Download as pdf or txt
Download as pdf or txt
You are on page 1of 13

Applied Mathematical Modelling 45 (2017) 179–191

Contents lists available at ScienceDirect

Applied Mathematical Modelling


journal homepage: www.elsevier.com/locate/apm

New mathematical model for computing natural frequencies


of retaining walls considering soil–structure interaction
Mohammad Saeed Ramezani∗, Ali Ghanbari, S.A.A. Hosseini
Faculty of Engineering, Kharazmi University of Tehran, No. 49 Mofattah Ave. Tehran, I.R. Iran

a r t i c l e i n f o a b s t r a c t

Article history: In this study, a new analytical model is proposed for estimating natural frequencies of
Received 16 June 2015 retaining walls, and accordingly three closed-form formulas are presented for calculating
Revised 26 October 2016
the first three natural frequencies. In the first step, rigid body motion of retaining walls
Accepted 15 December 2016
is taken into account and two closed-form formulas for calculating natural frequencies of
Available online 24 December 2016
the rigid mode of deformation are presented via exact solution. Next, applying the energy
Keywords: method, a new closed-form formula is developed for calculating the natural frequency of
Analytical model the flexural deformation mode. Another innovation of this paper is the modeling of the
Natural frequency foundation soil with both torsional and translational springs that make it possible to con-
Retaining wall sider rocking and sliding modes for retaining wall motion. Additionally, the effect of back-
Rigid mode fill soil interaction is taken into consideration by massless translational springs along the
Flexural mode wall. The results thus obtained from the proposed formulas are then compared with nu-
Soil–structure interaction merical analysis using ANSYS software, and a good agreement was observed.
© 2016 Elsevier Inc. All rights reserved.

1. Introduction

For an appropriate design, it is required to have accurate knowledge on the seismic behavior of structures. Natural
frequency is one of the most important parameters in the seismic design of retaining walls. As an example, if the natural
frequencies of the system are similar to the predominant vibratory frequency of the external excitation, the structures are
excited by resonant effects. In this situation, structures undergo large deformation, which can even lead to structure failure.
Therefore, introducing an appropriate technique to determine the natural frequencies of structures is worthwhile.
Retaining walls are generally designed to resist sliding, overturning, and structural failure due to lateral pressures gener-
ated by its backfill. Therefore, soil–wall interaction is important factor to be taken into consideration. Vibrating systems can
be characterized by the distribution of their mass as discrete or continuous. Discrete systems are those whose mass can be
considered to be concentrated at a finite number of locations. These systems have a finite number of degrees of freedom
and n-distinct natural frequencies of vibration. However, Retaining walls should be modeled in the form of continuous
system. The reason is that, as the mass, stiffness, and damping are distributed throughout the system, infinite numbers
of pointed mass continuously move against each other during vibration. Continuous systems have an infinite number of
natural frequencies with one mode shape corresponding to each natural frequency.
Three motion modes have been observed in the seismic analyses of retaining walls that depend on the structural type
of retaining wall; one of the aforementioned modes is crucial. These are rigid translational, rotational motion, and flexural
deformation modes [1–4].


Corresponding author.
E-mail addresses: saeed.ramezani66@yahoo.com (M.S. Ramezani), ghanbari@khu.ac.ir (A. Ghanbari), ali.hosseini@khu.ac.ir (S.A.A. Hosseini).

http://dx.doi.org/10.1016/j.apm.2016.12.019
0307-904X/© 2016 Elsevier Inc. All rights reserved.
180 M.S. Ramezani et al. / Applied Mathematical Modelling 45 (2017) 179–191

List of notations

E, Ec wall (concrete) modulus of elasticity


Es soil modulus of elasticity
H backfill width
I moment of inertia of uniform wall
I(x) moment of inertia of non-uniform wall
k(x) Winkler spring stiffness
K1 translational stiffness of the backfill
K2 translational stiffness of the foundation
K3 rotational stiffness of the foundation
L height of wall
m mass per unit length of uniform wall
m(x) mass per unit length of non-uniform wall
J0 mass moment of inertia
Tmax maximum kinetic energy
Tref reference kinetic energy
Vmax maximum potential energy
W wall width (constant section)
Wb wall base width
Wt wall top width
x, x(t) displacement of center of gravity of the wall
θ , θ (t) rotation of the wall
Y(x) shape function (trial function)
ρ wall density
ρs soil density
υs soil Poisson’s ratio
ω natural circular frequency
fr 1 translational rigid natural frequency (first rigid natural frequency)
fr 2 rotational rigid natural frequency (second rigid natural frequency)
ω f , ff fundamental flexural frequency (first flexural natural frequency)
Fc correction factor

In current paper, using the theory of beams on elastic foundations, a new analytical model for calculating natural
frequencies of retaining walls is proposed. The impact of backfill soil interaction is modeled with elastic translational
springs along the wall. Torsional and translational springs are also used to consider flexibility of foundation. Finally, three
closed-form formulas are presented for calculating natural frequencies of rigid and flexural modes.

