Influence of Dilation Angle on Drained Shallow Circular

Anchor Uplift Capacity

Downloaded from ascelibrary.org by The State University of New York at Buffalo (SUNY - Buffalo) on 06/08/16. Copyright ASCE. For personal use only; all rights reserved.

Joseph R. Giampa, S.M.ASCE1; Aaron S. Bradshaw, Ph.D., P.E., A.M.ASCE2;

and James A. Schneider, Ph.D., P.E.3

Abstract: An experimental study of uplift capacity of 22 circular helical anchors installed in sand with peak friction angles between 40
and 50° was performed. Laboratory triaxial tests indicated that the dilation angle varied between 10 and 25° for these peak friction
angles. To account for soil behavior exhibiting nonassociated flow (NAF), in which the dilation angle is much less than the friction
angle, a limit equilibrium plane strain analytical solution for plate anchor uplift was updated and extended to axisymmetric conditions.
Anchor test results were compared with upper bound (UB) plasticity solutions (based on associated flow) and the newly developed NAF
limit equilibrium model. The UB solution overpredicted uplift capacity by more than a factor of 2, whereas the limit equilibrium model
had a ratio of calculated to measured capacity of 1.15 and a coefficient of variation of 0.14. Although additional study is warranted, the
consistency among the numerical, analytical, and experimental results gives confidence in the further application of the NAF limit equi-
librium analytical solution presented in this paper. DOI: 10.1061/(ASCE)GM.1943-5622.0000725. © 2016 American Society of Civil
Engineers.

Introduction anchor plate. It should be noted that submerged unit weight ( g 0 )

would replace dry unit weight in Eq. (1) for saturated sands.
This paper focuses on the uplift capacity of single-helix anchors in Shallow anchor problems are a rigorous verification problem for
sands, which have a wide range of engineering applications both numerical analyses, in that unique numerical upper bound (UB) and
onshore and offshore (Lutenegger 2011; Byrne and Houlsby 2015). lower bound (LB) solutions have been shown for strip (Smith 1998)
Like other foundation elements, the design of a helical anchor and circular anchors (Merifield et al. 2006). The UB solution also
requires the prediction of its axial capacity. When the helix is em- has a convenient numerical form, which represents the weight of
bedded at shallow depths, uplift of the anchor will cause the failure soil in a wedge with an inclination angle (u ) equal to the friction
surface to extend toward the ground surface, but at greater depths angle above the anchor (Murray and Geddes 1987; White et al.
the failure becomes more localized. This transition from “shallow” 2008).
to “deep” failure is not distinct and is thought to range from depth Common failure modes assumed for the analysis of drained
to diameter ratios (D/B) of 4 for low friction angle sands to 10 for loading of shallow helical anchors are shown in Fig. 1. Traditional
high friction angle sands (Meyerhof and Adams 1968). For this analyses of these and other proposed failure modes are typically
study the primary focus is on the assessment of shallow anchors based solely on peak friction angle ( f 0p) (Meyerhof and Adams
with an embedment of less than or equal to 7 diameters in sands 1968; Vesic 1971; Sarac 1989; Ghaly and Hanna 1994; Ilamparuthi
with peak dilation angles ( c p) between 10 and 25°. Anchor uplift et al. 2002; Merifield et al. 2006). These analyses are either based
capacity in dry sand is commonly expressed as a breakout factor on the principle of associated flow (AF), where c p = f 0p, or they
(N g ) defined as inherently assume that the friction angle can uniquely capture the
influence of dilation angle. The analyses and interpretation dis-
Qu cussed in this paper suggest that the dilation angle itself needs to be
Ng ¼ (1)
g DA included in design formulations for assessment of uplift capacity of
circular and helical anchors, because the previously mentioned
where Qu = peak uplift resistance; g = dry unit weight; D = depth assumptions do not reflect drained soil behavior. Although compu-
from the soil surface to the anchor plate; and A = bearing area of the tationally rigorous, the assumption of AF is known to overpredict
drained foundation capacity in soils (Davis 1968; Drescher and
Detournay 1993; Loukidis et al. 2008; Krabbenhoft et al. 2012;
Sloan 2013), thus dilation angle should be directly included in the
1 analyses. Additionally, peak friction angle is controlled by the
Ph.D. Candidate, Dept. of Civil and Environmental Engineering,
Univ. of Rhode Island, Kingston, RI 02881 (corresponding author). critical state friction angle ( f 0c) and dilation angle (Bolton 1986).
E-mail: [email protected] The critical state friction angle varies by sand type (Cho et al.
2
Assistant Professor, Dept. of Civil and Environmental Engineering, 2006) and the relationship between the difference in peak and
Univ. of Rhode Island, Kingston, RI 02881. E-mail: [email protected] critical state friction angle and dilation angle also varies with
3
Civil Engineer, Ocean Facilities Dept., 1100 23rd Ave., NAVFAC
sand type (Simoni and Houlsby 2006; Liu and Lehane 2012b).
EXWC, Port Hueneme, CA 93043.
Note. This manuscript was submitted on December 11, 2015; approved
Therefore, changes in friction angle cannot uniquely capture the
on April 19, 2016; published online on June 8, 2016. Discussion period influence of the dilation angle on the uplift capacity of shallow
open until November 8, 2016; separate discussions must be submitted for circular anchors.
individual papers. This paper is part of the International Journal of Furthermore, experimental studies on shallow pipes by Cheuk et
Geomechanics, © ASCE, ISSN 1532-3641. al. (2007) and plate anchors by Liu et al. (2012a) suggest that the

© ASCE 04016056-1 Int. J. Geomech.

Int. J. Geomech., 04016056

Downloaded from ascelibrary.org by The State University of New York at Buffalo (SUNY - Buffalo) on 06/08/16. Copyright ASCE. For personal use only; all rights reserved.

