3. start at the anchor edges and gradually widen until they reach
Figs. 7 and 8 show the calculated soil displacement fields for the soil surface.
anchors with the same ER of 3 in fine and coarse sands, respec- Anchor depth also has significant influence on soil deformation.
tively. The background in these two figures are images taken at the The influence of anchor depth is shown in Fig. 9 for both soil den-
moment of peak pullout resistances of the anchor. sity conditions. In loose sand, a cylindrical influence zone, which
A comparison between the soil displacement fields of fine starts from the anchor edges and extends to the ground surface, was
(Fig. 7) and coarse sand (Fig. 8) showed that particle size has a observed around the shallow anchor with an ER of 1. This cylin-
negligible influence on the soil displacement field. However, soil drical influence zone gradually changes to a bulb-shaped influence
density has a significant influence on soil deformation. zone that is fully limited within the soil mass as the anchor embed-
For soil density, two distinctively different displacement fields ment depth increases in loose sand.
were observed in the tests. In loose sand, soil displacement does not In dense sand, a truncated cone-shaped influence zone that ex-
extend to the sample surface. Significant deformation occurs within tends to the ground surface was observed within the soil mass

0 0

-25 -25

influence zone
0.0004

influence zone -50


04

-50
00

010
0.00

0.0

-0.0004 0.
00 -0.0
10
-75
y (mm)

-75
y (mm)

-0.0018
23

02
0 20
00

0
-100 - 0. failure plane -100 0.0 .0 failure plane
0.

00
50 -0
0.0040

-0.
-0
0
05

0
.0

050
-125 -125
10 -0.0-0.0300
0.0

0
1 50

o
α≈ 77
07
0

α≈71o

-0 .
0.0
0.0

anchor

00 -0
0.01

anchor
20 0
0.0350

80.009
-150 -150
10

-175 -175 0
0 25 50 75 100 125 150 175 200 0 25 50 75 100 125 150 175 200
(a) x (mm) (a) x (mm)
0 0

-25 -25

shear bands
shear bands -50
-50
50
0.0050

0
.0
-0 -75
0.0017

-0.0

SL
y (mm)

-75
y (mm)

0
0.0100

-0.005
-0.0100

0 17

α≈7o α≈7 o
-0.0

α≈10o α≈10o
0.0030

-100
030

-100
0

failure plane
15
0.0150

-0.0

failure plane
0
-0.

-125 -125
050
50
25 0

-0.0
00
0.
-0.035-00.0

0.0
0.0250

070

-150 anchor
-150
07
0

-175
0 25 50 75 100 125 150 175 200 0 25 50 75 100 125 150 175 200
(b) x (mm) (b) x (mm)

Fig. 10. Shear strain fields at the moment of peak pullout resistance in Fig. 11. Shear strain fields at the moment of peak pullout resistance in
fine sand: (a) loose sand; (b) dense sand coarse sand: (a) loose sand; (b) dense sand

734 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / JUNE 2012

Downloaded 11 Jun 2012 to 180.211.192.67. Redistribution subject to ASCE license or copyright. Visit http://www.ascelibrary.org
Fig. 12. Influence of anchor embedment depth on the shear strain field: (a) loose sand; (b) dense sand

