Download as pdf or txt
Download as pdf or txt
You are on page 1of 5

Technical Note

Upper Bound Solution for Pullout Capacity of Vertical


Anchors in Sand Using Finite Elements and Limit Analysis
Jyant Kumar1 and Jagdish Prasad Sahoo2
Downloaded from ascelibrary.org by University of California, San Diego on 05/20/13. Copyright ASCE. For personal use only; all rights reserved.

Abstract: The horizontal pullout capacity of vertical anchors embedded in sand has been determined by using an upper bound theorem of the
limit analysis in combination with finite elements. The numerical results are presented in nondimensional form to determine the pullout
resistance for various combinations of embedment ratio of the anchor (H∕B), internal friction angle (ϕ) of sand, and the anchor-soil interface
friction angle (δ). The pullout resistance increases with increases in the values of embedment ratio, friction angle of sand and anchor-soil
interface friction angle. As compared to earlier reported solutions in literature, the present solution provides a better upper bound on the
ultimate collapse load. DOI: 10.1061/(ASCE)GM.1943-5622.0000160. © 2012 American Society of Civil Engineers.
CE Database subject headings: Anchors; Failure loads; Limit analysis; Plasticity; Sand (soil type); Finite element method; Pullout.
Author keywords: Anchors; Failure load; Limit analysis; Plasticity; Sand.

Introduction anchor, friction angle of the soil, and anchor-soil interface friction
angle were examined. The results obtained from the analysis were
Vertical plate anchors are often used for generating passive support compared with those from the literature.
to retaining walls, sheet piles, and bulkheads. A number of inves-
tigations have been performed by different researchers to evaluate
the ultimate pullout capacity of vertical plate anchors. Meyerhof Definition of the Problem
(1973) has used the limit equilibrium method in which the failure
mechanism was assumed. By employing the concept of the equiv- A rigid strip anchor plate of height B is buried vertically in sand
alent free surface, Neely et al. (1973) has used the method of char- with horizontal ground surface as shown in Fig. 1(a). The thickness
acteristics; the stresses along the equivalent free surface, however, of the plate is assumed to be negligible as compared to its height.
need to be assumed. Rowe and Davis (1982) have used the The bottom of the anchor plate lies at a depth H from ground sur-
displacements-based elastoplastic finite element method. Murray face. The interface of the anchor plate and surrounding soil mass,
and Geddes (1989) and Kumar (2002) employed the upper bound both front and back interfaces, has a friction angle δ. The horizontal
limit analysis; in these two approaches, however, the collapse pullout resistance (Pu ) of the anchor plate per unit length of anchor
mechanism needs to be assumed. Merifield and Sloan (2006) have plate needs to be determined. The soil mass is assumed to be per-
obtained the solution of this problem by using the lower and upper
fectly plastic, and it obeys Mohr-Coulomb failure criterion and an
bound theorems of the limit analysis in combination with finite
associated flow rule.
elements. However, the upper bound magnitude of the collapse
load from this approach becomes greater than that determined
by using the simple upper bound limit analysis based on the
Problem Domain and Velocity Boundary Conditions
assumption of the collapse mechanism (Kumar 2002; Kumar
and Kouzer 2008). With an intention of improving the existing The rectangular domain along with the different velocity boundary
upper bound numerical solution, in this note, an upper bound theo- conditions are shown in Fig. 1(b). The vertical and horizontal boun-
rem of limit analysis in conjunction with finite elements has been
daries OY, CX, and OX need to be located at sufficient distances
used to compute the horizontal pullout capacity of vertical anchors
away from the anchor plate. The locations of these boundaries with
buried at shallow depths in sand. The limit analysis approach of
D ¼ B to 3B, LB ¼ 2B to 6B, and LF ¼ 2H to 8H are found to be
Sloan and Kleeman (1995) based on discontinuous velocity fields
generally sufficient; where, the parameters D, LB , and LF are
has been used. The sizes of the elements approaching toward the
edges of the anchor plate were gradually reduced. The variations of defined in Fig. 1(b). Along the boundaries OX, OY and XC, the
the pullout resistance with changes in embedment ratio of the velocities of all the associated nodes are taken equal to zero.
The distances D, LB , and LF are derived in a way such that an
1
Associate Professor, Civil Engineering Dept., Indian Institute of increase in the values of these parameters do not cause any change
Science, Bangalore 560012, India (corresponding author). E-mail: in the magnitude of the failure load. A fully bonded contact
jkumar@civil.iisc.ernet.in between the anchor plate and the surrounding soil mass is modeled.
2
Research Fellow, Civil Engineering Dept., Indian Institute of Science, In order to simulate this anchor-soil interface condition, for the
Bangalore 560012, India. E-mail: jagdish@civil.iisc.ernet.in front and back nodes (with the same coordinates) lying along
Note. This manuscript was submitted on February 8, 2011; approved on the anchor plate, it was specified that V i ¼ 0 and ui ¼ V 0 ; where,
July 8, 2011; published online on July 11, 2011. Discussion period open
until November 1, 2012; separate discussions must be submitted for indi-
(1) V 0 = velocity of the anchor plates; and (2) ui and vi = nodal
vidual papers. This technical note is part of the International Journal of velocities at ith node in x and y directions, respectively. Because
Geomechanics, Vol. 12, No. 3, June 1, 2012. ©ASCE, ISSN 1532-3641/ the analysis is performed for perfectly cohesionless medium, it will
2012/3-333–337/$25.00. not be possible to have a stable unsupported vertical face (cut) of