Fig. 2. Experimental setup: (a) helical anchor installation and loading

frame; (b) installation of helical anchor; (c) installation of MCPT; and
(d) test anchors with B = 152 mm (left) and B = 254 mm (right) (images
by Joseph Giampa)

Koutsabeloulis and Griffiths 1989; White et al. 2008; Smith 2012)

has received more attention than for circular anchors. Links
between plane strain and axisymmetric analyses are presented in
Fig. 1. Problem geometry and shape of assumed failure mechanisms this paper, and consistency with those previous studies is reflected
typically assumed for analysis of drained loading of shallow anchors: in the analytical solution developed.
(a) anchor geometry; (b) failure mechanism assuming AF; and (c) fail-
ure mechanism assuming NAF
Experimental Program

inclination angle of the slip surface at failure seems to correspond to Twenty-two circular, intermediate-scale, single-helix anchor
the peak dilation angle of the soil. White et al. (2008) used the tests were performed at various embedment depths in dry sand.
assumption of a wedge of soil uplifted at the dilation angle to de- Additional details can be found in Giampa (2014). Three test
velop an analytical solution based on limit equilibrium for the trenches were prepared to different dry unit weights ( g ). Anchor
assessment of strip anchor uplift capacity. The solution takes the depth to diameter ratios (D/B) varied from 1.8 to 7.1. The frame
following form: shown in Fig. 2(a) was used for installation and load testing of the

 anchors [Fig. 2(b)], as well as for the miniature cone penetrometer
D tests (MCPTs) [Fig. 2(c)]. The frame was equipped with a
N g ¼ 1 þ Fps (2)
B mounted electric winch and hydraulic ram to control the installa-
tion or uplift of the anchors. The anchors were installed at a pene-
tration rate equal to the average pitch of the anchor multiplied by
Fps ¼ tan c p þ C1 ðtan f 0p  tan c p Þ (3)
the rotation rate. After a pause of approximately 5 minutes fol-
lowing installation, the anchor was pulled out at a rate ranging
where Fps = uplift factor for plane strain conditions; B = anchor from 1.70 to 3.80 mm/s. The 300-mm stroke hydraulic ram,
width or diameter; and C1 = constant for assessing the normal mounted to the same frame, was used for uplift tests in Trenches 1
stress on the failure plane. For the AF case, where c p = f 0p, Eq. 3 and 2, and the same electric winch used for installation control
simplifies to an UB solution, where Fps = tan f 0p (White et al. was used for uplift tests in Trench 3. Using the winch for uplift
2008). testing increased test efficiency and allowed the load to be meas-
White et al. (2008) proposed an expression of C1 = (1 þ K0)/2 – ured during the full anchor extraction.
(1 − K0)cos(2 c p)/2, where K0 is the lateral earth pressure coefficient
at rest. They assume that the normal stress on the failure surface
Test Anchor Details
does not change during uplift loading. However, based on argu-
ments presented in Smith (2012) this paper reassesses the constant Uplift tests were performed on two different diameter single-helix
for assessing normal stress on the failure plane and extends the for- anchors shown in Fig. 2(d). The diameters of the anchors were 152
mulation of White et al. (2008) to the axisymmetric case of circular and 254 mm, with a pitch at the location of the shaft of 22 and
anchors. 38 mm, respectively. The anchors had an untrue helix and thus the
The influence of dilation angle on plane strain plate anchor uplift pitch increased with radial distance from the shaft to a pitch
capacity (Rowe and Davis 1982; Vermeer and Sutjiadi 1985; of 48 mm at the perimeter of the 152-mm diameter helix and

© ASCE 04016056-2 Int. J. Geomech.

Int. J. Geomech., 04016056

Downloaded from ascelibrary.org by The State University of New York at Buffalo (SUNY - Buffalo) on 06/08/16. Copyright ASCE. For personal use only; all rights reserved.

Fig. 3. Relationship between peak friction and dilation angles for

Golden Flint sand measured in triaxial tests Fig. 5. Calibration of Bolton stress-dilatancy Q and R parameters

Table 1. Properties of Test Sand

Property Value
g max (kN/m )3
17.68
g min (kN/m3) 14.24
emax 0.847
emin 0.487
Gs 2.68
D50 (mm) 0.25
Cu 1.61
Cc 1.13
f 0 c (°) 33.9
b 0.64
Af 3.6
Q 9.5
R −0.68

Triaxial tests were performed on dry and saturated Golden

Flint sand at relative densities (Dr) between 0 and 0.80. Mean
Fig. 4. Assessment of Bolton Af parameter from triaxial tests effective stresses at failure (p0f) varied from 30 to 450 kPa. Stress-
dilatancy fitting parameters for estimating peak friction and dila-
tion angles from Dr and estimated p0f were based on recommenda-
76 mm at the perimeter of the 254-mm diameter helix. The shaft tions by Bolton (1986) and Salgado et al. (2000). Bolton (1986)
diameters were 35 and 44 mm for the 152- and 254-mm diameter empirically relates the peak dilation angle to the difference in
anchors, respectively. The thickness of the helix plate was 6 mm peak and critical state friction angles by a parameter b (i.e.,
for both anchors. f 0p − f 0c = b c p); Bolton further expresses that f 0p − f 0c =
Af(Dr[Q − ln(p0f)] − R). Fig. 3 plots f 0p against c p to estimate f 0c
Sand Properties and the b parameter, which correspond to the y-intercept and
slope of the line, respectively. The Af parameter is determined as
Each test trench was filled with fine uniformly graded Golden Flint the slope of the line shown in Fig. 4. Finally, to estimate Bolton
sand. Characterization of the sand included standard index testing (1986) fitting parameters Q and R in Fig. 5, relative density is
and 28 triaxial tests performed among three laboratories: University plotted against ( f 0p − f 0c)/Af þ Drln(p0f), where Q = slope of the
of Rhode Island (URI) (this study), University of Tennessee, line; and R = y-intercept. The results indicated that values of f 0c =
Knoxville (UTK) (Alshibli 2015), and California State University, 33.9° are typical for subangular silica sands, and values of Af =
Los Angeles (CSLA) (Tufenkjian and Yee 2006). Interpretations of 3.6 and b = 0.64 are in agreement with values discussed by
Bolton (1986) stress-dilatancy parameters are shown in Figs. 3–5. Chakraborty and Salgado (2010). Although the Q parameter of
Note that values of c p and volumetric strain (« v) were not measured 9.5 was near the typical value of 10 for silica sands (Bolton
for the CSLA tests; therefore they are not included in Figs. 3 and 4. 1986), negative R values, although atypical, have been previously
Triaxial test data from all laboratories are included in Fig. 5. reported by Salgado et al. (2000) and Chakraborty and Salgado
Measured sand properties are summarized in Table 1. (2010).

© ASCE 04016056-3 Int. J. Geomech.