2. Literature review

Estimating the natural frequency of walls has been subject of research for quite some time. One of the first studies was
conducted by Matsuo and Ohara [5]. They solved differential equation of wall motion based on two initial assumptions and
defined two limiting boundaries for predominant natural frequency that they believed the real solution lied somewhere
in between. Using a numerical approach, Wood [6] calculated natural frequencies of homogeneous and elastic soil mass
embedded between two rigid walls under plane strain conditions. He presented plot of backfill natural frequency as a
function of backfill width for different Poisson’s ratio values.
Scott [7] modeled the backfill soil as a one-dimensional shear beam attached to the wall by Winkler springs and
determined the predominant frequency of wall. Wu [8] derived a closed-form expression for estimating the undamped
natural frequencies of the backfill model under small-amplitude vibration. His analysis was based on the shear beam
analogy with zero shear stress at the surface and the assumption of no vertical normal stress throughout the backfill. He
assumed that the nonlinear behavior of soil under strong ground motions would reduce the fundamental frequency of the
wall-backfill system. Accordingly, he suggested that by evaluating the reduced shear modulus of backfill at larger strain
levels, his formula could be extended to stronger input ground motions.
Veletsos and Younan [9] examined the deficiencies of the Scott [7] model and discussed the sources of inaccuracies.
They proposed that the deficiencies of this model may be removed by modeling the restraining action of the medium by
a series of elastically supported, semi-infinite horizontal bars with distributed mass, rather than by springs of constant
stiffness that simulate the action of finite-length, massless, and unconstrained bars.
Elgamal et al. [10] conducted a dynamic full-scale test on a cantilever wall-backfill system and measured the soil–wall
system response over a wide range of resonances through a frequency sweep test. Therefore, the resonant frequencies and
M.S. Ramezani et al. / Applied Mathematical Modelling 45 (2017) 179–191 181

Fig. 1. The retaining wall with variable cross section and its equivalent model.

the associated resonant mode shapes were estimated. They also modeled two types of wall using finite element method and
concluded that a plane-strain seismic analysis is quite adequate for wall–soil systems with uniform heights and properties.
However, wall–soil systems of variable heights might display additional 3-D resonances. Hatami and Bathurts [11] analyzed
a reinforced-soil retaining wall model numerically and studied the effects of some structural design parameters on the
fundamental frequency. They reported that the fundamental frequency of reinforced-soil walls can be estimated accurately
from a few available formulas based on linear elastic wave theory.
Based on energy methods and by modeling the soil as series of linear springs, Ghanbari et al. [12] presented a new
formulation to calculate the flexural fundamental natural frequency of retaining walls. They considered the behavior of
the wall as a cantilever beam (rigid base), so the effect of soil–structure interaction was not considered in their research.
Applying the same methods, Abbasi et al. [13] obtained an analytical formula for flexural fundamental frequency of
geosynthetic-reinforced soil retaining walls. They used translational springs to include the effect of foundation flexibility
during sliding, but they did not consider the rocking mode for retaining wall motion. Moreover, they failed to present any
formula for calculating natural frequencies of rigid modes. Ramezani et al. [14] improved Abbasi et al. [13] approach and
they offered a new analytical solution for calculating natural frequencies of geosynthetic-reinforced with a full-height rigid
concrete facing (GRS-FHR walls).
In the current paper, for the first time, the rigid body motion of retaining walls is investigated. Another innovation is
the modeling the soil of foundation with both rotational and translational springs that make consideration of rocking and
sliding modes possible.
The proposed method also considers the vertical cross sectional width change, and hence, enables us to calculating the
natural frequency of retaining walls with different types of backfill.
Several other studies on dynamic behavior of soil–wall systems and free vibration analysis of structures were carried out
in recent years. Researches like: Gazetas et al. [15]; Choudhury and Chatterjee [16]; Lanzoni et al. [17]; Yesilce and Catal
[18]; Kamgar and Saadatpour [19]; Malekinejad and Rahgozar [20]; Wu [21]; Ansari et al. [22]; Komak Panaha et al. [23].

