Journal of Rock Mechanics and Geotechnical Engineering 10 (2018) 347e357

Contents lists available at ScienceDirect

Journal of Rock Mechanics and

Geotechnical Engineering
journal homepage: www.rockgeotech.org

Full Length Article

Behavior of ring footing resting on reinforced sand subjected to

eccentric-inclined loading
Vaibhav Sharma*, Arvind Kumar
Dr. B.R. Ambedkar National Institute of Technology, Jalandhar, 144011, India

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

Article history: Ring footings are suitable for the structures like tall transmission towers, chimneys, silos and oil storages.
Received 14 July 2017 These types of structures are susceptible to horizontal loads (wind load) in addition to their dead weight.
Received in revised form In the literature, very little or no effort has been made to study the effect of ring footing resting on
21 September 2017
reinforced sand when subjected to eccentric, inclined and/or eccentric-inclined loadings. This paper aims
Accepted 31 October 2017
Available online 13 February 2018
to study the behavior of ring footing resting on loose sand and/or compacted randomly distributed fiber-
reinforced sand (RDFS) when subjected to eccentric (0B, 0.05B and 0.1B, where B is the outer diameter of
ring footing), inclined (0
, 5
, 10
, 15
, 5
, 10
and 15
) and eccentric-inclined loadings by using a
Keywords:
Soil reinforcement
finite element (FE) software PLAXIS 3D. The behavior of ring footing is studied by using a dimensionless
Ring footing factor called reduction factor (RF). The numerical model used in the PLAXIS 3D has been validated by
Numerical method conducting model plate load tests. Moreover, an empirical expression using regression analysis has been
Randomly distributed fibers presented which will be helpful in plotting a load-settlement curve for the ring footing.
Load inclination Ó 2018 Institute of Rock and Soil Mechanics, Chinese Academy of Sciences. Production and hosting by
Load eccentricity Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/
Model tests licenses/by-nc-nd/4.0/).
Reduction factor (RF)

1. Introduction
Ring footings are suitable for the structures which are circular in
plan and all the external or internal forces are transferred through
their wall into the foundation before they get transferred into the
ground. Tall transmission towers, chimneys, silos and oil storages
are suggested to be founded on ring foundations. As ring footings
are more suitable for these types of structures, which are supposed
to be taller than the existing structures around them, these struc-
tures are more susceptible to horizontal loads/forces like wind
loads. The ring foundation has to bear not only the vertical loads
but also the horizontal loads. The combination of horizontal and
vertical loads acting on the structure will result in an inclined load.
Moreover, the location of horizontal load (centroid) anywhere
along the height of the structure may produce eccentricity of the
resultant load’s point. Hence, ring footings can be subjected to both
eccentric and/or inclined loadings. Many investigators have studied
the behaviors of footings resting on unreinforced and/or reinforced
soil beds, other than ring footings subjected to eccentric and
* Corresponding author.
E-mail address: civil.vaibhav.sharma@gmail.com (V. Sharma).
Peer review under responsibility of Institute of Rock and Soil Mechanics, Chi-
nese Academy of Sciences.
inclined loadings (Meyerhof and Hanna, 1978; Sastry and Meyerhof,
1987; Saran and Agarwal, 1989, 1991; Taiebat and Carter, 2002;
Hjiaj et al., 2004; Patra et al., 2006; El Sawwaf, 2007, 2009; Saran
et al., 2007; Jao et al., 2008; Behera et al., 2013; Ornek, 2014;
Badakhshan and Noorzad, 2015, 2017a, b). From their studies, one
can see that as the eccentricity of the load increased, the bearing
capacity of the footing decreased significantly, and more tilting of
footing was also observed. Reinforcing the soil, however, can
improve the bearing capacity of footing, and reduce the settlement
of footing.
The majority of experimental, numerical and/or analytical
studies conducted on ring footings resting on reinforced and/or
unreinforced soil beds merely focused on concentric loading
(Bowles, 1987; Al-Sanad et al., 1993; Ismael, 1996; Boushehrian and
Hataf, 2003; Laman and Yildiz, 2003, 2007; Kumar and Ghosh,
2005; Zhao and Wang, 2008; Lee and Eun, 2009; Benmebarek et al.,
2012; Naderi and Hataf, 2014; Kumar and Chakraborty, 2015; Naseri
and Hosseininia, 2015; Lee et al., 2016; Keshavarz and Kumar, 2017;
Sharma and Kumar, 2017a). Bowles (1987) used finite element (FE)
method to calculate the bearing capacity and settlement of ring
footing. Laman and Yildiz (2003) conducted model tests on ring
footing resting on geogrid-reinforced sand. It was concluded that
the ring footing with ratio of inner to outer radius of ring (ri/
ro ¼ 0.3), which gave similar results as that of full circular footing,