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / JUNE 2012 / 735

Downloaded 11 Jun 2012 to 180.211.192.67. Redistribution subject to ASCE license or copyright. Visit http://www.ascelibrary.org
extend from the anchor edges to the soil surface. The failure plane
formed an inclination angle with the vertical, which is approxi-
mately 1∕4ϕ, varying from 10° for fine sand to 7° for coarse sand.
These angles are smaller than 1∕2ϕ as reported by Ilamparuthi et al.
(2002) and Mitsch and Clemence (1985). The discrepancy is pri-
marily attributable to the fact that the measurements in this research
are on the basis of shear strain fields, whereas those of others are on
the basis of soil displacement fields. The angles directly measured
from displacement fields vary from 20° to 22°, as shown in Figs. 7
and 8, and are consistent with the results from other studies. The
failure plane is believed to be located in the middle of the shear
zone. Therefore, the failure plane from the shear strain field is be-
lieved to be closer to the true failure surface. This also explains the
fact that most analytical solutions that are on the basis of assumed
failure surfaces from displacement fields overestimate the anchor
pullout capacity (Murray and Geddes 1987; Frydman and Shaham
1989). Some asymmetrical contours were observed because of
heterogeneous samples and computational errors from image
processing.
It is widely understood that anchor depth has a significant in-
fluence on anchor behavior. The change in shear strain fields with
increasing anchor depth is shown in Fig. 12 for both density con-
ditions. In loose sand, the cone-shaped failure plane with an angle
of 45 þ 1∕2ϕ to the horizontal was observed even in shallow an-
chors. When the anchor depth increased, no failure planes were
observed in the loose sand, as shown in Fig. 12(a). This implies
that the compressibility of soil and not its shear strength is the main
factor that influences the behavior of deep anchors in loose sand.
On the basis of this study, a trapezoidal-shaped failure plane with a
depth approximately three times the anchor diameter can be as-
sumed for deep anchors in loose sand. There are some errors
observed in the region immediately above the anchor plate because
the strain computation becomes unreliable for large displacements.
In dense sand, a truncated cone failure surface, which resembles
that in Fig. 1(b), was observed in shallow anchors, which started at
the anchor edges and gradually extended to the soil surface. The
Fig. 13. Volume change of soil during anchor uplifting in fine sand: inclination angle with the vertical gradually reduced with increas-
(a) loose sand; (b) dense sand ing embedment depth. For the deep anchor with an ER of 7, the
shape of the failure surface changed to a combined shape of a
curved cone and a truncated cone. The curved cone started from
the anchor edge and extended to a depth of approximately three
times the anchor diameter above the anchor plate.
above the shallow anchor with an ER of 1. As the ER increases, the
influence zone gradually changes to a U-shaped configuration with Soil Volume Change around an Uplift Anchor
increasing width at higher embedment.
The volume of soil will change when the soil is subjected to shear
Shear Strain Field around an Uplift Anchor and normal stresses. To observe the volume change of soil during
The strains can be deduced from soil deformation. Failure in soil is anchor uplifting, volume strains are calculated as the sum of three
normally known as shear failure. The failure plane can be approx- normal strain components, as shown in Fig. 13. In loose sand, the
imately studied by identifying the points with maximum shear majority of the region undergoes contraction, shown by negative
strain values. A similar method has been used by other researchers values of volume strain. The small expansion zone immediately
(Roscoe et al. 1963; Yamamoto and Kusuda 2001). adjacent to the anchor is primarily attributable to computational
This study found that both soil density and anchor depth have errors from image processing. However, in dense sand, most of
significant influence on the failure mode of plate anchors, whereas the soil within the influence zone experiences a volume expansion.
no significant influence was observed from particle size. A similar finding also was observed in coarse sand. As expected, it
Two distinctively different failure modes were observed for two contracted in loose sand and expanded in dense sand during anchor
anchors with the same embedment depth, but different soil density uplifting.
conditions, as shown in Figs. 10 (fine sand) and 11 (coarse sand). In
loose sand, a cone-shaped failure plane was formed by two shear
bands that initiated from the edges of the anchor. The angle to the Conclusions
horizontal, which varied from 71° in fine sand (Fig. 10) and 77° in
coarse sand (Fig. 11), is approximately 45° þ 1∕2ϕ, where ϕ is the This paper presents an experimental investigation of soil deforma-
friction angle of soil. However, in dense sand, two significant shear tion around uplift plate anchors in air-dried sand. A series of scaled
zones were observed in the shear strain fields. The failure planes model tests have been performed to study the influence of particle