INTERNATIONAL JOURNAL OF GEOMECHANICS © ASCE / MAY/JUNE 2012 / 333

Int. J. Geomech. 2012.12:333-337.


Anchor plate

H
B Pu
Tie rod

Back surface Front surface

(a)
LB LF
Anchor Anchor
Y C Back nodes Front nodes
Downloaded from ascelibrary.org by University of California, San Diego on 05/20/13. Copyright ASCE. For personal use only; all rights reserved.

ui = V0 ui = V0
vi = 0 vi = 0
(a)
H

B V0
u=0 u=0
v=0 v=0
y(v)
D

O x(u) X
Soil nodes

u=v=0
(b)

Fig. 1. (a) Definition of problem; (b) chosen domain and associated


velocity boundary conditions
(b)

the sand mass when a separation is considered between the back Fig. 2. Chosen finite element meshes for H∕B ¼ 5 and δ ¼ 0 with
surface of the anchor plate and the surrounding soil mass. (a) ϕ ¼ 30°; (b) ϕ ¼ 40°
A friction angle δ is imposed between the interface of the anchor
plate and the surrounding soil mass. The velocity boundary condi-
tions along with the interface friction angle applicable on the remains always a rigorous upper bound on the exact solution, the
anchor-soil interfaces are shown in Fig. 1(b). The soil and anchor linearized polygon is circumscribed to the parent yield circle. While
nodes at the same level have the same coordinates. A velocity dis- performing the computations, the value of p is kept equal to 24 in
continuity is imposed between the soil and anchor nodes with the all the cases; it was observed that for p ≥ 24, hardly any improve-
velocity jump making an angle δ with the soil-anchor interface. ment occurs in the results. The upper bound problem is formulated
as a linear programming problem in which the magnitude of Pu
needs to be minimized subjected to a number of linear equality
Finite Element Mesh and inequality constraints. All the steps for formulating the upper
The domain is discretized into a number of three-noded triangular bound limit analysis as a linear programming problem and the
elements. Each interface between the adjacent elements is kept as associated computational procedure are described in detail in the
the line of the velocity discontinuity. Typical finite element meshes paper of Kumar and Kouzer (2007). For solving the problem,
chosen for different combinations of H∕B, δ, and ϕ are illustrated in the necessary computer code was written in Matlab.
Figs. 2(a) and 2(b); in these figures, the parameters, E, N, and Dc
refer to the total number of (1) elements, (2) nodes, and (3) discon-
tinuities, respectively. Top and bottom edges of the anchor plate 20
become singular points. The sizes of the elements are gradually δ=0
7
reduced approaching these two singular points. The total number 6
of elements, nodes, and discontinuities are increased with increases 15
5
in H∕B, ϕ, and δ. The meshes were generated by writing the
necessary preprocessing code in Matlab. 4