Int. J. Geomech., 04016056

 Table 2. Summary of Local Soil Properties and Helical Anchor Test
Results

g
Test (kN/ f 0 pa c pa B D Qu
ID m3) (°) (°) (m) (m) D/B (N)
1a 14.9 42.2 12.9 0.254 0.785 3.1 3,943
1b 14.7 40.5 10.3 0.254 0.787 3.1 3,688
Downloaded from ascelibrary.org by The State University of New York at Buffalo (SUNY - Buffalo) on 06/08/16. Copyright ASCE. For personal use only; all rights reserved.

1c 14.8 41.3 11.6 0.254 0.762 3 3,625

1d 14.8 41.3 11.6 0.254 0.762 3 3,809
1i 14.7 40.5 10.3 0.254 0.775 3.1 3,837
1j 14.7 40.5 10.3 0.254 0.775 3.1 3,754
2a 15.7 48.5 22.8 0.254 0.737 2.9 6,583
2b 15.8 49.3 24.0 0.254 0.721 2.8 6,696
2c 15.8 49.2 23.8 0.254 0.775 3.1 7,471
2f 15.6 47.7 21.6 0.152 0.762 5 4,707
2g 15.6 47.7 21.6 0.152 0.756 5 4,540
2h 15.5 47.7 21.6 0.254 0.46 1.8 1,501
2i 15.6 48.6 22.9 0.152 0.448 2.9 1,575
2j 15.2 45.3 17.8 0.152 0.435 2.9 1,292
3a 15.5 46.4 19.6 0.152 1.08 7.1 7,303
3b 15.5 46.5 19.7 0.152 1.03 6.8 7,103
3d 15.4 46.2 19.2 0.152 0.756 5 4,499
3e 15.4 46.2 19.2 0.254 0.775 3.1 5,758
3f 15.3 45.4 18.0 0.254 0.762 3 5,778
3h 15.3 45.4 17.9 0.254 0.781 3.1 5,771
3i 15.3 46.0 18.9 0.254 0.488 1.9 2,020
Fig. 6. Sand pluviator used for test trench preparation (image by 3j 15.3 46.1 19.1 0.152 0.438 2.9 1,056
Joseph Giampa)
The dry unit weight with depth was back-calculated at each
MCPT location using the maximum and minimum unit weights of
Trench Preparation the Golden Flint sand (Table 1). A median dry unit weight was then
Testing was conducted in a reinforced concrete trench with dimen- taken at each MCPT location from the surface to a corresponding
sions 1.5  21  1.5 m, filled with 60 t of fine uniformly graded anchor depth. Last, by using the median MCPT dry unit weights on
Golden Flint sand. The long trench allowed for multiple anchor either side of an anchor test, a second median was calculated to
uplift tests and cone penetration tests to be performed in a single determine the dry unit weight at the exact anchor location, which is
sand sample. A customized sand-dispensing system (Tufenkjian summarized in Table 2. Profiles of MCPT tip resistance (qc) and
and Yee 2006) was used to air pluviate the sand into the test trench, inferred dry unit weight for Trenches 1, 2, and 3 are shown in Figs. 7
as shown in Fig. 6. and 8, respectively. In Fig. 7, the gray lines indicate the range of dry
The pluviator was mounted on top of the trench on which it unit weight estimated from sand-specific, fall height–hopper open-
could then traverse the full length and deposit the sand in a con- ing size–Dr correlations, and the dark black lines indicate the dry
trolled manner. As the pluviator proceeds down the trench, the sand unit weight estimated from the sand-specific MCPT qc − Dr correlation.
is sieved, dispensed through an adjustable opening, falls, and con- Generally, MCPT profiles are relatively uniform across the test
tacts a rotating distribution drum, from which it flows to a dispens- trenches. Trench 1, in which sand placement methods were being
ing tray that drops it into the trench from a controlled height. Dry refined, exhibits a slightly higher variability in sand density across the
unit weight was controlled by the flow rate of the sand, angle of the trench. The nonhomogeneity in the dry unit weight in the upper
dispensing tray, and fall height. Calibrations relating hopper param- 300 mm for Trench 2 resulted from the maximum height of the dispens-
eters to dry unit weight were performed, such that a target dry unit ing tray above the sand trench being reached. A shorter specimen was
weight could be achieved within each trench. Fall height was meas- prepared for Trench 3 to minimize this effect. Table 2 summarizes the
ured periodically during sand placement and used to estimate the local soil properties and results from each helical anchor uplift test.
range of in-place dry unit weight.
Analytical and Numerical Solutions
Properties of Test Trenches
The uniformity of the sand in each test trench was assessed by Limit Equilibrium Analytical Solution
MCPTs with a 2-cm2 area, pushed between each anchor location. To assess N g for circular anchors, a new axisymmetric (as) uplift
Vertical and horizontal variability were noted, and localized dry unit capacity model was derived for circular helical anchors using non-
weights for each anchor test were evaluated. The dry unit weights associated flow (NAF) limit equilibrium analysis adapted from the
for each anchor test were estimated from a sand-specific MCPT White et al. (2008) plane strain (ps) solution.
qc − Dr (tip resistance-relative density) correlation (J. A. Schneider, To aid in the assessment of the constant C1, interpretations from
J. R. Giampa, A. S. Bradshaw, and J. T. Newgard, “Calibrating CPT previous analyses on strip anchors in a NAF Coulomb soil (Rowe
relative density and strength correlations for a laboratory fine sand,” and Davis 1982; Vermeer and Sutjiadi 1985; Koutsabeloulis and
submitted, Balkema, Rotterdam, Netherlands). Griffiths 1989; White et al. 2008; Smith 2012) is required. Main

© ASCE 04016056-4 Int. J. Geomech.