https://doi.org/10.1016/j.jrmge.2017.11.005
1674-7755 Ó 2018 Institute of Rock and Soil Mechanics, Chinese Academy of Sciences. Production and hosting by Elsevier B.V. This is an open access article under the CC BY-
NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
348 V. Sharma, A. Kumar / Journal of Rock Mechanics and Geotechnical Engineering 10 (2018) 347e357

can be a good option from the economic point of view in practical
applications. Zhao and Wang (2008) calculated the bearing capacity
factor Ng for ring footings with different radius ratios using finite
difference method. More recently, Keshavarz and Kumar (2017)
used the stress characteristics method to obtain the bearing ca-
pacity factors Nc, Nq and Ng as functions of internal friction angle (f)
of soil and ri/ro of ring footing. From the literature review, it has
been found that the optimum ri/ro value of ring footing which be-
haves similar or better performance than circular footing lies in the
range of 0.2e0.4, which is not a unique value.
Very limited effort has been made to understand the behavior of
ring footing subjected to eccentric loading (El Sawwaf and Nazir,
2012). El Sawwaf and Nazir (2012) experimentally studied the
behavior of eccentrically loaded ring footing resting on a com-
pacted replaced geogrid-reinforced sand layer that overlies an
extended loose sand layer. It was concluded that the reinforcement
not only reduced the depth of replaced sand, but also increased the
bearing capacity of eccentrically loaded ring footing significantly.
Hence, there is a need to study the effects of load eccentricity
and inclination on the response of ring footings. This paper aims at
studying the behavior of ring footing, resting on unreinforced and/
or fiber-reinforced sand beds, when subjected to concentric,
eccentric and inclined loadings using FE package PLAXIS 3D.
Moreover, by replacing the loose sand up to some depth with
compacted randomly distributed fiber-reinforced sand (RDFS) (fi-
bers distributed randomly in the soil to obtain a homogeneous soil-
fiber mixture), the influence of sand reinforcement on the behavior
of ring footing can be studied. Ring footing used in this study has
radius ratio (ri/ro) of 0.4, with outer and inner radii of 150 mm and
60 mm, respectively. Fiber percentage and depth of reinforced sand
used in this study are 1% and 1B, respectively, where B is the outer
diameter of ring footing. Both parameters are kept constant,
because the effect of these parameters on ring footing subjected to
concentric loading only has already been documented (Sharma and
Kumar, 2017a, b). The load eccentricities of 0B, 0.05B and 0.1B, and
inclinations of 0
, 5
, 10
, 15
, 5
, 10
and 15
, are used in this
study. Furthermore, an empirical expression is developed using
regression technique which will be useful in plotting load-
settlement curve for eccentric, inclined and/or eccentric-inclined
loaded ring footings. Finally, the numerical results are validated
with the model plate load test results.
positive when it is measured anticlockwise and negative when it is
clockwise with respect to the Z-axis, as shown in Fig. 1a. The ratio of
difference between settlements at opposite sides to width of the
footing is taken as the tilt of the footing (Fig. 1b).
In order to apply vertical loads to the ring footing, uniformly
distributed load is converted into multiple closely spaced point
loads and then applied in a circular and symmetrical fashion, as
shown in Fig. 2. Similarly, load inclination and eccentricities are
given to these multiple closely spaced point loads to study their
effects on the behaviors of ring footings.
3. Numerical study
Numerical study of ring footings resting on unreinforced and/or
fiber-reinforced sand beds is carried out using FE package PLAXIS
3D. The width of the soil bed is taken as five times the footing
width, and the depth is taken as three and a half times the footing
width. Footing and soil bed model used in the FE analysis is shown
in Fig. 3, in which RD1 is the relative density of top RDFS layer, and
RD2 is the relative density of bottom unreinforced sand layer.
PLAXIS is designed for the two- or three-dimensional analyses
of deformation, stability and groundwater flow in geotechnical
engineering. Geotechnical applications require advanced constitu-
tive models for the simulation of the nonlinear, time-dependent,
and anisotropic behaviors of soils and/or rocks. Although the
modeling of soil itself is an important issue, many geotechnical
projects involve the modeling of structures and the interaction
between the structures and the soil. PLAXIS is equipped with