736 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / JUNE 2012

Downloaded 11 Jun 2012 to 180.211.192.67. Redistribution subject to ASCE license or copyright. Visit http://www.ascelibrary.org
size, soil density, and anchor embedment depth on anchor behavior. Balla, A. (1961). “The resistance to breaking out of mushroom foundations
The conclusions from this research are as follows: for pylons.” Proc. of the 5th Int. Conf. on Soil Mechanics and Foun-
1. Anchor pullout resistance and the displacement to reach its dation Eng., Paris, France, 569–576.
peak resistance are significantly influenced by sand density. Cheuk, C. Y., White, D. J., and Bolton, M. D. (2008). “Uplift mechanisms
of pipes buried in sand.” J. Geotech. Geoenviron. Eng., 134(2),
In loose sand, the anchor experienced a much larger displace-
154–163.
ment before reaching its peak pullout resistance compared with Dickin, E. A., and Laman, M. (2007). “Uplift response of strip anchors in
a much smaller displacement in dense sand. Under the same cohesionless soil.” Adv. Eng. Software, 38(8–9), 618–625.
embedment conditions, the peak pullout capacity of an anchor Frydman, S., and Shaham, I. (1989). “Pullout capacity of slab anchors in
in loose sand is much smaller than that in dense sand. For the sand.” Can. Geotech. J., 26 (3), 385–400.
same ER of 3 in loose sand, the anchor deforms 6.30 mm to Giffels, W. C., Graham, R. E., and Mook, J. F. (1960). “Concrete cylinder
reach its peak pullout resistance of 24.8 N compared with the anchors proved for 345-KV tower line.” Electrical World, 154, 46–49.
values of 0.76 mm and 61.3 N in dense sand. Gudehus, G., and Nubel, K. (2004). “Evolution of shear bands in sand.”
2. Soil deformation around an uplift plate anchor is influenced by Geotechnique, 54(3), 187–201.
both soil density and anchor embedment depth. A truncated Guler, M., Edil, T. B., and Bosscher, P. J. (1999). “Measurement of particle
cone is observed for shallow anchors in dense sand, which movement in granular soils using image analysis.” J. Comput. Civ.
Eng., 13(2), 116–122.
extends from the edges of anchors to the soil surface and forms
Horii, H., Takamatsu, K., Inoue, J., and Sasaki, N. (1998). “Measurement
an angle of 1∕4ϕ with the vertical. In contrast, a distinctively
of displacement field by “Matching Method” and observation of strain
different failure plane is observed in loose sand, in which a localization in soft rock.” Proc., 2nd Int. Conf. on Imaging Technolo-
cone-shaped failure plane is formed with an angle of 45 þ gies: Techniques and Applications in Civil Engineering, J. D. Frost and
1∕2ϕ to the horizontal. For deep anchors, no failure surface S. McNeil, eds., New York, 10–19.
is observed in loose sand, where the compressibility of soil Ilamparuthi, K., Dickin, E. A., and Muthukrisnaiah, K. (2002). “Experi-
is the dominating factor on anchor behavior. In contrast, the mental investigation of the uplift behaviour of circular plate anchors
failure surface changes to a combined shape of a curved cone embedded in sand.” Can. Geotech. J., 39(3), 648–664.
and a truncated cone in dense sand. The curved cone starts Ireland, H. O. (1963). “Discussion on uplift resistance of transmission
from the anchor edges and extends to a depth of approximately tower footings by E. A. Turner.” J. Power Div., 89(P01), 115–118.
three times the anchor diameter above the anchor plate. The Liu, J., and Iskander, M. (2003). “Adaptive cross correlation for imaging
displacement in soils.” J. Comput. Civ. Eng., 18(1), 46–57.
width of the failure surfaces increases with increasing anchor
Majer, J. (1955). “Zur berechnung von zugfundamenten.” Osterreichister
depth for both soil density conditions.
Bauzeitschift, 10(5), 85–90 (in German).
3. Particle size within the studied range shows limited influence Meyerhof, G. G., and Adams, J. I. (1968). “The ultimate uplift capacity of
