Вы находитесь на странице: 1из 5

International Journal of Computer Applications in Engineering Sciences

[VOL II, ISSUE IV, DECEMBER 2012]

[ISSN: 2231-4946]

Reinforced Slope Stability Analysis Using


Natural Element Method (NEM)
E. Ghazavi Baghini #, M.M. Toufigh *
#

Civil Engineering Department, Shahid Bahonar University of Kerman, 22 Bahman Blv., Kerman, Iran
#

ghazavi.emad@gmail.com
*toufigh@mail.uk.ac.ir

Abstract In reviewing the stability of reinforced


slopes, which are often the main components of vital
arteries like roads, finding the most probable slip surface
and factor of safety is the main target. This paper is based
on finding existing stresses in the soil and the geogrid
under static loading and linear elastic zone with using
NEM, which is one kind of meshfree methods. By
combining numerical analysis with limit equilibrium
methods, factor of safety is obtained. As a result, a
reinforced slope that is analyzed with NEM is compared
with the result of Geostudio 2004 software, which indicates
good accuracy in analyzing and locating failure.
Keywords Natural Element Method (NEM), reinforced
slope, meshfree, geogrid

I. INTRODUCTION
Slope stability analysis to determine the most
probable slip surface or the least factor of safety is one
of the most important issues in geotechnical engineering.
The importance of this is more apparent when failure of
a slope in mines or dams causes great loss of life and
finances. For many years, it was used for unreinforced
slopes.
Although the design of unreinforced slope is simple
and their materials are available, their performance is
limited. For this reason, construction of reinforced soil
slope was developed. Art of soil stabilized with dried
branches of trees, bamboo, straw, and similar materials
has been common since ancient times throughout the
world; however, a French civil engineer and architect,
Henry Vidal, introduced a model for a general
configuration of reinforced embankment built in 1963.
This plan rapidly was accepted throughout the world for
retaining walls, foundations of bridges, and other
projects.

Slope stability analysis in terms of factor of safety for


stability of a slope is important. Limit equilibrium
methods and finite element are the most common
methods used for slope stability analysis in statics. Limit
equilibrium methods are widely used for slope stability
analysis. Before analyzing, a possible slip surface is
assumed, and then, limit equilibrium analysis is done
according to the soil volume that is above the probable
slip surface. There are many methods based on this
process, for example, Bishop (1955), Janbu (1957),
Morgenstern-Price (1965), Spencer (1967), and Sarma
(1979).
One of the major defects of finite element method is
the requirement of mesh generation throughout the
geometry of problem. In the problems with variable
boundary conditions (for example, spread out crack on
surfaces), it is necessary that mesh generation constantly
be updated; that requires plenty of time to be spent on
[1]. Hence, in recent years, a new branch of
computational mechanics, which is named meshfree
methods, has been developed. The main reason of this
idea in mechanical problems is such as heat transfer and
Lagranges formula. Meshfree method is based on the
definition of shape functions, and the quality of
minimizing approximation functions (strong-weak form)
is classified [2].
II.

NATURAL ELEMENT METHOD

Natural Element Method (NEM) is a procedure in


which a mesh-free updated Lagrangian Approximation is
permitted without mesh distortion. The main
characteristics of NEM is its independency upon the
mesh regularity in contrast to the FEM, in which mesh
quality has remarkable influence on the results [3]. The
spectacular properties of NEM have led to the usage of
this method in the variety of engineering problems [4,5].
One of the advantages of NEM is the imposition of
essential boundary conditions (Dirichlet Boundaries),
only by relocation in a system of linear equations [6].
NEM interpolant is linear between nodes on the

410 | P a g e

Baghini et. al.

boundary of convex hull which facilitates the imposition


of essential boundary conditions.
NEM is based on natural neighbors interpolation
scheme [7,8], and correlation among nodes is computed
by local distribution and density of nodes [9]. These
interpolants are based on the Voronoi cells of one set of
nodes which are constructed both in problem domains
and along boundaries. The discrete model of the domain
() consists of Voronio diagrams tessellated the entire
domain.
Sukumar showed the main advantages of this method
in elastostatic problems and the interesting property of
NEM is the acceptable accuracy in the data interpolation
[9].
A. Natural neighbors
Sibson introduced the concept of natural neighbors for
interpolation scattered data [7]. Voronoi diagram and
Delaunay tessellation are the main theories to obtain
natural neighbors. A given domain () is partitioned
with special subdomains (i) called Voronoi cells [9].
The Voronoi tessellation is one of the most basic and
useful geometric structures which can be used in an
irregular set of nodes.
A given set of nodes N = {n1,n2,n3, . . . ,nm} is
assumed in R2 in which the Voronoi cell is a tessellation
of plane into zone TI, which is associated with node nI.
TI Defines a domain where the distance between each
node and nI is less than the other nodes in N. A Voronoi
diagram for nI in mathematical terms is defined as:
)

) JI}

(1)

Where d(x,xi) is the Euclidean distance between x, xi.