3. Proposed model

A retaining wall with variable cross section and its equivalent model are shown in Fig. 1. In this model, the wall behaves
as a beam with free end support condition on elastic foundation and under free transverse vibration. Using the theory of
beams on elastic foundation, the backfill is modeled by elastic translational springs along the wall. Furthermore, the effect
of earth soil interaction is considered using rotational and translational springs allow both rocking and sliding modes to
be considered for retaining wall motions [24]. The proposed model results in two modes of rigid body motion and infinite
modes of flexural motion.
The following assumptions are made to simplify the model formulations:

1. The granular backfill is dry with a constant modulus of elasticity at all points.
2. The backfill and foundation soil are assumed massless. To simplify the problem most researchers have accepted the
massless hypothesis. Nevertheless, research on the mass contribution of soil on vibration of structures continues.
3. The effects of backfill stiffness are modeled by continuous elastic compression springs with constant stiffness along the
height of the wall, so that the tensile stiffness of the backfill is assumed zero.
182 M.S. Ramezani et al. / Applied Mathematical Modelling 45 (2017) 179–191

Fig. 2. The rigid body motion of the wall. (a) Analytical proposed model. (b) Free body diagram.

4. The effects of foundation are modeled by torsional and translational elastic springs.
5. The behavior of soil is assumed linear elastic. Other advanced conditions, like non-constant and nonlinear spring stiffness,
are also possible.

4. Natural frequencies of rigid modes

The rigid body motion of the retaining wall and its free body diagram are shown in Fig. 2. Displacement of gravity-center
and rotation of the wall are denoted by x(t) and θ (t), respectively.
From the free body diagram shown in Fig. 2, the force equilibrium equation in the horizontal direction can be written
as:
 L1  L2
mẍ + K2 (x − L1 θ ) + K1 (x − |y|θ )dy + K1 (x + |y|θ )dy = 0. (1)
0 0

Moreover, rotational dynamical equilibrium around the gravity-center can be expressed as:
 L1  L1
J0 θ̈ + K3 θ − K2 (x − L1 θ )L1 − K1 (x − |y|θ )y dy + K1 (x + |y|θ )y dy = 0. (2)
0 0

Simplifying the expressions leads to the following equations:


θ 
mẍ + K2 (x − L1 θ ) + K1 x(L2 + L1 ) + K1 L2 2 − L1 2 = 0, (3)
2
x 2  θ 
J0 θ̈ + K3 θ − K2 (x − L1 θ )L1 + K1 L2 − L1 2 + K1 L2 3 + L1 3 = 0. (4)
2 3
In the equations above, parameters m and J0 are the mass and mass moment of inertia of the wall, respectively; other
parameters are defined in Fig. 2. Eqs. (3) and (4) can be re-arranged and re-written in matrix form as:
         
m 0 ẍ K2 + K1 (L2 + L1 ) −K2 L1 + K1
L2 2 − L1 2 x 0
+   2
  = . (5)
0 J0 θ̈ −K2 L1 + K1
2
L2 2 − L1 2
K3 + K2 L1 2 + K1
3
L2 3 + L1 3 θ 0

For a free vibration analysis, the solution is assumed to be:

x(t ) = X cos(ωt + ∅ ) (6)

θ (t ) =  cos(ωt + ∅) (7)
M.S. Ramezani et al. / Applied Mathematical Modelling 45 (2017) 179–191 183

Fig. 3. Arbitrary differential element of wall.

Substitution of Eqs. (6) and (7) into Eq. (5) yield:


      
−mω2 + K2 + K1 (L2 + L1 ) −K2 L1 + K1
L2 2 − L1 2 X 0
  2
  = . (8)
−K2 L1 + K1
2
L2 2 − L1 2
−J0 ω2 + K3 + K2 L1 2 + K1
3
3
L2 + L1 3
 0

Equating the determinant of the coefficient matrix to zero, the frequency equation can be derived as:
1

Jmω4 − L3 K1 m − L2 L1 K1 m + LL1 2 K1 m + L1 2 K2 m + JLK1 + J K2 + K3 m ω2


3
1 4 2 1 3
+ L K1 + L K1 K2 + LK1 K3 + K2 K3 = 0 (9)
12 3
The natural frequencies of the rigid mode can be derived from the frequency equation (Eq. (9)) for any arbitrary cross
section of wall. Therefore, for a retaining wall with constant cross section, the natural frequencies of the rigid mode can be
calculated using the following formulae:

2 λ2 − λ3 0.5 1
fr1 = × , (10)
λ1 2π

2 λ2 + λ3 0.5 1
fr2 = × . (11)
λ1 2π
In these equations, fr 1 and fr 2 are natural frequencies of translational and rotational rigid mode (Hz), respectively.
Parameters λ1 , λ2 and λ3 are defined as follow:
 
λ1 = 4ρ LW L2 + 1 , (12)

λ2 = 2L3 K1 + 8L2 K2 + LK1 + 2K2 + 24K3 , (13)

λ3 = 64L4 K2 2 − 8L3 K1 K2 + 32L2 K2 2 + 192L2 K2 K3 + L2 K1 2 + 4LK1 K2 − 48LK1 K3 + 4K2 2 − 96K2 K3 + 576K3 2 . (14)
In Eqs. (12)–(14), parameters W, L, and ρ are the width, height and density of the wall respectively. Parameter K1 is
the translational stiffness of the backfill; K2 and K3 indicate the translational and rotational stiffness of the foundation,
respectively. Similarly, another formula can be obtained for retaining wall with variable cross-section.