Int. J. Geomech., 04016056

 For dilation angles between 10° and f 0p, using Eq. 4 produces
Fps within 4% of factors in Smith (2012) for a peak friction angle up
to 50° and within 7% of Fps factors in Smith (2012) for peak friction
angles up to 60°.
The following derivation extends the plane strain formulation to
that of a circular plate, primarily to allow for calculations in which
the dilation angle does not equal the friction angle. The primary
assumption is that the failure surface is represented as a truncated
Downloaded from ascelibrary.org by The State University of New York at Buffalo (SUNY - Buffalo) on 06/08/16. Copyright ASCE. For personal use only; all rights reserved.

cone above the anchor, in which the inclination angle (u ) is equal to

the peak dilation angle of the soil [Fig. 1(c)]. The uplift capacity is
then a combination of the weight of the soil wedge plus the resist-
ance along the failure surface. The weight of the soil wedge (Ww) is
defined by
"
2 #
p B B 
Ww ¼ g D þ D tan c p þ B þ D tan c p (5)
3 2 2

Consistent with the assumptions made by White et al. (2008),

only frictional energy dissipation is considered on the failure sur-
Fig. 7. MCPT qc profiles face. Thus, the friction coefficient equals tan f 0p – tan c p, and the
peak-mobilized shear stress (t f) can be defined by

t f ¼ C1 g Dðtan f 0p  tan c p Þ (6)

Integrating Eq. (6) along the failure surface and combining with
Eq. (5), the peak uplift resistance can be calculated as
"
2 #
p B B  p
Qu ¼ g D þ D tan c p þ B þ D tan c p þ g D2 C1
3 2 2 2


2
ðtan f 0p  tan c p Þ B þ D tan c p (7)
3

Thus, the breakout factor for circular anchors can be simplified

as


2
D D
N g ¼ 1 þ Fas1 þ Fas2 (8)
B B

Fas1 ¼ 2½tan c p þ C1 ðtan f 0p  tan c p Þ ¼ 2Fps (9)

Fig. 8. Average dry unit weight profiles inferred from MCPT data and
fall height measurements 4 h  i 4
Fas2 ¼ tan c p tan c p þ C1 tan f 0p  tan c p ¼ Fps tan c p
3 3
(10)
relevant conclusions from the previously mentioned plane strain
strip anchor uplift analyses include the following:
 Initial earth pressure coefficient at rest (K0) has a minor influ- where Fas1 and Fas2 are uplift factors for axisymmetric conditions.
ence on uplift capacity (Rowe and Davis 1982). As with the plane strain solution discussed by White et al.
 When c p = 0, Fps  sin f 0p, i.e., C1 = cos f 0p (Rowe and (2008), the combination of Eq. (8) with Eqs. (9) and (10), become
Davis 1982; Vermeer and Sutjiadi 1985; Koutsabeloulis and the UB solution when assuming AF. For the UB solution, Fps = tan
Griffiths 1989). f 0p, Fas1 = 2 tan f 0p, and Fas2 = 4/3 tan2 f 0p, which is in agreement
 When c p = f 0p, Fps = tan f 0p (Rowe and Davis 1982; Vermeer with a reorganization of the UB solution of Murray and Geddes
and Sutjiadi 1985; Murray and Geddes 1987; Smith 1998; (1987).
Koutsabeloulis and Griffiths 1989; White et al. 2008; Smith
2012). Numerical Analysis
Based on a review of these previous studies, a first-order approx- A series of 72 axisymmetric numerical analyses were performed
imation of C1 for this study is proposed that differs from White using the software OptumG2 (Optum CE 2014a, b) to assess the
et al. (2008). Because C1 is approximately cos f 0p for c p = 0, and influence of dilation angle on anchor uplift capacity. Numerical
C1 can be unity for c p = f 0p, thus C1 is taken as models were set up to bound the soil and anchor conditions from the
experimental study, as summarized in Table 3. Initially, UB and LB
C1 ¼ cos ð f 0p  c p Þ (4)
numerical limit analyses were used as part of the numerical

© ASCE 04016056-5 Int. J. Geomech.

Int. J. Geomech., 04016056

 Table 3. Summary of Numerical Analyses

Analytical Numerical
0
Trench B (m) D (m) D/B f p (°) c p (°) UB Eq. (8) UB LB AF NAF
T1 0.152 0.457 3 40 9.5 14.5 7.0 14.7 14.4 14.3 7.5
T1 0.152 0.760 5 40 9.5 32.9 12.6 33.2 32.6 32.8 15.4
T1 0.152 1.06 7 40 9.5 58.7 19.6 58.6 57.5 57.1 29.6
Downloaded from ascelibrary.org by The State University of New York at Buffalo (SUNY - Buffalo) on 06/08/16. Copyright ASCE. For personal use only; all rights reserved.

T1 0.254 0.457 1.8 40 9.5 7.1 4.2 7.1 7.0 6.8 3.5
T1 0.254 0.760 3 40 9.5 14.5 7.0 14.6 14.3 14.2 6.8
T1 0.254 1.06 4.2 40 9.5 24.6 10.2 24.5 24.1 24.0 11.6
T2 0.152 0.457 3 50 25.0 25.2 14.0 25.7 25.0 24.2 12.6
T2 0.152 0.760 5 50 25.0 60.3 29.7 60.9 59.3 59.3 27.4
T2 0.152 1.06 7 50 25.0 110.5 51.0 110.2 108.0 105.4 46.1
T2 0.254 0.457 1.8 50 25.0 11.4 7.3 11.6 11.3 10.4 5.7
T2 0.254 0.760 3 50 25.0 25.2 14.0 25.2 24.8 24.0 12.8
T2 0.254 1.06 4.2 50 25.0 44.4 22.8 43.8 43.1 42.8 19.6
T3 0.152 0.457 3 45 17.5 19.0 10.0 19.4 18.9 18.5 9.8
T3 0.152 0.760 5 45 17.5 44.3 19.9 43.9 43.9 44.1 19.5
T3 0.152 1.06 7 45 17.5 80.3 32.9 80.0 78.6 77.5 38.0
T3 0.254 0.457 1.8 45 17.5 8.9 5.6 9.0 8.9 8.6 5.0
T3 0.254 0.760 3 45 17.5 19.0 10.0 19.1 18.7 18.5 9.6
T3 0.254 1.06 4.2 45 17.5 32.9 15.6 32.7 32.0 31.9 15.9