It is clear that each Voronoi diagram 1st is the
intersection of many open half-spaces, each being
limited by perpendicular bisector. Consequently, the
Voronoi cells are closed and convex. (See Fig. 1a).
The Delaunay triangle resulted from Voronoi diagram
is constructed by the nodes whose Voronoi cells have
common boundaries (See Fig. 1b). The duality between
the two implies that a Delaunay edge between the two
nodes in the plane

Fig. 1. Geometric concept of Natural Neighbor: (a) Voronoi Diagram


1st. (b) Delaunay Tessellation. (c) Circumcircle criteria. (d)Voronoi
Diagram 2nd.

should exist; if and only if their Voronoi cells share


common edge. Empty circle criterion is the main feature
of Delaunay triangles. If DT(nJ,nk,nI) is considered to be
a Delaunay triangle of the set N, then the circle of DT
includes no other node of N, and nj, nk, nI are the three
vertices of this triangle and the center of the circle is on
the incidence of Voronoi cell boundaries which formed
the Delaunay triangle [10]. According to Fig. 1c, two
specified circles include x; hence the three vertices in
every circle are indicated as natural neighbors of x (node
= 1, 4, 5, 6).
The concept of first-order Voronoi diagram is
extendable to higher order Voronoi diagrams in the
plane. The second order Voronoi diagram of nodes set is
exploited in the formation of natural neighbor
interpolants. The second order Voronoi diagram of one
set of nodes is a subdivision of plane into the zone T IJ,
where each zone is accompanied by a nodal pair (ni,nj).
Moreover, TIJ is the location of all points which have
ni as the nearest neighbor and nj as the second nearest
neighbor [1]. The mathematical term for higher order
Voronoi diagram (2nd) is as follow:
)

(2)
Also, the second order Voronoi diagram for node x is
illustrated as a domain with bold blue lines in Fig. 1d.

411 | P a g e

Reinforced Slope Stability Analysis Using Natural Element Method (NEM)

B. Sibson interpolation
Natural neighbor coordinate is applied to interpolate
the domain of the problem. As shown in Fig. 1d, the
natural neighbor coordinate of x, related to node I (I = 1)
is defined as the ratio of overlap area of their Voronoi
cells to the total area of the second Voronoi cell. Based
on related areas, the shape function for node (I) can be
defined as [7,8]:

Where L is differential operator, is stress tensor, b is


body force, n is normal stress on domain,
is
predetermined
forces,
is predetermined
displacements,
is border of applied force,
is
border of supports.
Constitutive equation based on Hookes law for
elastic materials is written:
,

(8)

(3)
Stress-strain relation can be shown as:
Where AI(x) is the area related to node I, A(x) is the
area of the 2nd Voronoi diagram, and
is the shape
function formed for node I.
The Sibson shape function has special properties that
make NEM as a useful tool to solve PDEs:
(X)
)=

(4)
,

(5)

(X)=1,

(6)

Eq. (3) expresses that the shape function is locally


compact support. It means that the shape function ()
interpolates only in Voronoi 2nd cell. Eq. (4) shows that
the Sibson shape functions possess kronockr which is
a useful property to impose essential boundary
conditions and Eq. (5) gives the property of unity
partition.
Sibsonian shape functions are
at any location and
on the given node. In addition, in one-dimension,
natural neighbor interpolation as described earlier, is
identical to linear finite elements [1].
In this paper, reinforced elements are modeled linear.
III. CONSTITUTIVE EQUATION AND ITS WEAK FORM SHAPE

Partial differential equations and boundary conditions


for two-dimensional problem with small displacements
on domain which has encircled with
and
is as
follow [11]:

(9)

Due to the fact that shape functions


are not
always linear, and in general, Natural Neighbor
Interpolation scheme
logically is consisted of
polynomials, system equation is obtained from the
overall weak form [12]. The standard weak form of
equation (7) can be determined as follow [11]:

L u DLu d u
T

bd u td 0 (10)

By substitution of shape function in above equation


and follow a series of mathematical operations, a
discrete system of linear equations are obtained as
follow [11]:
Kd =

(11)

As:
=

(12)
bd

(13)

In above equations,
that is shape functions
derivatives matrix is defined as:

0
=

on

(7)

= 0

(14)

u=

412 | P a g e

Baghini et. al.

Material constants matrix D for isotropic and linear


elastic material in plane strain is given by:

according to the angle of the sliding surface in nth.


fragment, and P is ultimate tensile strength of the
reinforcement element.

0
V. EXAMPLE

D=

1
0

IV.

(15)

DETERMINATION THE FACTOR OF SAFETY

After obtaining the stresses in the problem domain,


surfaces is assumed S shape. In this article these surfaces
are circular, the factor of safety (F.S) on this surfaces are
calculated according to the definition overload (that
defines the factor of safety as ratio of the total resistance
forces to total driving forces along a specific sliding
surface) [13]:
(16)

F.