5. Natural frequency of flexural mode

In order to think up a suitable approach for computing the natural frequency of the flexural mode, the first step is
developing the dynamic equation of motion.
An arbitrary differential element of the wall in dynamic equilibrium is shown in Fig. 3. The dynamic equilibrium
equation for this system yields:

∂2 ∂ 2y ∂ 2y
− 2 EI (x ) 2 − K (x )y = m(x ) 2 . (15)
∂x ∂x ∂t
Here, EI(x) is the flexural stiffness of the wall, y is wall transverse displacement, K(x) is the Winkler spring stiffness
parameter, and m(x) is the mass per unit length of wall. Taking advantage of the method of separation of variables, the
differential equation above can be separated as:
y(x, t ) = Y (x )F (t ). (16)
Substituting the Eq. (16) into Eq. (15) and simplifying leads to the following expression:
1 d2   k(x ) F̈ (t )
− EI (x )Y  (x ) − = . (17)
m(x )Y (x ) dx 2 m (x ) F (t )
184 M.S. Ramezani et al. / Applied Mathematical Modelling 45 (2017) 179–191

Considering the right side of the Eq. (17) and equaling it to a constant value, a solution for F(t) is driven as follows:

F (t ) = C cos (ωt −
). (18)
Where, C and
are constant coefficients that are obtained from the initial conditions of problem, and ω is the frequency
of vibration. Accordingly, Eq. (17) can be simplified as follow:

d2  
2
EI (x )Y  (x ) − k(x )Y (x ) = ω2 m(x )Y (x ). (19)
dx
This governing differential equation is difficult to solve. In such cases, Rayleigh’s method can be used most conveniently
to find the approximate fundamental natural frequency of vibration of a system without having to solve the governing
differential equation of motion [25]. Actually, Rayleigh’s method is a technique for finding natural frequencies by equating
the maximum kinetic energy of a system with the maximum potential energy. Rayleigh’s principle can be interpreted as
follows [26]:
Vmax
Vmax = Tmax ⇒ ω f 2 = minR(Y (x ) ) = . (20)
Tre f
Here, Vm ax is the maximum potential energy, Tmax the maximum kinetic energy, ωf the fundamental flexural frequency of
system, Tref the reference kinetic energy, and finally Y(x) is the shape function. In the Rayleigh method, it is important to
select an appropriate shape function (trial function). The closer the shape function resembles the first natural mode, the
better the estimate of the fundamental frequency. Here, the normalized form of the first vibration mode of beam with
free-end support condition is selected as a proper shape function for retaining walls [27]:
4.73
4.73
4.73
4.73

Y (x ) = cos x + cos h x − 0.98 sin x − 0.98 sin h x . (21)


L L L L
Reference kinetic energy and maximum potential energy are calculated as below [26]:
 L
1
Tre f = m(x )Y (x )2 dx, (22)
2 0

 2 
1 L
d2 Y ( x ) 1 L
1
Vmax = EI(x ) dx + k1 · Y(x )2 · dx + k2 Y2 ( x = 0 )
2 0 dx2 2 0 2
2
1 dy
+ k3 (x = 0 ) (23)
2 dx

Where EI(x) is flexural stiffness of wall, m(x) is mass per unit length of wall, parameters K1 , K2 and K3 indicate
translational stiffness of the backfill, translational and rotational stiffness of the foundation, respectively. Applying Rayleigh’s
principle, the value of ω2 is computed as follow:

2  dy 2
d 2 Y (x )
∫L0 EI (x ) dx + ∫L0 k1Y (x ) dx + k2Y 2 (x = 0 ) + k3
2
d x2 dx (x = 0 )
ω2 = . (24)
∫L0 m(x )Y (x ) dx
2

Assuming a constant cross section for the retaining wall, this expression is simplified as:

41.7EW 3 + 0.5K1 L4 + 4K2 L3 + 86.4K3 L


ωf 2 = . (25)
ρ LW
The equation above was obtained using only one mode of shape function. In general, the accuracy of Rayleigh’s method
depends on the choice of the proper shape function and the number of considered-modes used in the problem solving
process. By considering a larger number of modes, the approximation can be made more accurate. Accordingly, a correction
factor (Fc ) is proposed for taking into account the effect of higher modes. Therefore, the final formula for calculating the
fundamental flexural frequency of the retaining wall with constant cross section is proposed as follow:

