Professional Documents
Culture Documents
2000
Japanese Geotechnical Society
ABSTRACT
The effects of stabilizing piles on the stability of a slope are predicted by the three-dimensional elasto-plastic shear
strength reduction finite element method. The soil-pile interaction is simulated with zero-thickness elasto-plastic inter-
face elements. The numerical results are compared with those obtained by Bishop's simplified method, where the reac-
tion force of the piles is determined by Ito-Matsui's equation. The effects of the pile spacing, pile head conditions,
bending stiffness, and pile positions on the safety factor are analyzed. The shear strength reduction finite element
method shows that the pile head conditions and bending stiffness can considerably influence the stability of the slope,
but this cannot be indicated by the limit equilibrium method. The positions of the pile row have significant influence
on the stability of the slope. The shear strength reduction finite element method indicates that the pile row needs to be
installed in the middle of the slope for maximum safety. By contrast, Bishop's simplified method shows that the piles
should be installed slightly closer to the top of the slope.
Key words: finite element method, interface element, limit equilibrium method, pile, slope stability, three-dimension
(IGC: E6)
73
the analysis. This method, however, can only be used in where F is the shear strength reduction factor. The
purely cohesive slopes undergoing creep. Ito and Matsui reduced shear strength parameters Cp and cfJF replace the
(1975) proposed a theoretical method to analyze the shear strength parameters c and cjJ of Mohr-Coulomb's
growth mechanism of lateral force acting on stabilizing failure criterion in elasto-plastic finite element analysis.
piles when the soil is forced to squeeze between piles. The Firstly, a gravity turn-on is implemented under elastic
method was developed to specifically calculate the pres- state to determine the initial stress distribution inside the
sures acting on the passive piles in a row. The force that slope. Then, stresses and strains are calculated by the
the failing mass exerts on a row of piles can be expressed elasto-plastic finite element method. The shear strength
as a function of the soil strength, pile diameter, and pile reduction factor, F, is initially selected to be so small that
spacing. If a portion of that force is assumed to counter- the materials of the slope are under elastic conditions.
act the driving force of the slope, the safety factor of the The shear strength reduction factor, F, is then increased
slope stabilized with piles can be calculated as a function incrementally until the global failure of the slope is
of the pile size and position based on the limit equilibri- achieved, which means that the finite element calculation
um method (Ito et al., 1979, 1981, 1982; Yamagami and diverges under a physically real convergence criterion.
Yamakawa, 1995; Hassiotis et al., 1997). Although this The global safety factor at failure lies between the shear
approach appears useful, the model is derived for rigid strength reduction factor, F, at which the iteration limit
piles, which may not represent the actual piles in the field is reached, and the immediately previous value.
as they are unlikely to be rigid. The model may also pro- For a slope stabilized with piles, the slope completely
vide doubtful solutions when the piles are closely spaced. consists of the soil when the shear strength reduction fac-
In the present paper, the effects of the pile spacing, pile tor, F, is less than unity. The material of the solid ele-
head conditions, bending stiffness, and pile positions are ments within the extent of the pile is changed from the
numerically analyzed with the three-dimensional elasto- soil into the pile material when the shear strength reduc-
plastic shear strength reduction finite element method. tion factor, F, reaches unity. The stresses in the pile are
The soil-pile interaction is simulated by zero-thickness assumed to be zero when the material is changed. Then,
three-dimensional interface elements. The results of the the shear strength reduction factor increases once again,
shear strength reduction finite element method are com- step by step until the failure of the slope.
pared with Bishop's simplified method, where the reac- One of the main advantages of the shear strength reduc-
tion force by the piles is determined by Ito-Matsui's equa- tion finite element method is that the safety factor
tion. emerges naturally from the analysis without the user hav-
ing to commit to any particular form of the mechanism a
priori. For a slope reinforced with piles, the slip surface
ANALYSIS METHOD may not be circular or continuous due to the piles. This
Shear Strength Reduction Finite Element Method situation cannot be sufficiently considered by the limit
The slope stability is commonly assessed using limit equilibrium method, but is not difficult for the shear
equilibrium methods (Bishop, 1955; Morgenstern and strength reduction finite element method. The shear
Price, 1965). The ability of the limit equilibrium methods strength reduction finite element method can predict the
to determine the stability of slopes reinforced with piles safety factor of a slope reinforced with piles as a slope
may be in doubt because of the soil-pile interaction. without piles, and the pile behavior characteristics can be
However, the elasto-plastic shear strength reduction obtained simultaneously.