validation study. These were followed by elastoplastic small dis- Numerical limit analysis was performed using LB and UB ele-
placement finite-element analyses (FEAs) assuming AF, and finally ments. The LB elements are triangles with a linear variation in stress
by analyses assuming NAF conditions. between the corner nodes (Optum CE 2014a). The UB elements are
OptumG2 solves problems in terms of variational principles, triangles with quadratic interpolation of displacements and linear
in which the governing equations are formed as an optimization interpolation of stresses (Optum CE 2014a). The FEAs were per-
problem that is maximized or minimized while being con- formed using 15-node Gauss triangles with quartic interpolation of
strained by boundary conditions, such as external supports and displacements and cubic interpolation of stresses (Optum CE
those imposed by the stiffness matrix and flow rule (Optum CE 2014a). Discontinuities between all element types are handled
2014b). An iterative solution of the series of equations is per- numerically using a zero thickness element to join adjacent ele-
formed using SONIC, which is a second-order cone program- ments (Krabbenhoft et al. 2005; Lyamin et al. 2005a, 2011; Optum
ming algorithm within OptumG2 based on principles similar to CE 2014a).
MOSEK (Andersen et al. 2003). Solution methods for limit anal- All anchors were modeled in an area with a radius of 2 m and
ysis are similar to those discussed by Lyamin (1999), Lyamin depth of 1.4 m. It is noted that for numerical analysis of the anchor
and Sloan (2002a, b), Krabbenhoft and Damkilde (2003), and uplift problem, localization will tend to occur at a boundary
Krabbenhoft et al. (2008). For elastoplastic FEAs, OptumG2 between the anchor/uplifted soil and the surrounding soil
uses solution methods similar to those described by Krabbenhoft (Krabbenhoft et al. 2012). These localizations may result in the nu-
et al. (2007a, b, 2012), and Krabbenhoft (2009). merical solution being mesh dependent. To minimize effects of
To assess collapse loads, a unit vertical uplift stress was mesh geometry on analysis results, numerical models within this
applied to the anchor plate, which was constrained from horizon- paper used the adaptive meshing strategies based on shear dissipa-
tal translation and rotation. Multiplier loads are amplified within tion within OptumG2 (Optum CE 2014a). Similar adaptive meshing
each step until collapse occurs. This procedure is used within schemes are discussed by Lyamin et al. (2004, 2005b). Example
OptumG2 for limit analysis and is termed the multiplier elasto- meshes are shown in Figs. 10(a and b); however, it is noted that
plastic analysis for small displacement FEAs (Optum CE only the first meter of the mesh radius is shown for illustrative
considerations.
2014a). The automatic load stepping scheme within OptumG2
was used for small displacement FEAs using 10 elastic and 10
plastic steps. Collapse loads were selected based on the average Material Properties
collapse multiplier at the end of the analysis, as indicated in Rather than model each specific anchor test, numerical studies were
Figs. 9(a and b). A clear ultimate load is apparent for AF analy- performed in a manner to assess if similar conclusions evolved
ses [Fig. 9(a)]; however, some oscillations are observed for from both numerical and experimental studies. An advantage of nu-
anchor uplift using a NAF soil model [Fig. 9(b)], as also shown merical analysis is that the material input parameters are known,
by Loukidis and Salgado (2009) and Krabbenhoft et al. (2012), whereas material parameters need to be estimated for interpreting
among others. These oscillations are not necessarily a numerical experimental studies. Material properties that were constant with
artifact but a consequence of the NAF problem not having a depth were used in analyses for direct comparison with analytical
unique solution and the numerical solution switching between solutions. Three sets of material parameters were selected for analy-
slightly different failure modes (Krabbenhoft et al. 2012; Optum ses: those generally representative of Trench 1, Trench 2, and
CE 2014a). Although it is noted that some judgment is required Trench 3.
when selecting collapse loads for NAF numerical analyses Soil conditions were modeled as an isotropic linear elastoplastic
(Sloan 2013), Fig. 9(b) indicates that bounds should be within material failing according to a Mohr-Coulomb failure criterion with
610% in these cases. peak friction and dilation angles constant with depth. Peak friction

© ASCE 04016056-6 Int. J. Geomech.

Int. J. Geomech., 04016056

 equilibrium analytical solution, which can account for the dilation
angle not equaling friction angle.

Numerical Model Validation and Discussion

Main observations from the numerical studies indicate the
following:
 Limit analyses, shown in Table 3, are in agreement with
Downloaded from ascelibrary.org by The State University of New York at Buffalo (SUNY - Buffalo) on 06/08/16. Copyright ASCE. For personal use only; all rights reserved.

Merifield et al. (2006) and considered validated in that the

8 Numerical UB is within 1% of the analytical UB;
8 Numerical LB is 1–3% less than numerical UB; and
8 Limit analysis based on the assumption of AF shows an
essentially unique solution with equal upper and lower
bounds.
 The elastoplastic FEAs are considered validated because the
analysis assuming AF are within 4% of the analytical UB and
have a coefficient of variation (COV) in results of 0.02 (Tables
3 and 4).
 The shape of the failure mechanism is indicated by concentra-
tions of shear strain shown in Fig. 11. Only the first meter of
the 2-m radius model area is shown for illustrative purposes.
In agreement with the primary assumption of the analytical so-
lution, the shape of the failure mechanism is a truncated cone
at the angle of dilation for both AF and NAF analyses.
 The NAF analysis results in uplift resistance that is approxi-
mately one-half of the analytical UB (Table 3); however, the
ratio is not unique and varies with friction angle and depth.
 The median ratio of N g from the NAF truncated cone analyti-
cal solution [Eqs. (8)–(10)] to the NAF numerical analysis was
1.03, assuming C1 = cos ( f 0p − c p) (Eq. 4). The COV in that
ratio was 0.15. As the dilation angle approached zero, the
COVs in the analyses increased and median bias decreased.
Numerical analyses support the backing assumption of the analyti-
cal solution that the uplift wedge occurs at an angle approximately
equal to the dilation angle of the soil. The analyses also support the
use of Eq. (4) to estimate the normal stress on the failure plane.