2 41.7EW 3 + 0.5K1 L4 + 4K2 L3 + 86.4K3 L 1


f f = Fc × . (26)
ρ LW 2π
In the equation above, parameters E, ρ , W, and L are elasticity module, density, width, and height of the wall, respec-
tively. Parameter K1 is the translational stiffness of the backfill; K2 and K3 indicate the translational and rotational stiffness
of the foundation, respectively. Similarly, for a retaining wall with variable cross-section (Fig. 1), the following formula is
presented:

2 μ2 + μ3 1
f f = Fc × . (27)
μ1 2π
M.S. Ramezani et al. / Applied Mathematical Modelling 45 (2017) 179–191 185

Table 1
Coefficients of correction factor equation.

Height of wall (m) α1 α2 α3 α4

0<L<3 21.3 −27.5 11.4 −0.5


3≤L<5 3.15 −6.5 4.3 −0.1
5≤L<7 0.8 −2.28 2 −0.4
7 ≤ L < 10 0.37 −1.22 1.27 −0.6
10 ≤ L < 15 0.07 −0.32 0.49 −0.7
L ≥ 15 0.04 −0.21 0.35 −0.8
w

Backfill

For wall with constant cross-section: L


If : W < 5L ⇒ Chose β = W
If : W > 5L ⇒ Chose Fc = 1

Foundation

wt

Backfill

For wall with variable cross-section: L


If : W b+
2
Wt
< 5L ⇒ β = W b+
2
Wt

If : W b+
2
Wt
> 5
L
⇒ Chose F c =1

wb

Foundation

Parameters μ1 , μ2 , and μ3 are defined as follow:


μ1 = ρ L4 (Wb + Wt ), (28)
 
μ2 = E 13.17Wb3 + 13.17Wt 3 + 28.54Wb 2Wt + 28.54WbWt 2 , (29)

μ3 = K1 L4 + 8K2 L3 + 172.8K3 L (30)


Where, Wt and Wb are the widths of the wall in top and bottom. The correction factor (Fc ) can be calculated from
following equation:
Fc = α1 β 3 + α2 β 2 + α3 β + α4 . (31)
The coefficients of correction factor equation introduced in Eq. (31) are given in Table 1.

6. Comparison of the analytical solution with FEM

In order to assess the accuracy of the proposed method, the obtained results were compared with the numerical result
obtained from the finite element software ANSYS V.15. For this purpose, modal analysis was performed with the block
Lanczos method. The retaining wall was modeled by Beam-188 element. The backfill and soil foundation were modeled by
means of Combination-14 elements with linear behavior. The interfaces between the retaining wall and the soil foundation
were modeled by free-end supports with two degrees of freedom. Fig. 4 shows the finite element model deformation of
the retaining wall in first rigid mode (translational rigid mode), second rigid mode (rotational rigid mode), and first flexural
mode (fundamental flexural mode).
In the first comparison, the variations in the geometric properties of the wall were examined and thus soil properties
were considered constant. The geometrical and mechanical properties of the retaining walls are listed in Table 2.
186 M.S. Ramezani et al. / Applied Mathematical Modelling 45 (2017) 179–191

First Rigid Second Rigid First Flexural


Mode Mode Mode

Fig. 4. Deformation of the FEM model of the retaining wall in different natural modes.

Table 2
Geometrical dimensions and mechanical properties of the re-
taining walls for comparison with FEM.

Wall (concrete) modulus of elasticity Ec = 20 GPa

Wall density ρ = 2320 kg/m3


Wall height L = 3, 5, 7, 10 m
Wall width (constant cross-section) W = (0.1 L) m

The results of the comparison between the proposed method and FEM for calculating the natural frequencies of rigid
modes and flexural mode are listed in Table 3 and plotted in Fig. 5. These results show that the solutions using the
proposed formulas are in acceptable agreement with FEM solutions. Since the natural frequencies of rigid modes (Eqs. (10)
and (11)) were obtained from the exact solution of the problem, the conformity of these exact results with the numerical
solution can also verify the validity of the FEM model.
As a second comparison, variations in the soil stiffness were evaluated. Here, the length and width of the wall were
considered 7 and 0.7 m respectively; other properties were selected according Table 2. The results are presented in
Table 4 and Fig. 6. This comparison shows that the translational rigid natural frequencies are in excellent agreement with
FEM solutions. However, in connection with rotational rigid frequency and first flexural frequency, it was observed that
with increasing the soil stiffness, the disparity between the results of proposed method and FEM became greater. This can
be due to the variations in wall behavior, as foundation soil stiffness increases. In such a situation, the behavior of the wall
should be considered as that of a typical cantilever beam. Therefore, it is recommended, when the foundation soil stiffness
is very high, to use the formula proposed by Ghanbari et al. [12]. They reported an analytical formula for calculating the
fundamental natural frequency of retaining walls, assuming the wall behaves as a cantilever beam.