2. Load application, sign conventions and definitions for load

inclination and tilting

Fig. 1a shows the sign conventions for measuring the resultant

load (Qr) inclination with respect to the Z-axis. Load inclination is Fig. 2. Multiple closely spaced point loads applied on ring footing.

Fig. 1. Definitions for (a) load inclination and (b) footing tilt.
 V. Sharma, A. Kumar / Journal of Rock Mechanics and Geotechnical Engineering 10 (2018) 347e357 349

Table 2
Input parameters used for loose sand and RDFS.

Relative Content of Young’s Cohesion, Internal Unit

density fiber (%) modulus, c (kPa) friction weight
(%) Es (kPa) angle, f (
) (kN/m3)

30 0 6500 1 27 14.4
70 1 8900 10.4 35 15.7

6.0

5.5

Ulmate load (kN)

5.0 Selected mesh density
Fig. 3. Footing and soil bed model used in the numerical analysis.

4.5
features to deal with various aspects of complex geotechnical
structures.
Out of the various constitutive models available in the PLAXIS, 4.0
the elastic-perfectly plastic Mohr-Coulomb model and non- 0 20000 40000 60000 80000
associated flow rule are used in this study. The ring footing has Number of elements in mesh
been modeled as rigid plate and is considered to be very stiff and
rough in the analysis. Input parameters used in this study (provided Fig. 4. Sensitive study of mesh density for ring footing.
by the supplier) for ring footing are presented in Table 1. Various
input parameters of unreinforced and fiber-reinforced sands were
Table 3
obtained by conducting drained triaxial shear tests (Sharma and
Test scheme.
Kumar, 2017a) and are presented in Table 2. As fiber-
reinforcement cannot be modeled directly in PLAXIS, the soil is Series Sand RD1 (%) RD2 (%) e/Bb i (
) Number
condition of tests
assumed as cohesive-frictional sand. The term ‘cohesion’ in fiber-
reinforced sand does not correspond to the intermolecular cohe- I Sand only 30 30 0, 0.05, 0.1 0, 5, 10, 15 12
sion, but it corresponds to apparent cohesion due to the fact that II Sand only 30 30 0, 0.05, 0.1 5, 10, 15 9
III Sand þ fibersa 70 30 0, 0.05, 0.1 0, 5, 10, 15 12
sand grains adjacent to the fibers are gripped on fiber’s surface. IV Sand þ fibersa 70 30 0, 0.05, 0.1 5, 10, 15 9
PLAXIS incorporates fully automatic mesh generation proced- Total 42
ure, in which the model is divided into elements of basic type, and a
Fiber content is 1%, and the depth of fiber-reinforced sand is 1B from the footing
compatible structural elements. Five different mesh densities are base.
available in PLAXIS (very coarse, coarse, medium, fine, and very b
e is the load eccentricity.
fine). Preliminary trials were carried out on these five mesh den-
sities for footing resting on unreinforced sand (Fig. 4). There was
Table 4
not much significant difference between the medium and fine
Ultimate load (Qu) and RF obtained in the present study.
mesh densities. Hence, the medium mesh density was selected. In
areas where large stress concentration or large deformation Soil type i (
) Qu (kN) RF
gradient is expected, it is desirable to have finer element mesh. For e/B ¼ 0 e/B ¼ 0.05 e/B ¼ 0.1 e/B ¼ 0 e/B ¼ 0.05 e/B ¼ 0.1
this, automatic mesh refinement was done near the edges of Loose sand 0 4.6 4.3 3.5 1 0.93 0.76
foundation. Automatic boundary conditions were applied to the 5 4.4 3.92 3.2 0.96 0.85 0.7
model geometry. PLAXIS generates automatic boundary conditions 10 3.65 3.27 2.7 0.79 0.71 0.59
in which there is fully fixity at the base and smooth condition at the 15 2.8 2.5 2.2 0.61 0.54 0.48
5 4.55 4.25 3.6 0.99 0.92 0.78
vertical sides. The scheme for the FE analysis is presented in Table 3.
10 3.85 3.7 3.4 0.84 0.8 0.74
15 2.95 2.85 2.7 0.64 0.62 0.59
Compacted 0 43 40.9 38.1 1 0.95 0.89
4. Results and discussion RDFS 5 41.9 40.5 37.9 0.97 0.94 0.88
10 39.7 38.7 36.2 0.92 0.9 0.84
To study the effects of load inclination, eccentricity and fiber 15 37 36.2 33.5 0.86 0.84 0.78
reinforcement, a dimensionless term called reduction factor (RF) is 5 41.8 39.9 37.5 0.97 0.93 0.87
10 39 38 36.2 0.91 0.88 0.84
introduced, which is defined as the ratio of the ultimate load of an
15 37 36.3 34.8 0.86 0.84 0.81
eccentric, inclined or eccentric-inclined loaded foundation (Qu1) to