10
3

Analysis 2
5
With the incorporation of the plastic strains, both within elements 1
and along all the chosen velocity discontinuities, the upper bound H/B
finite limit analysis is used. Following Bottero et al. (1980), 0
20 25 30 35 40 45 50
Mohr-Coulomb failure criterion is linearized by drawing a regular
p-sided polygon in x, y-stress coordinates; where, (1) X ¼ σx  σy φ (Degree)
and Y ¼ 2τ xy ; and (2) σx and σy = normal stresses, and τ xy = shear
Fig. 3. Variation of F γ with ϕ for different H∕B with δ ¼ 0
stress acting on the x, y-plane. In order that the numerical solution

334 / INTERNATIONAL JOURNAL OF GEOMECHANICS © ASCE / MAY/JUNE 2012

Int. J. Geomech. 2012.12:333-337.


Results and Comparisons (3) five values of δ, namely, 0, 0:25ϕ, 0:5ϕ, 0:75ϕ, and ϕ. The numeri-
cal results obtained from the analysis are summarized herein.
The nondimensional pull out factor F γ is defined by using the
following expression: Variation of F γ
The variation of F γ with ϕ for different values of H∕B, and δ is
Pu
Fγ ¼ ð1Þ presented in Figs. 3–7. The factor F γ increases continuously with
BγH increases in the values of ϕ, H∕B, and δ. The rate of increase of F γ
with ϕ increases continuously with an increase in ϕ. On the other
The computations were performed for (1) different values of H∕B hand, the rate of increase of F γ with H∕B decreases with an in-
ranging from 0 to 7; (2) different values of ϕ varying from 20 to 45°; and crease in H∕B. The values of F γ obtained from the present analysis
Downloaded from ascelibrary.org by University of California, San Diego on 05/20/13. Copyright ASCE. For personal use only; all rights reserved.

25 7 40 7
δ = 0.25 φ 6 35
δ=φ 6

20 5
5 30
4
4 25
15 3


20 2

3
10 15
2 1
10
5 H/B
1
5
H/B
0 0
20 25 30 35 40 45 50
20 25 30 35 40 45 50
φ (Degree)
φ (Degree)

Fig. 4. Variation of F γ with ϕ for different H∕B with δ ¼ 0:25 Fig. 7. Variation of F γ with ϕ for different H∕B with δ ¼ 1:00

35
16
δ = 0.5 φ 7 UBFE (Merifield & Sloan, 2006)
30 14 Present analysis
6
25 12
5
δ=0
4 10
20

3 8
15
2 6
10
1 4
5
H/B 2
0 0
20 25 30 35 40 45 50 0 1 2 3 4 5 6 7 8
φ (Degree)
(a)
Fig. 5. Variation of F γ with ϕ for different H∕B with δ ¼ 0:50
30
UBFE (Merifield & Sloan, 2006)
Present analysis
25

40
20 δ=
35
δ = 0.75 φ 7
6

15
30
5
4
25 10
3

20
2 5
15
1 0
10
0 1 2 3 4 5 6 7 8
5 H/B
0 (b)
20 25 30 35 40 45 50
φ (Degree) Fig. 8. Comparison of present F γ values with upper bound finite
elements limit analysis approach of Merifield and Sloan (2006) for
Fig. 6. Variation of F γ with ϕ for different H∕B with δ ¼ 0:75 (a) δ ¼ 0; (b) δ ¼ ϕ

INTERNATIONAL JOURNAL OF GEOMECHANICS © ASCE / MAY/JUNE 2012 / 335

Int. J. Geomech. 2012.12:333-337.