7. Comparison with previous studies

In this section, the results obtained from proposed method are compares with the results of other researches related
to the wall natural frequency. Accordingly, in Fig. 7, variation of the first natural frequency versus the height of wall was
compared with the results reported by Scott [7], Wu [8], Ghanbari et al. [12] and Abbasi et al. [13] assuming geometrical
and mechanical properties of the retaining wall and soil in Table 5.
M.S. Ramezani et al. / Applied Mathematical Modelling 45 (2017) 179–191 187

Table 3
The results of comparison between the proposed method and numerical solution for calculating the natural frequencies (K1 = 9 MN/m, K2 = 9 MN/m and
K3 = 3 MN m/rad). (a) Translational rigid mode. (b) Rotational rigid mode. (c) First flexural mode.

L (m) (a) Translational rigid natural frequency (fr 1 ) (Hz) Relative error (%)

Proposed method (Eq. (10)) FEM (ANSYS)

3 13.07 13.40 2.55


5 9.95 10.12 1.74
7 8.37 8.48 1.29
10 7.00 6.99 0.07
L (m) (b) Rotational rigid natural frequency (fr 2 ) (Hz) Relative error (%)
Proposed method (Eq. (11)) FEM (ANSYS)
3 24.26 24.30 0.18
5 15.90 15.69 1.30
7 12.22 12.03 1.58
10 9.37 9.25 1.29
L (m) (c) First flexural natural frequency (ff ) (Hz) Relative error (%)
Proposed method (Eq. (26)) FEM (ANSYS)
3 99.26 101.80 2.56
5 60.56 60.83 0.44
7 44.96 43.70 2.79
10 31.66 30.65 3.19

Table 4
The results of comparison between the proposed method and numerical solution for variation in soil stiffness.

Es Kx = K1 = Kθ = K3 Translational rigid natural Rotational rigid natural First flexural natural frequency (ff )
(MPa) K3 (MN/m) (MN m/rad) frequency (fr 1 ) (Hz) frequency (fr 2 ) (Hz) (Hz)

Proposed FEM Proposed FEM Proposed FEM


method method method

10 9 3 8.38 8.48 12.22 12.03 44.96 43.70


30 26 8 14.24 14.41 20.76 19.92 48.22 46.92
60 53 17 20.34 20.57 29.65 27.39 53.02 52.01
90 80 27 24.99 25.28 36.44 32.61 57.46 56.00
120 107 35 28.90 29.27 42.13 36.76 61.52 57.71

Table 5
Geometrical dimensions and mechanical properties of the retaining
walls, comparison of the proposed method with previous studies.

Soil modulus of elasticity Es = 10 MPa

Soil Poisson’s ratio υ s = 0.4


Soil density ρ s = 1600 Kg/m3
Wall modulus of elasticity Ec = 20 GPa
Wall density ρ = 2320 Kg/m3
Width of backfill H=4L

The results show that using Scott and Wu methods the first natural frequency of the wall predict lower than proposed
method, which is because wall rigidity is not considered in their approach; they only considered the effects of backfill soil
properties in their formulation. Moreover, applying Ghanbari et al. and Abbasi et al. formulas, the first natural frequency is
calculated to be greater than that using the proposed method because their methods neglect foundation interaction effects.

8. Results obtained from proposed method

Variations of the first rigid frequency (translational rigid mode frequency) and the first flexural natural frequency versus
the flexibility of the retaining wall (W/L ratio) are compared in Fig. 8. The data show that as wall flexibility increases, the
translational rigid natural frequency tends to the first flexural natural frequency. The reason is that, in the flexural mode,
the rigidity of system is due to the stiffness of wall and soil; while, in the rigid mode, the rigidity of system is only due
to the stiffness of soil. Therefore, it can be seen that when the wall stiffness decrease (in case of flexible walls), the flexural
natural frequency close to rigid natural frequency.
In Fig. 9, for different soil stiffness conditions, variations of first natural frequency versus the height of the retaining wall
are plotted. As can be seen, the value of the soil stiffness generally plays an important role in the free vibration response
and has a significant effect on the natural frequency. As an example, for a wall with 5 m height, the first natural frequency
188 M.S. Ramezani et al. / Applied Mathematical Modelling 45 (2017) 179–191

a
fr1- Proposed Method fr1- Numerical Method
fr2- Proposed Method fr2- Numerical Method
30

25

Natural Frequency (Hz)


20

15

10

Rigid Modes
5 Kx=K1=K2= 9 MN/m
Kθ=K3= 3 MN-m/rad
Width of Wall: W=0.1L
0
0 2 4 6 8 10 12
Height Of Wall, L (m)
b
ff- Proposed Method ff- Numerical Method
120