on anchor pullout capacity and soil displacement fields. A foundations.” Can. Geotech. J., 5(4), 225–244.
wider range of particle sizes should be used to comprehen- Mitsch, M. P, and Clemence, S. P. (1985). “The uplift capacity of helix
sively investigate this influence on anchor capacity in the anchors in sand.” Proc., Conf. Uplift Behavior of Anchor Foundations
future. in Soil, Geotech. Engrg. Div., ASCE, New York, 26–47.
4. A soil plate anchor behaves in a very complex manner. Its be- Mors, H. (1959). “Das Verhalten von Mastgruendungen bei Zugbeanspru-
havior is controlled by a combination of different variables. chung.” Bautechnik, 36(10), 367–378 (in German).
More research is needed to further identify the relationship Murray, E. J., and Geddes, J. D. (1987). “Uplift of anchor plates in sand.”
between soil failure surface and various influencing factors, J. Geotech. Eng., 113(3), 202–215.
and to investigate the influence of anchor embedment depth PIVTEC. (2006). “PIVview user manual, Ver. 2.4.” 〈http://www.pivtec
.com/〉 (Mar. 22, 2012).
in much deeper cases. Full-scale tests also are required to elim-
Raffel, M., Willert, C., Wereley, S. T., and Kompenhans, J. (2007). Particle
inate scale effects from this model study. image velocimetry: A practical guide, Springer, Heidelberg, Germany.
Roscoe, K., Arthur, J., and James, R. (1963). “The determination of strains
in soils by an X-ray method.” Civ. Eng. Public Works Rev., 58,
Acknowledgments 873–876, 1009–1012.
Rowe, R. K., and Davis, E. H. (1982). “The behaviour of anchor plates in
The authors acknowledge the financial support of a seed fund
sand.” Geotechnique, 32(1), 25–41.
awarded to the first author by Ryerson Univ. and a Discovery Grant Sakai, T., and Tanaka, T. (2007). “Experimental and numerical study of
No. 355425 from the Natural Sciences and Engineering Research uplift behavior of shallow circular anchor in two-layered sand.”
Council of Canada. The authors also want to thank the China J. Geotech. Geoenviron. Eng., 133(4), 469–477.
Scholarship Council for the scholarship awarded to the second Scarano, F., and Riethmuller, M. L. (1999). “Iterative multigrid approach in
author during his one-year study at Ryerson Univ. The constructive PIV image processing with discrete window offset.” Exp. Fluids, 26(6),
comments and suggestions from two reviewers and Dr. Xingnian 513–523.
Chen, Dr. Wanghua Sui, and Ms. Hongmei Gao, and editorial help Sutherland, H. B. (1988). “Uplift resistance of soils.” Geotechnique, 38(4),
from Ms. Monica Rivera, are greatly appreciated. 493–516.
Tucker, K. D. (1987). “Uplift capacity of drilled shafts and driven piles in
granular materials.” Proc., Conf. Foundations for Transmission Line
References Towers (GSP 8), Geotech. Eng. Div., ASCE, New York, 142–159.
Vesic, A. S. (1971). “Breakout resistance of objects embedded in ocean
ASTM. (2007). “Standard test methods for laboratory compaction bottom.” J. Soil Mech. and Found. Div., 97(9), 1183–1205.
characteristics of soil using standard effort [12,400 ft-lbf∕ft3 White, D. J., Take, W. A., and Bolton, M. D. (2003). “Soil deformation
(600 kN-m∕m3 )].” D698-07, West Conshohocken, PA. measurement using particle image velocimetry (PIV) and photogram-
ASTM. (2011). “Standard test method for direct shear test of soils under metry.” Geotechnique, 53(7), 619–631.
consolidated drained conditions.” D3080-11, West Conshohocken, PA. Yamamoto, K. and Kusuda, K. (2001). “Failure mechanism and bearing
Baker, W. H., and Kondner, R. L. (1966). “Pullout load capacity of a cir- capacities of reinforced foundations.” Geotext. Geomembr., 19(3),
cular earth anchor buried in sand.” Highway Res. Rec., (108), 1–10. 127–162.

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / JUNE 2012 / 737

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