Comparison of Analytical Expressions with

Fig. 9. Normalized load displacement response for FEAs assuming (a) Experimental Results
AF conditions and (b) NAF conditions
Table 5 shows the tabulated results comparing the UB and NAF an-
alytical solutions to the experimental data. Statistical assessment for
and dilation angles are presented in Table 3 and cover dilation each trench is shown in Table 6. The UB analytical solution
angles from approximately 10 to 25° for NAF analyses and 40 to assumes an AF rule, such that the peak dilation angle is equal to the
50° when considering AF (limit analysis and AF finite-element peak friction angle. Fig. 12(a) shows the N g ratio of predicted to
modeling). The dry unit weight was constant with depth and varied measured that overpredicts the experimental results by a factor of 2
slightly with each trench approximating Fig. 8: Trench 1 = or greater with a median bias of 2.29. This bias is consistent with
14.75 kN/m3, Trench 2 = 15.6 kN/m3, and Trench 3 = 15.35 kN/m3. the numerical results comparing the analytical UB with the NAF
Elastic modulus (E) was constant with depth and varied with each FEAs that resulted in a median bias of 2.05. Capacity predictions
trench: Trench 1 = 3 MPa, Trench 2 = 12 MPa, and Trench 3 = from the UB analytical solution become slightly better with increas-
6 MPa. Poisson ratio () of 0.25 and K0 of 0.45 was used for all cases. ing f 0p for D/B ranging from 2 to 5. Therefore, the overpredictions
Anchors were included as a weightless rigid plate at depth with a could be explained by the underlying assumption that c p = f 0p,
weightless rigid shaft extending to the surface. The anchor plate had which overestimates the size of the failure wedge.
a rough interface, whereas the anchor shaft had a completely Conversely, Fig. 12(b) shows the N g ratio of predicted to meas-
smooth interface. Since the shaft is included within the failure ured using the analytical solution assuming NAF ( c p < f 0p). The
mechanism for the shallow anchor problem, the shaft roughness, N g of predicted to measured decreases with an increasing peak di-
assessed through parametric studies, had no effect on uplift lation angle. The NAF analytical solution predicts the experimental
capacity. capacity with a median bias of 1.15 and COV of 0.14. The bias
It is noted that further numerical analysis of anchor response in relating the analytical solution and the experimental results is com-
layered or spatially variable soils is warranted as well as analyses parable to that found in the numerical results with a median bias of
with more detailed constitutive models (Sakai and Tanaka 1998). 1.03 and a COV of 0.15. The closer approximation of uplift capacity
These additional analyses are beyond the scope of this paper, as the resulting from the NAF analytical solution is considered primarily
primary purpose is to assess the performance of the derived limit to be explained by the inclination angle of the failure wedge being

© ASCE 04016056-7 Int. J. Geomech.

Int. J. Geomech., 04016056

Downloaded from ascelibrary.org by The State University of New York at Buffalo (SUNY - Buffalo) on 06/08/16. Copyright ASCE. For personal use only; all rights reserved.

Fig. 10. Adaptive meshes at failure for f 0p = 50°, c p = 25°, B = 254 mm, D = 760 mm: (a) AF FEAs (1,250 elements); (b) NAF FEAs (1,250
elements)

Table 4. Statistical Assessment of Analytical Expressions Based on AF plane strain limit equilibrium solution to drained anchor uplift, (2)
and NAF Numerical Analyses extension of the model to axisymmetric conditions, (3) comparison
of analytical expressions to numerical analysis for AF and NAF
Eqs. (8)–(10)/
UB/AF FEAs UB/NAF FEAs NAF FEAs conditions, and (4) comparison of analytical expressions to experi-
mental studies for AF and NAF conditions. Developing conclusions
Trench Median COV Median COV Median COV by comparing both numerical and experimental studies with the
T1 1.02 0.01 2.07 0.04 0.90 0.20 same simplified expressions was intended to minimize limitations
T2 1.04 0.02 2.10 0.08 1.11 0.06 of both validation techniques; i.e., performing solely numerical
T3 1.03 0.01 2.02 0.08 1.02 0.08 analysis tends to oversimplify soil response, and performing solely
All 1.03 0.02 2.05 0.07 1.03 0.15 experimental studies tends to add uncertainty in validation through
selection of material parameters as well as significance of installa-
tion disturbance.
equal to the peak dilation angle, thus better capturing the effects of The primary conclusions of this study indicate the following:
dilation during uplift. Although this is a relatively small difference,  There is an updated stress coefficient for the White et al.
it may highlight the influence of factors ignored in both the numeri- (2008) plane strain anchor solution that better matches other
cal and analytical studies, particularly, depth-dependent strain lev- numerical analyses on the drained shallow plane strain anchor
els for reaching f 0p and postpeak reduction in dilation and friction problem in nonassociated soils within about 10%.
angles.  Extension of the plane strain analysis to axisymmetric condi-
tions adds a second term that is a function of (D/B)2. Because
of this term, UB solutions that rely on the assumption of AF
Conclusions tend to overpredict circular anchor response to a greater
degree, over a factor of 2 in most cases, than for the plane
The objective of this study was to develop and validate an analytical strain case.
expression that will accurately predict the drained uplift capacity of  Although FEAs of soils with an AF rule were within 4% of the
shallow circular helical anchors. During drained loading the friction analytical solution, having a COV of 0.02, larger variation was
and dilation angles are not equal; therefore rigorous UB and LB sol- observed for FEAs with a NAF rule. Median values for the
utions overpredict capacity. Model development and validation nonassociated case were equal to the analytical solution, but
involved (1) reassessment of normal stress coefficients within a the COV increased to 0.15. Although the NAF numerical

© ASCE 04016056-8 Int. J. Geomech.

Int. J. Geomech., 04016056

Downloaded from ascelibrary.org by The State University of New York at Buffalo (SUNY - Buffalo) on 06/08/16. Copyright ASCE. For personal use only; all rights reserved.

Fig. 11. Shear strain contours for case with f 0 p = 50°, c p = 25°, B = 254 mm, D = 760 mm: (a) AF FEAs; (b) NAF FEAs

Table 5. Comparison of Experimental Results to Analytical Expressions Table 6. Statistical Assessment of Analytical Expressions Based on
Experimental Data
Test Experiment UB/ Eq. (8)/
ID Ng UB experiment Eq. (8) experiment UB Eqs. (8)–(10)
1a 6.7 17.1 2.55 8.5 1.27 Trench Median COV Median COV
1b 6.3 15.6 2.48 7.5 1.19 T1 2.42 0.03 1.17 0.04
1c 6.4 15.6 2.43 7.7 1.20 T2 2.02 0.15 1.11 0.17
1d 6.7 15.6 2.32 7.7 1.15 T3 2.24 0.28 1.18 0.16
1i 6.6 15.6 2.37 7.5 1.14 All 2.29 0.21 1.15 0.14
1j 6.5 15.6 2.40 7.5 1.15
2a 11.2 21.9 1.95 12.2 1.09
2b 11.6 21.6 1.86 12.2 1.05
analyses provide some validation of the analytical solution,
2c 12.1 25.3 2.09 13.9 1.15
particularly when compared with the UB results, it is acknowl-
2f 21.7 52.3 2.41 24.8 1.14
edged that more judgment is required to assess collapse loads
2g 21.1 52.3 2.48 24.8 1.18
for NAF FEAs compared with the AF case.
2h 4.2 10.2 2.42 6.5 1.54  On the experimental side, estimation of the peak dilation
2i 12.4 22.0 1.77 12.2 0.98
angle, and to a lesser degree peak friction angle, proved to be
2j 10.7 18.3 1.71 9.8 0.91
highly uncertain without sufficient laboratory testing. A series
3a 23.9 90.2 3.77 38.2 1.60
of triaxial tests to determine sand-specific Bolton parameters
3b 24.4 83.8 3.43 35.9 1.47
was needed. It was necessary to reassess not only Q and R, but
3d 21.1 47.7 2.26 21.9 1.04
also Af and b .
3e 9.5 21.4 2.25 11.3 1.19
 Comparison of analytical expressions with both experimental
3f 9.8 19.4 1.98 10.3 1.05
3h 9.6 20.4 2.13 10.7 1.11
and numerical analysis resulted in the same conclusions; the
3i 5.3 10.1 1.90 6.2 1.17
assumption of AF tends to overpredict anchor uplift capacity
3j 8.6 19.2 2.23 10.3 1.20
by over a factor of 2, with bias that was friction angle and
depth dependent. The newly developed analytical solution
agrees with the NAF numerical analyses as well as experimen-
tal studies within approximately 15%. The COVs, when