100
Natural Frequency (Hz)

80

60

40

First Flexural Mode


20 Kx=K1=K2= 9 MN/m
Kθ=K3= 3 MN-m/rad
Width of Wall: W=0.1L
0
0 2 4 6 8 10 12

Height Of Wall, L (m)


Fig. 5. Variations of natural frequencies versus the height of the retaining wall. (a) Rigid modes. (b) First flexural mode.

varied from 9.9 to 34.3 Hz when the soil stiffness varied from 10 to 120 Mpa. Moreover, it is observed that with increasing
the height of wall, the natural frequency decreases in contrast.

9. Conclusion

Using an analytical model, three closed-form formulas for calculating the natural frequencies of retaining wall were
presented. For this purpose, the retaining wall was modeled as a plane strain beam with free-end support conditions on
an elastic foundation and the effects of backfill soil interaction were modeled by massless elastic translational springs
along the wall. Moreover, the effects of earth soil interaction were modeled with rotational and translational springs
that enabled consideration of the rocking and sliding deformation modes for retaining wall motion. In the first step,
two analytical formulas (Eqs. (10) and (11)) for calculating the natural frequencies of the rigid mode of deformations of
retaining walls were developed using exact solution method. Then, applying the energy method (Rayleigh’s method), new
M.S. Ramezani et al. / Applied Mathematical Modelling 45 (2017) 179–191 189

fr1- Proposed Method fr1- Numerical Method


fr2- Proposed Method fr2- Numerical Method
ff- Proposed Method ff- Numerical Method
70

60

Natural Frequency (Hz)


50

40

30

20

10 Height of Wall: L=7m


Width of Wall: W=0.1L
0
0 20 40 60 80 100 120 140
Soil Modulus, Es (Mpa)
Fig. 6. Variations of natural frequencies versus the soil modulus of elasticity.

40
Abbasi et al. [13]
35 Ghanbari et al. [12]
Proposed Method
First Natural Frequency (Hz)

30 Sco [7]
Wu [8]
25

20

15

10

0
0 1 2 3 4 5 6 7 8 9 10 11
Height Of Wall, L (m)
Fig. 7. Comparison between results of proposed method with previous studies.

analytical formula was proposed (Eqs. (26) and (27)) for calculating the fundamental natural frequency of the flexural
mode. The accuracy of the suggested formulas was assessed by comparing the present solutions with the results obtained
from numerical analyses using the ANSYS V.15 finite element software. The comparison revealed that the results obtained
using the proposed formulas are in acceptable agreement with results obtained using the FEM solutions. It was observed
that with increasing soil stiffness, the discrepancy between analytical results and FEM for rotational rigid mode and first
flexural mode, increased. This may be due to the variation of the wall behavior with increasing foundation soil stiffness. In
such a situation, the behavior of the wall should be considered as that of a cantilever beam. Therefore, it is recommended
that when the foundation soil stiffness is very high, to use the formula proposed by Ghanbari et al. [12] for calculating
fundamental natural frequency. The findings in this study show that with an increase in the flexibility of the retaining
wall, the translational rigid frequency tends toward the flexural natural frequency. It was also observed that the value of
soil stiffness and the height of the retaining wall have a significant effect on the natural frequency. Increasing in the soil
stiffness leads to greater natural frequencies; however, increasing in the height of wall results in lower natural frequencies.
190 M.S. Ramezani et al. / Applied Mathematical Modelling 45 (2017) 179–191

First rigid mode First flexural mode

30

25

Natural Frequency (Hz)


20

w
15
Backfill

10 L

Foundation

0
0.1 0.08 0.06 0.04 0.02 0
W/L
Fig. 8. Variations of translational rigid frequency (first rigid frequency) and first flexural natural frequency versus the flexibility of the retaining wall.

60

(ES)backfill = (ES)foundaon
50 ES = Soil Elascity
First Natural Frequency (Hz)

Width of Walls = 0.1L

40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Height Of Wall, L (m)
Fig. 9. Variations of first natural frequency versus the height of retaining wall for different soil stiffness.