Sliding surfaces are divided into n fragment that have


each one equal length L, and since all fragment have
the same length, above equation can be simplified as
follows:
F.S = min (

(17)

Where
is mobilized shear stress of each fragment,
is shear strength of each fragment. and (based on
the Mohr-Columbus criterion) can be written in the
following form:
= C+
= -0.5 (

tan +Prcos

tan

+Prsin

In the following example, reinforced slope domain is


determined with 684 nodes )Figure 2). Soil behavior is
assumed linear elastic, and structure is analyzed under
the influence of its weight without considering pore
water pressure )Figure 2).
Shape functions are calculated based on the Watson
algorithm, and for numerical integration calculations, we
use Delaunay tessellation for cells. We assume that
sliding surfaces are circular )Figure 3).
Finally, sliding surface related to the lowest factor of
safety is presented as the most probable slip surface. In
order to compare the results with other numerical
methods, the same example is analyzed with the
Geostudio2004 software (refer to Figure 5).
Boundary conditions and natural element model for a
reinforced slop with 684 nodes are shown in Figure 2.
We assume that the soil shear strength parameters are
C=41.7 kN
and
= , density is
also

Elastic parameters are E = 16200 kN


and
; also, the ultimate tensile strength of geogrid
is 100 kNm. Height of slope is H =20 m, and slope
face angle is 76 .
The algorithm is written in Matlab, which all related
calculations of shape function and stresses as well as
factor of safety have been done. After entering the data
related to the geometry and the mechanical properties of
materials, the factor of safety equal to 1.382 from NEM
is obtained.

(18)
(19)

)+(

) cos 2 +

sin

2 (20)
Where c and are cohesion and internal friction angle
and
is the normal stress on nth. fragment
,
and
are the effective stresses on nth. fragment.
is angle between new coordinate axis and general
coordinate axis or angle between intersection place of
failure line and reinforcement element, that is calculated

413 | P a g e

Figure 2: Boundary conditions and nodal arrangement for the analysis


of reinforced slop from NEM

Reinforced Slope Stability Analysis Using Natural Element Method (NEM)

Figure 3: Assumed slide surfaces

of most probable failure surface in reinforced slope


problems, the factor of safety was determined with high
accuracy, insomuch as shape function of linear element
in natural element method (NEM) and finite element are
the same. Therefore, the modeling of reinforcement does
not have any difference with the finite element, and the
only difference is in planting of total stiffness matrix.
Mesh generation was an important step in previous
studies. However, in this paper, the conventional
methods for mesh generation are omitted, and more
flexibility in solution is achieved.

REFERENCES
[1]

[2]

[3]

[4]
Figure 4: Delaunay tessellation
[5]

[6]

[7]
[8]
[9]

[10]
[11]
[12]

Figure 5: Analysis of slop with the Geostudio 2004 software that the
factor of safety is calculated 1.347

VI. CONCLUSION

[13]

Sukumar N., Moran B., Belytshko T., The Natural Element


Method in Solid Mechanics International Journal for Numerical
Method in Engineering, 839-860. (1998).
Idelsohn, S.R. and onate, E., and et al. Meshless Finite Element
Ideas Fifth World Congress on Computational Mechanics.
Vienna, Austria. (2002).
D. Gonzalez, E. Cueto, F. Chinesta, M. Doblar, A natural
element updated Lagrangian strategy for free-surface fluid
dynamics, J. Comput. Phys. 223(2007) 127150.
J. Cho, H. Lee, 2-D large deformation analysis of nearly
incompressible body by natural element method, Comput. Struct.
84 (2006) 293304.
C. Yongchang, Z. Hehua, A meshless local natural neighbour
interpolation method for stress analysis of solids, Eng. Anal.
Boundary Elem. 28 (2004) 607613.
E. Cueto, J. Cegonino, B. Calvo, M. Doblar, On the imposition
of essential boundary conditions in natural neighbour Galerkin
methods, Commun.Numer. Methods Eng. 19 (2003) 361376.
R. Sibson, A Vector Identity for the Dirichlet Tessellation,
Cambridge University Press, 1980.
R. Sibson, A brief description of natural neighbor interpolation,
Interpreting multivariate data, 21 (1981).
N. Sukumar, B. Moran, A. Yu Semenov, V. Belikov, Natural
neighbor Galerkin methods, Int. J. Numerical Methods Eng. 50
(2001) 127.
C.L. Lawson, Software for C1 interpolation (1977).
Lui, G. R, (2002), Mesh Free Methods: Moving Beyond the
Finite Element Method, CRC Press LLC.
Yongchang, C. and Hehua, Z., (2004). A Meshless Local
Natural Neighbour Interpolation Method for Stress Analysis of
Solid, Engineering Analysis with Boundary Elements, v.28, pp.
607-613.
Zheng, H., Tham, L. G. and Liu, D., (2006), On Two Definition
of the Factor of Safety Commonly Used in the Finite Element
Slope Stability Analysis, Computer and Geotechnics,v. 33, pp.
188-195.

Apropos of the results, in the analysis of reinforced


slope, it has been observed that NEM meshfree method
with shape functions, which change softly around nodes,
in stress-strain analysis, has good accuracy. Also, NEM
method in calculation of integrals uses Delaunay
tessellation, which has simple but principal concepts,
does not cause mesh generation problems. Also, with
combination of this analysis and the concepts of limit
equilibrium method for determining the factor of safety

414 | P a g e

Вам также может понравиться