© ASCE 04016056-9 Int. J. Geomech.

Int. J. Geomech., 04016056

 University of Western Australia, Centre for Offshore Foundation
Systems. Last, the authors thank Marcus Rasulo for the useful
comments that improved this paper.

References

Alshibli, K. A. (2015). “Stress-dilatancy response of fine golden flint sand.”

Downloaded from ascelibrary.org by The State University of New York at Buffalo (SUNY - Buffalo) on 06/08/16. Copyright ASCE. For personal use only; all rights reserved.

Rep. Prepared for Ocean Engineering Div., NAVFAC, Port Hueneme,

CA.
Andersen, E. D., Roos, C., and Terlaky, T. (2003). “On implementing a pri-
mal-dual interior-point method for conic quadratic optimization.” Math.
Program., 95(2), 249–277.
Bolton, M. (1986). “The strength and dilatancy of sands.” Geotechnique,
36(1), 65–78.
Byrne, B. W., and Houlsby, G. T. (2015). “Helical piles: An innovative
foundation design option for offshore wind turbines.” Philos. Trans. R.
Soc. London, Ser. A, 373(2035), 20140081.
Chakraborty, T., and Salgado, R. (2010). “Dilatancy and shear strength of
sand at low confining pressures.” J. Geotech. Geoenviron. Eng., 10
.1061/(ASCE)GT.1943-5606.0000237, 527–532.
Cheuk, C., White, D., and Bolton, M. (2007). “Uplift mechanisms of pipes
buried in sand.” J. Geotech. Geoenviron. Eng., 10.1061/(ASCE)1090
-0241(2008)134:2(154), 154–163.
Cho, G., Dodds, J., and Santamarina, J. (2006). “Particle shape effects on
packing density, stiffness, and strength: natural and crushed sands.” J.
Geotech. Geoenviron. Eng, 10.1061/(ASCE)1090-0241(2006)132:
5(591), 591–602.
Davis, E. H. (1968). “Theories of plasticity and the failure of soil masses.”
Soil mechanics: selected topics, I. K. Lee, ed., Butterworths, London,
341–380.
Drescher, A., and Detournay, E. (1993). “Limit load in translational failure
mechanisms for associative and non-associative materials.” Geotechnique,
43(3), 443–456.
Ghaly, A., and Hanna, A. (1994). “Ultimate pullout resistance of single ver-
tical anchors.” Can. Geotech. J., 31(5), 661–672.
Giampa, J. R. (2014). “Interpretation of shallow helical anchor capacity in
sand.” M.S. thesis, Univ. of Rhode Island, Kingston, RI.
Ilamparuthi, K., Dickin, E. A., and Muthukrisnaiah, K. (2002).
“Experimental investigation of the uplift behavior of circular plate
anchors embedded in sand.” Can. Geotech. J., 39(3), 648–664.
Koutsabeloulis, N. C., and Griffiths, D. V. (1989). “Numerical modelling of
the trap door problem.” Geotechnique, 39(1), 77–89.
Krabbenhoft, K. (2009). “A variational principle of elastoplasticity and its
application to the modeling of frictional materials.” Int. J. Solids Struct.,
Fig. 12. Predicted to measured uplift capacity using (a) UB analytical 46(3–4), 464–479.
solution and (b) NAF analytical solution Krabbenhoft, K., and Damkilde, L. (2003). “A general non-linear optimiza-
tion algorithm for lower bound limit analysis.” Int. J. Numer. Methods
Eng., 56(2), 165–184.
comparing the newly developed analytical solution with nu- Krabbenhoft, K., Karim, M. R., Lyamin, A. V., and Sloan, S. W. (2012).
“Associated computational plasticity schemes for nonassociated fric-
merical analyses using a NAF rule and with experimental stud-
tional materials.” Int. J. Numer. Methods Eng., 90(9), 1089–1117.
ies, were 0.15 and 0.14, respectively. Krabbenhoft, K., Lyamin, A. V., Hjiaj, M., and Sloan, S. W. (2005). “A
Although further assessment in different sand types and soil new discontinuous upper bound limit analysis formulation.” Int. J.
layering configurations is warranted, as well as numerical studies Numer. Methods Eng., 63(7), 1069–1088.
with more complex soil constitutive models, the agreement in con- Krabbenhoft, K., Lyamin, A. V., and Sloan, S. W. (2007a). “An interior-
sistency among the numerical, analytical, and experimental results point method for elastoplasticity.” Int. J. Numer. Methods Eng., 69,
gives confidence in the further application of the NAF limit equilib- 592–626.
rium analytical solution presented in this paper. Krabbenhoft, K., Lyamin, A. V., and Sloan, S. W. (2007b). “Formation and
solution of some plasticity problems as conic programs.” Int. J. Solids
Struct., 44(5), 1533–1549.
Acknowledgments Krabbenhoft, K., Lyamin, A. V., and Sloan, S. W. (2008). “Three-
dimensional Mohr-Coulomb limit analysis using semi-definite pro-
gramming.” Commun. Numer. Methods Eng., 24(11), 1107–1119.
The authors acknowledge Optum CE for an advanced version of
Liu, J., Liu, M., and Zhu, Z. (2012a). “Sand deformation around an uplift
Optum G2 that includes axisymmetric analyses. The first and plate anchor.” J. Geotech. Geoenviron. Eng., 10.1061/(ASCE)GT.1943
second authors acknowledge that this material is based upon work -5606.0000633, 728–737.
partially supported by the National Science Foundation under Grant Liu, Q. B., and Lehane, B. M. (2012b). “The influence of particle shape on
No. 1300142. The third author acknowledges support for work on the (centrifuge) cone penetration test (CPT) end resistance in uniformly
this paper through a Martin Fahey Visiting Fellowship at the graded granular soils.” Geotechnique, 62(11), 973–984.