finite element method, in which the definition of the
global safety factor is identical to that in the convention- Simulation of Soil-Pile Interaction
al limit equilibrium methods, can analyze the slope stabil- The isoparametric interface element has been de-
ity under a general frame. A numerical comp'!rison has scribed by Beer (1985). Such an isoparametric interface
shown that the shear strength reduction finite element element is fully compatible with isoparametric solid ele-
method can yield nearly the same safety factor and corre- ments. The interface stresses are characterized by the nor-
sponding critical sliding surface as the conventional limit mal stress, an, and two shear stresses, Ts, and, Tt. The nor-
equilibrium methods for slopes without piles under either mal and shear stresses are related with the constitutive
two-dimensional or three-dimensional conditions (U gai, matrix to the normal and tangential relative displace-
1989; Ugai and Leshchinsky, 1995). ments of the interface element. For the elastic constitu-
The global safety factor of slopes, defined in the shear tive matrix, the following simple expression is chosen:
strength reduction finite element method, is identical to
the one in limit equilibrium methods. The reduced shear
D= 0
Kn 0 0]0
Ks (3)
strength parameters Cp and cfJF are defined as:
[
c 0 0 Kt
Cp=- (1)
F where Kn is the elastic normal stiffness, and Ks and Kt are
tan c/J) the s-direction and !-direction shear stiffness, respec-
c/Jp=tan- 1 ( F (2) tively. The interface stiffness should be chosen such that
the initial slope of the load displacement relationship
the tangential direction of the slip circle at the pile posi- with the depth and is not influenced by the pile positions.
tion and the horizontal direction, and Q is the total force The simplex reflection technique, an optimization
exerted on the sliding body by the piles, which is the in- method, is used for locating the critical slip circle that
tegral of the lateral force per unit thickness acting on the has the lowest safety factor. When the slope is stabilized
pile from the top of the pile to the depth of the slip circle with the piles, the critical slip surface is found after addi-
at the pile position. The lateral force per unit thickness tion of the resisting moment by the piles. Thus a smaller
acting on the pile, q, has been shown by Ito and Matsui safety factor can be obtained than that considering the
(1975) to be: effect of the piles with the original critical slip surface
without piles.
q=cA [
N<P tan~
1 1
{exp (D -D N<P
D2
2
tan~ tan(!!._+
8 4
~ ))
1/2 2 tan "-+2N
'P
112
<P + N-
112
<P ]
RESULTS AND DISCUSSIONS
- 2N <P tan ~- 1} + N 1; 2 N Model Slope
<P tan~+ <P- 1
112 An idealized slope with a height of 10 m and a gradient
+N- 112
- c D 1 2tan"-+2N
'P <P <P 2D2N - 112
)
of 1 V:1.5 Hand a ground thickness of 10m is analyzed
( 12
N ~ tan ~ + N<P- 1 <P with a three-dimensional finite element mesh, as shown
0
E
ci
......
X
Lx
35.0m
1.7
I
I
I I I
-e- Bishop
I
I
I
1.6 ------ FEM(free)
-B- FEM(hinged)
1.5
en
~
1.4
1.3
1.2
0 2 4 6 8
Fig. 3. Vectors of nodal displacement induced by shear strength
reduction and critical slip circle located by Bishop's simplified Dl/D
method for slope without piles
Fig. 4. Effects of pile spacing on safety factor
10 ---- Dl/D=2
-a- Dl/D=3
------- Dl/D=4
-e- Dl/D=6
(a)
II
---- Dl/D=2
10 -a- Dl/D=3
~ Dl/D=4
-e--- D 1/D=6
(b)
Fig. 5. (a) Pile behavior characteristics for various pile spacing (free head), (b) Pile behavior characteristics for various pile spacing (hinged head)
should be a coincidence that the safety factors obtained valid because the analytical results of the piles under mov-
by the two methods compare well with each other for the ing soil show that the first extreme point of the distribu-
free head piles because of the existence of negative pres- tion of the shear force in the piles is developed at the level
sure on the free head piles, as indicated in Fig. 5(a). The of the slip surface (Ito et al., 1981; Poulos, 1995;
negative pressure implies that this part of the pile cannot Hassi otis et al., 1997). The nodal displacements due to
supply resisting force to the slope; on the contrary, it the shear strength reduction and the critical slip surface
makes the safety factor of the slope decrease. But it is im- located by Bishop's simplified method are shown in Fig.
possible for Eq. (10) to obtain a negative pressure on the 6. By comparing the magnitude and distribution of the
piles. nodal displacements inside the slope, especially at the
The lateral shear force along the x-direction in the pile pile position, as shown in Figs. 3 and 6, it is reasonable
reaches the first extreme point at a critical depth, as that the critical depth is taken as the level of the slip sur-
shown in Fig. 5. The critical depth can be regarded as the face. Table 2 shows the depth of the slip surface at the po-
level of the slip surface although the shear strength reduc- sition of the piles, determined by the foregoing methods.