Table 1. Comparison of F γ from Present Analysis with that Reported by Kumar (2002) for H∕B ¼ 3, 5, and 7
H∕B ¼ 3 H∕B ¼ 5 H∕B ¼ 7
Kumar (2002) Kumar (2002) Kumar (2002)
ϕ δ∕ϕ Present analysis Bilinear Log-spiral Present analysis Bilinear Log-spiral Present analysis Bilinear Log-spiral
30 0 4.17 4.22 4.27 6.34 6.45 6.71 8.28 8.47 9.04
0.5 5.40 5.55 5.51 7.97 8.08 8.22 9.85 10.35 10.80
1 5.99 6.72 6.21 8.60 9.43 8.93 10.57 11.85 11.59
40 0 6.87 6.88 6.94 10.85 10.89 11.24 14.41 14.59 15.39
0.5 10.90 11.26 10.99 16.09 16.48 16.37 20.27 21.24 21.57
1 13.76 16.54 13.86 19.22 22.78 19.4 23.85 28.46 24.90
Downloaded from ascelibrary.org by University of California, San Diego on 05/20/13. Copyright ASCE. For personal use only; all rights reserved.

Table 2. Comparison of F γ from Present Analysis with that Reported by Chen (1975), Soubra (2000), and Kumar (2002) for H∕B ¼ 1
Kumar (2002) Chen (1975)
ϕ δ∕ϕ Present analysis Bilinear Log-spiral Bilinear Log-spiral Soubra (2000)
30 0.0 1.37 1.50 1.50 1.50 1.50 1.50
0.5 2.15 2.28 2.28 2.28 2.28 2.27
1.0 2.80 3.14 3.07 3.14 3.07 2.97
40 0.0 2.24 2.30 2.30 2.30 2.30 2.30
0.5 4.64 4.76 4.74 4.75 4.75 4.70
1.0 7.42 8.71 8.02 8.7 8.01 7.51

for various combinations of H∕B and ϕ were compared with the the vertical wall. The results obtained from the analysis for
numerical upper bound solution of Merifield and Sloan (2006) H∕B ¼ 1, were compared in Table 2 with the existing solutions
for δ ¼ 0 and δ ¼ ϕ; the analysis of Merifield and Sloan (2006) by using (1) the upper bound limit analysis (Chen 1975; and Soubra
is also based on with the usage of the finite elements, but by em- 2000); and (2) the upper bound analysis based on bilinear and
ploying a mesh comprising of uniform elements. The associated composite logarithmic spiral rupture surfaces (Kumar 2002). It
comparison is presented in Fig. 8. In all the cases, as compared can be seen that as compared to all earlier reported upper bound
to the computational results of Merifield and Sloan (2006), the solutions, the present analysis provides always a little smaller
present analysis provides a little lower value of the pullout factor.
The difference between the present solution and that provided by
Merifield and Sloan increases continuously with increases in the
values of ϕ, δ, and H∕B. The present numerical solution is found
to be better in all the cases. This observation is also aligned with
that made earlier by Kumar and Kouzer (2008) while computing
the upper bound solution for finding the vertical uplift resistance
of horizontal anchors in sand with the usage of the finite elements
limit analysis.
For H∕B ¼ 3, 5, and 7, the present numerical results are also
compared with that obtained by Kumar (2002) based on the
assumption of the (1) bilinear collapse mechanism; and (2) logarith-
mic spiral collapse mechanism. The associated comparison is pro-
vided in Table 1. The present results are found to be always lower
than Kumar (2002) in all the cases. As compared to the bilinear
collapse mechanism, the logarithmic spiral rupture surface provides
a better match with the present results. The values of F γ obtained on
the basis of the logarithmic spiral rupture surface are even lower
than provided by Merifield and Sloan (2006). The higher values
of F γ obtained by Merifield and Sloan (2006) are because through-
out the problem domain constant sizes of the elements were being
chosen. This comparison indirectly demonstrates the requirement
of choosing smaller sizes as well as larger number of elements
around the singular point while performing the upper bound finite
elements limit analysis.
If the contribution of the active thrust exerted by sand on
the back surface of the anchor surface is ignored, for H∕B ¼ 1,
Fig. 9. Variation of nodal velocity patterns for H∕B ¼ 5 and ϕ ¼ 30°
the problem of finding the pullout resistance of the anchor would
with (a) δ ¼ 0; (b) δ ¼ ϕ
become the same as that of obtaining the passive earth pressure on