© ASCE 04016056-10 Int. J. Geomech.

Int. J. Geomech., 04016056

 Loukidis, D., Chakraborty, T., and Salgado, R. (2008). “Bearing capacity of Optum CE. (2014a). Optum G2, analysis, Optum Computational
strip footings on purely frictional soil under eccentric and inclined Engineering, Newcastle, New South Wales, Australia.
loads.” Can. Geotech. J., 45(6), 768–787. Optum CE. (2014b). Optum G2, theory, Optum Computational
Loukidis, D., and Salgado, R. (2009). “Bearing capacity of strip and circular Engineering, Newcastle, New South Wales, Australia.
footings in sand using finite elements.” Comput. Geotech., 36(5), 871–879. Rowe, R. K., and Davis, E. (1982). “The behaviour of anchor plates in
Lutenegger, A. J. (2011). “Historical development of iron screw-pile foun- sand.” Geotechnique, 32(1), 25–41.
dations: 1836–1900.” Int. J. Hist. Eng. Technol., 81(1), 108–128. Sakai, T., and Tanaka, T. (1998). “Scale effect of a shallow circular anchor
Lyamin, A. V. (1999). “Three-dimensional lower bound limit analysis using in dense sand.” Soils Found., 38(2), 93–99.
Salgado, R., Bandini, P., and Karim, A. (2000). “Shear strength and stiff-
Downloaded from ascelibrary.org by The State University of New York at Buffalo (SUNY - Buffalo) on 06/08/16. Copyright ASCE. For personal use only; all rights reserved.

nonlinear programming.” Ph.D. thesis, Univ. of Newcastle, New Castle,

Australia. ness of silty sand.” J. Geotech. Geoenviron. Eng., 10.1061/(ASCE)1090
Lyamin, A. V., Krabbenhoft, K., Abbo, A., and Sloan, S. W. (2005a). -0241(2000)126:5(451), 451–462.
“General approach to modelling discontinuities in limit analysis.” Sarac, D. Z. (1989). “The uplift capacity of shallow buried anchor slabs.”
Proc., IACMAG 11, Politecnico di Torino, Turin, Italy. Proc., 12th Int. Conf. on Soil Mechanics and Foundation Engineering,
Lyamin, A. V., Krabbenhoft, K., and Sloan, S. (2011). “Interface modelling Vol. 2, Balkema, Rotterdam, Netherlands, 1213–1218.
in computational limit analysis.” Recent Dev. Innovative Appl. Comput. Simoni, A., and Houlsby, G. T. (2006). The direct shear strength and
Mech., 321–330. dilatancy of sand-gravel mixtures.” Geotech. Geol. Eng., 24(3),
Lyamin, A. V., Krabbenhoft, K., Sloan, S. W., and Hjiaj, M. (2004). “An 523–549.
adaptive algorithm for upper bound limit analysis using discontinuous Sloan, S. W. (2013). “Geotechnical stability analysis.” Geotechnique,
velocity fields.” Proc., 4th European Congress on Computational 63(7), 531–572.
Methods in Applied Sciences and Engineering, ECCOMAS, P. Smith, C. (1998). “Limit loads for an anchor/trapdoor embedded in an asso-
Neittaanmaki, T. Rossi, K. Majava, and O. Pironneau, eds., Jyvaskyla, ciative Coulomb soil.” Int. J. Numer. Anal. Methods Geomech., 22(11),
Finland, 1–11. 855–865.
Lyamin, A. V., and Sloan, S. W. (2002a). “Lower bound limit analysis using Smith, C. C. (2012). “Limit loads for a shallow anchor/trapdoor embedded
non-linear programming.” Int. J. Numer. Methods Eng., 55, 573–611. in a non-associative Coulomb soil.” Geotechnique, 62(7), 563–571.
Lyamin, A. V., and Sloan, S. W. (2002b). “Upper bound limit analysis using Tufenkjian, M., and Yee, E. (2006). “Soil friction angle and relative density
linear finite elements and non-linear programming.” Int. J. Numer. Anal. of sand from minicone penetration tests at shallow depth.” Rep. pre-
Methods Geomech., 26(2), 181–216. pared for Ocean Engineering Div., Naval Facilities Engineering Service
Lyamin, A. V., Sloan, S. W., Krabbenhoft, K., and Hjiaj, M. (2005b). Center, Port Hueneme, CA.
“Lower bound limit analysis with adaptive remeshing.” Int. J. Numer. Vermeer, P. A., and Sutjiadi, W. (1985). “The uplift resistance of shallow
Anal. Methods Geomech., 63, 1961–1974. embedded anchors.” Proc., 11th Int. Conf. on Soil Mechanics and
Merifield, R., Lyamin, A., and Sloan, S. (2006). “Three-dimensional lower- Foundation Engineering, Vol. 3, Balkema, Rotterdam, Netherlands,
bound solutions for the stability of plate anchors in sand.” 1635–1638.
Geotechnique, 56(2), 123–132. Vesic, A. S. (1971). “Breakout resistance of objects embedded in ocean bot-
Meyerhof, G., and Adams, J. (1968). “The ultimate uplift capacity of foun- tom.” Soil Mech. Found. Eng. Div., 97(9), 1183–1205.
dations.” Can. Geotech. J., 5(4), 225–244. White, D. J., Cheuk, C. Y., and Bolton, M. (2008). “The uplift resistance of
Murray, E., and Geddes, J. (1987). “Uplift of anchor plates in sand.” J. pipes and plate anchors buried in sand.” Geotechnique, 58(10),
Geotech Engrg., 10.1061/(ASCE)0733-9410(1987)113:3(202), 202–215. 771–779.

© ASCE 04016056-11 Int. J. Geomech.

Int. J. Geomech., 04016056