tion finite element method cannot predict a clear slip sur- For free head piles, the smaller the spacing of the piles,
face like the limit equilibrium method. The definition is the deeper the level of the slip surface predicted by the
10 ----Free
-a- Unrotated
-e-- Hinged
~Fixed
- - lto-Matsui
15
(a)
10 ---- Free
-a- Unrotated
-e-- Hinged
--e- Fixed
-- Ito-Matsui
············ Ito-Matsui(fs)
15 -----'-----------'
(b)
Fig. 7. (a) Pile behavior characteristics for various pile head conditions (EP = 60 GPa), (b) Pile behavior characteristics for various pile head condi-
tions (EP = 200 GPa)
larger bending stiffness (Ep=200 GPa). Table 3. Effects of pile head conditions and bending stiffness on safe-
Even for piles with larger bending stiffness (Ep=200 ty factor
GPa), Fig. 7(b) shows that the pressure on the piles varies Young's Shear strength reduction FEM Bishop's
considerably with different pile head conditions because modulus simplified
of piles (GPa) Free Unrotated Hinged Fixed method
of the relative displacements between the soil and pile
(Fig. 8(b)), that is in turn, the mobilized shear strength of 60 1.36 1.45 1.54 1.55 1.37
the soil-pile interface. The pressure on piles with larger 200 1.55 1.57 1.55 1.56 1.37
bending stiffness, predicted by the shear strength reduc-
tion finite element method, is significantly larger than
that estimated by lto-Matsui's equation, as shown in Fig. ment method, is significantly larger than that obtained by
7(b), where the notation Ito-Matsui(fs) means that the Bishop's simplified method, as shown in Table 3.
pressure is calculated by Ito-Matsui's equation using the The pile behavior characteristics for various pile head
reduced shear strength parameters Cp and cf>p. The larger conditions are shown in Figs. 7(a) and 7(b ). The hinged
pressure on the piles is one of the reasons why the safety head condition results in the smallest bending moment in
factor, predicted by the shear strength reduction finite ele- the piles, followed in order by the fixed, unrotated, and
15cm 15cm
~ L____j
- -=--
---
15cm I Scm
L___j L___j
- -:=.--
3cm 3cm
L___j L___j
3cm 3cm
L___j L___j
(a) (b)
Fig. 8. (a) Local top views of nodal displacements induced by shear strength reduction (Ep=60 GPa), (b) Local top views of nodal displacements
induced by shear strength reduction (Ep = 200 GPa)
free head condition. Because it is more possible for the can also result in a larger safety factor for slopes rein-
stabilizing piles to be yielded by the bending moment forced with flexible piles. If the restrained head condition
than by the shear force (Ito and Matsui, 1979), a re- cannot be obtained, the bending stiffness should be in-
strained pile head (hinged or fixed) is recommended, and creased in order to prevent possible negative pressure on
the free head condition should be avoided due to its maxi- the piles.
mum bending moment in the piles. A restrained pile head
1.6 show that when the piles are placed in the lower and mid-
dle portions of the slopes, the pressure on the piles is
1.5 larger because the numerical results show that the relative
soil-pile displacement is so large that the shear strength
1.4 of the soil-pile interface is sufficiently mobilized. But, it
is clear that the shear strength of the soil-pile interface is
~ 1.3 not sufficiently mobilized when the piles are installed in
the upper portions of the slopes, because the pressure on
1.2 -II- FEM(free) these piles is smaller than that on the piles in the lower
-B- FEM(hinged) and middle portions of the slopes. However, Eq. (10)
1.1 -e- Bishop shows that the influence of the pile positions on the pres-
sure on the piles cannot be considered by Ito-Matsui's
1.0 method, which is used to determine the reaction force of
0.0 0.2 0.4 0.6 0.8 1.0 the piles in Bishop's simplified method. Because the reac-
Lx/L tion force of the piles, determined by Ito-Matsui's equa-
tion, is the same regardless of the pile positions, Bishop's
Fig. 9. Effects of pile positions on safety factor simplified method predicts the maximum safety factor
when the piles are installed a little closer to the top of the
slopes. Figures 10(a) and 10(b) show that the pile behav-
Effect of Pile Positions ior characteristics are significantly influenced by the posi-
The pile positions in the slope are indicated with a tions of the piles, and that the bending moment and the
dimensionless ratio of the horizontal distance between shear force in the piles are maximum when the piles are
the slope toe and the pile position, Lx, to the horizontal placed in the middle of slopes.