References

[1] Y. Wu, Displacement-based Analysis and Design of Rigid Retaining Walls During Earthquake, University of Missouri-Rolla, Rolla, 1999 Ph.D. Dissertation.
[2] M. Sabermahani, M. Ghalandarzade, A. Fakher, Experimental study on seismic deformation modes of reinforced-soil walls, Geotext. Geomembr. 27
(2009) 121–136.
[3] D. Choudhury, A. Katdare, S. Shukla, B. Basha, P. Ghosh, Seismic behaviour of retaining structures, design issues and requalification techniques, Indian
Geotech. J. 44 (2014) 167–182.
[4] B.M. Das, Z. Luo, Principles of Soil Dynamics, Cengage Learning, US, 2016.
[5] H. Matsuo, S. Ohara, Lateral earth pressure and stability of quay walls during earthquake, Proceedings of the Second World Conference on Earthquake
Engineering, Science Council of Japan, Tokyo-Kyoto, Japan, 1960.
[6] J. Wood, Earthquake-Induced Soil Pressures on Structures, California Institute of Technology, Pasadena, 1973 Ph.D. Dissertation.
M.S. Ramezani et al. / Applied Mathematical Modelling 45 (2017) 179–191 191

[7] R. Scott, Earthquake-induced earth pressures on retaining walls, Proceedings of the Fifth World Conference on Earthquake Engineering, IAEE, June
1973.
[8] G. Wu, Dynamic Soil–Structure Interaction: Pile Foundations and Retaining Structures, University of British Columbia, Vancouver, 1994 Ph.D. Thesis.
[9] A.S. Veletsos, A.H. Younan, Dynamic modeling and response of soil-wall systems, ASCE J. Geotech. Eng. 120 (1994) 2155–2179.
[10] A. Elgamal, S. Alamapalli, P. Laak, Forced vibration of full-scale wall-backfill system, ASCE J. Geotech. Eng. 122 (1996) 849–858.
[11] K. Hatami, R.J. Bathurst, Effect of structural design on fundamental frequency of reinforced soil retaining walls, Soil Dyn. Earthq. Eng. 19 (20 0 0)
137–157.
[12] A. Ghanbari, E. Hoomaan, M. Mojallal, An analytycal method for calculating the natural frequency of retaining walls, Int. J. Civil Eng. 11 (2013) 1–9.
[13] O. Abbasi, A. Ghanbari, S.A. Hosseini, An analyticalmethod for calculating the natural frequency of reinforced retaining walls with soil structure inter-
action effect, Geosynth. Int. 21 (2014) 53–61.
[14] M.S. Ramezani, A. Ghanbari, S.A. Hosseini, Analytical method for calculating natural frequencies of geosynthetic-reinforced wall with full-height con-
crete facing, Geosynth. Int. (2016) Ahead of Print, doi:10.1680/jgein.16.0 0 011.
[15] G. Gazetas, P.N. Psarropoulos, I. Anastasopoulos, N. Gerolymos, Seismic behaviour of flexible retaining systems subjected to short-duration moderately
strong excitation, Soil Dyn. Earthq. Eng. 24 (2004) 537–550.
[16] D. Choudhury, S. Chatterjee, Dynamic active earth pressure on retaining structures, Sadhana Acad. Proc. Eng. Sci. 31 (2006) 721–730.
[17] L. Lanzoni, E. Radi, A. Tralli, On the seismic response of a flexible wall retaining a viscous poroelastic soil, Soil Dyn. Earthq. Eng. 27 (2007) 818–842.
[18] Y. Yesilce, H.H. Catal, Free vibration of piles embedded in soil having different modulus of subgrade reaction, Appl. Math. Model. 32 (20 08) 889–90 0.
[19] R. Kamgar, M.M. Saadatpour, A simple mathematical model for free vibration analysis of combined system consisting of framed tube, shear core, belt
truss and outrigger system with geometrical discontinuities, Appl. Math. Model. 36 (2012) 4918–4930.
[20] M. Malekinejad, R. Rahgozar, A Simple analytic method for computing the natural frequencies and mode shapes of tall buildings, Appl. Math. Model.
36 (2012) 3419–3432.
[21] J.-J. Wu, Free vibration analysis of a rigid bar supported by arbitrary elastic beams, Appl. Math. Model. 38 (2014) 1969–1982.
[22] R. Ansari, M. Faghih Shojaei, H. Rouhi, M. Hosseinzadeh, A novel variational numerical method for analyzing the free vibration of composite conical
shells, Appl. Math. Model. 39 (2015) 2849–2860.
[23] A. Komak Panaha, M. Yazdi, A. Ghalandarzadeh, Shaking table tests on soil retaining walls reinforced by polymeric strips, Geotext. Geomembr. 43
(2015) 148–161.
[24] T.k. Datta, Seismic Analysis Of Structures, John Wiley & Sons (Asia) Pte. Ltd., Singapore, 2010.
[25] S.S. Rao, Vibration of Continuous Systems, John Wiley & Sons, New Jersey, 2007.
[26] L. Meirovitch, Fundamentals of Vibrations, McGraw-Hill, Singapore, 2001.
[27] I.A. Karnovsky, O. Lebed, Formulas for Structural Dynamics Vibration, McGraw-Hill, New York, 2004.

You might also like