336 / INTERNATIONAL JOURNAL OF GEOMECHANICS © ASCE / MAY/JUNE 2012

Int. J. Geomech. 2012.12:333-337.


values of F γ . The difference between the present and the existing References
results increases with increases in the values of ϕ and δ.
Bottero, A., Negre, R., Pastor, J., and Turgeman, S. (1980). “Finite element
Nodal Velocity Patterns method and limit analysis theory for soil mechanics problem.” Comput.
Methods Appl. Mech. Eng., 22(1), 131–149.
The nodal velocity patterns were studied for H∕B ¼ 5 with
Chen, W. F. (1975). Limit analysis and soil plasticity, Elsevier, Amsterdam.
ϕ ¼ 30°; namely, 0 and ϕ. The nodal velocity patterns are shown Kumar, J. (2002). “Seismic horizontal pullout capacity of vertical anchors
in Figs. 9(a) and 9(b). The zone of influence reaches up to the in sands.” Can. Geotech. J., 39, 982–991.
ground surface in all the cases; the zone of influence refers to Kumar, J., and Kouzer, K. M. (2007). “Effect of footing roughness on
the region in the problem domain for which the magnitudes of bearing capacity factor Nγ.” J. Geotech. Geoenviron. Eng., 133(5),
the nodal velocities remain significant. The size of the zone of in- 502–511.
fluence increases with an increment in the values of δ and ϕ. The Kumar, J., and Kouzer, K. M. (2008). “Vertical uplift capacity of horizontal
boundary of the influence zones in all the cases is contained well anchors using upper bound limit analysis and finite elements.”
Downloaded from ascelibrary.org by University of California, San Diego on 05/20/13. Copyright ASCE. For personal use only; all rights reserved.

within the chosen problem domain for carrying out the analysis. Can. Geotech. J., 45(5), 698–704.
Merifield, R. S., and Sloan, S. W. (2006). “The ultimate pullout capacity of
anchors in frictional soils.” Can. Geotech. J., 43(8), 852–868.
Conclusions Meyerhof, G. G. (1973). “Uplift resistance of inclined anchors and piles.”
Proc., 8th Int. Conf. on Soil Mechanics and Foundation Engineering,
The horizontal pullout resistance of the vertical strip anchor plate Vol. 2, International Society for Soil Mechanics and Foundation
embedded in sand, has been determined by using the upper bound Engineering, 167–172.
finite elements limit analysis. The magnitude of the pullout factor Murray, E. J., and Geddes, J. D. (1989). “Resistance of passive inclined
increases substantially with increases in the embedment ratio anchors in cohesionless medium.” Gèotechnique, 39(3), 417–431.
Neely, W. J., Stuart, J. G., and Graham, J. (1973). “Failure loads of vertical
(H∕B), the friction angle (ϕ) of sand, and the anchor-soil interface
anchor plates in sand.” J. Soil Mech. Found. Div., 99(9), 669–685.
angle (δ).The size of the zone of the influence increases with an
Rowe, R. K., and Davis, E. H. (1982). “The behaviour of anchor plates in
increment in the values of H∕B, ϕ, and δ. The results obtained from sand.” Gèotechnique, 32(1), 25–41.
the analysis are found to be better as compared to that reported in Sloan, S. W., and Kleeman, P. W. (1995). “Upper bound limit analysis using
literature. The study also reveals that it is equally important in discontinuous velocity fields.” Comput. Methods Appl. Mech. Eng.,
the upper bound finite elements limit analysis to refine the mesh 127(1), 293–314.
as per the location of the singularity rather than employing a Soubra, A. H. (2000). “Static and seismic passive earth pressure coeffi-
uniform mesh. cients on rigid retaining structures.” Can. Geotech. J., 37(2), 463–478.

INTERNATIONAL JOURNAL OF GEOMECHANICS © ASCE / MAY/JUNE 2012 / 337

Int. J. Geomech. 2012.12:333-337.

You might also like