distance between the slope toe and slope shoulder, L, as
shown in Fig. 2. The equivalent Young's modulus of the
piles, EP = 60 GPa, and the center-to-center spacing CONCLUSIONS
D1 =3D. The influence of the pile positions on the safety The three-dimensional shear strength reduction finite
factor of a slope stabilized with piles is shown in Fig. 9. element method is used to predict the stability of a slope
The numerical results, obtained with the shear strength reinforced with piles, where the soil-pile interaction is
reduction finite element method, show that the safety fac- simulated by zero-thickness, three dimensional, elasto-
tor of slopes reinforced with hinged head piles is larger plastic interface elements. The numerical results obtained
than that with free head piles. The difference is more sig- by this method are compared with those based on
nificant when the piles are installed in the middle portion Bishop's simplified method where the reaction force of
of slopes. Therefore, a restrained pile head (hinged or the piles is determined by Ito-Matsui's equation. As the
fixed) is recommended once again. The improvement of results of the numerical analyses of the effects of the pile
the safety factor of slopes reinforced with piles is largest spacing, pile head conditions, bending stiffness, and pile
when the piles are installed in the middle of the slopes, positions, the following conclusions can be obtained:
regardless of pile head conditions. However, Bishop's (1) The stability of a slope can be improved with piles,
simplified method shows that the safety factor increases and as might be expected, the improvement of the
with the horizontal distance between the slope toe and safety factor increases with reducing the spacing be-
the pile position, Lx, until the piles are placed considera- tween the piles. The safety factor, obtained by the
bly closer to the top of the slopes. This is the same as the shear strength reduction finite element method, is
results of the limit equilibrium methods obtained by Ito significantly larger than that predicted by Bishop's
et al. (1979) and Hassiotis et al. (1997). When the piles simplified method for hinged head piles, which is
are placed in the lower portion of the slopes, the pressure closer to the assumption of the rigid piles in Ito-
on the piles is positive even for the free head condition Matsui's equation. The two methods can obtain a
due to the shallow sliding soil mass. Figure 9 shows that similar rate of change in the safety factor with
the difference in the safety factor obtained by these two decreasing the pile spacing.
methods is small when Lx/ L is less than 0.3. When the (2) The pile head conditions and the bending stiffness
hinged head piles are installed in the middle portion of influence the pressure on the piles, and thus the safe-
the slopes, the safety factor obtained by the shear ty factor of the slopes. If the piles are more flexible,
strength reduction finite element method is significantly the pressure on free head piles is negative over a cer-
larger than that predicted by Bishop's simplified method. tain depth. Therefore, the safety factor of the slope
This is mainly because the pressure on the piles, predict- is significantly smaller than that of a slope rein-
ed by the shear strength reduction finite element method, forced with restrained head piles. This implies that
is larger than that obtained by Ito-Matsui's equation, as a restrained pile head (hinged or fixed) should be
shown in Fig. lO(b). recommended, and the free head condition should
The pile behavior characteristics for various pile posi- be avoided when piles are used to stabilize slopes.
tions are shown in Figs. lO(a) and lO(b). These figures For restrained pile head conditions, the safety fac-
10
- - - Lx!L= 1/6
-a- Lx!L= 1/3
15 -e- Lx/L= 112
-e- Lx!L=2/3
_._ Lx!L=5/6
- - Ito-Matsui
20
(a)
5
s
::ci
-+-=' )
p..
CD
t::::1 ~
10
- - - Lx/L=l/6
-a- Lx!L= 1/3
-e- Lx/L= 112
15 ~ -e- Lx/L=2/3
- A - Lx!L=5/6
- - Ito-Matsui
j~
20
(b)
F1g. 10. (a) Pile behavior characteristics for various pile positions (free head), (b) Pile behavior characteristics for various pile positions (hinged
head)
tor predicted by Bishop's simplified method is over- of the slope for the largest safety factor.
ly conservative.
(3) The numerical results of the shear strength reduc-
tion finite element method show that the pressure REFERENCES
on piles is larger only when the piles are placed in 1) Baguelin, F., Frank, R. and Said, Y. H. (1977): "Theoretical study
the lower and middle portion of the slopes. This is of lateral reaction mechanism of piles," Geotechnique, Vol. 27,
because the relative displacements of the soil-pile in- No. 3, pp. 405-434.
2) Beer, G. (1985): "An isoparametric joint/interface element for
terface are so large that the effect of the piles is finite element analysis," Int. J. Numer. Meth. Engrg., Vol. 21, pp.
sufficiently mobilized. Therefore, piles should be in- 585-600.
stalled in the middle of slopes for the maximum safe- 3) Bishop, A. W. (1955): "The use of the slip circle in the stability of
ty factor. By contrast, the reaction force supplied slopes," Geotechnique, Vol. 5, No. 1, pp. 7-17.
by piles is assumed to mobilize sufficiently, regard- 4) Chow, Y. K. (1996): "Analysis of piles used for slope stabiliza-
tion," Int. J. Numer. Anal. Meth. Geomech., Vol. 20, No.9, pp.
less of the pile positions, in Ito-Matsui's equation. 635-646.
Therefore, Bishop's simplified method shows that 5) Day, R. A. and Potts, D. M. (1994): "Zero thickness interface ele-
the piles should be placed slightly closer to the top ments-numerical stability and application," Int. J. Numer. Anal.