Table 1
Input parameters for ring footing. that of a foundation where applied load is vertical and concentric
Outer Inner Thickness of Unit weight of Poisson’s Young’s (Qu2) (Patra et al., 2006; Ornek, 2014). Its mathematical form can be
radius, ro radius, ri plate, d (m) plate, g (kN/m3) ratio, n modulus, Ep written as
(m) (m) (GPa) Qu1
RF ¼ (1)
0.3 0.12 0.025 78.5 0.3 210 Qu2
350 V. Sharma, A. Kumar / Journal of Rock Mechanics and Geotechnical Engineering 10 (2018) 347e357

1.2 Fig. 6a and b shows the effect of load inclination on the ultimate
load of ring footing resting on loose sand and compacted RDFS,
1.0 respectively. It can be inferred from Fig. 6 that as the load inclina-
tion increases, there is a significant decrease in the ultimate load of
ring footing. At the load inclinations of 5
, 10
and 15
, the ultimate
0.8
loads Qu for loose sand are 3.2 kN, 2.7 kN and 2.2 kN, respectively,
and those for compacted RDFS are 37.9 kN, 36.2 kN and 33.5 kN,

RF
0.6
e/B = 0.00 0.4
e/B = 0.05
e/B = 0.10
0.2
Loose sand
Compacted RDFS
0.0
20 15 10 5 0 -5
i (°)
Fig. 5. Effect of load inclination on RF for different eccentricities and sand beds.
The effective area of the ring footing will not be constant as the
load eccentricity increases. Therefore, this study emphasizes the
load-settlement behavior of ring footing, instead of pressure-
settlement behavior. Once the ultimate load is known, the ulti-
mate bearing capacity can be calculated easily, if the effective area
of ring footing is computed. The ultimate load (Qu) and RF obtained
for ring footing are listed in Table 4.
4.1. Effect of load inclination
Fig. 5 shows the effect of load inclination on RF for different
eccentricities and sand beds. It can be inferred from Fig. 5 that as
the load eccentricity increases, there is a significant change in the
RF, whether load inclination is positive or negative. RF values for
loose sand are 0.7, 0.59 and 0.48 corresponding to load inclinations
of 5
, 10
and 15
, respectively. Under the same conditions, RF
values for compacted RDFS are 0.88, 0.84 and 0.78, respectively.
Similarly, RF values for loose sand at load inclinations of 5
, 10

and 15
are 0.78, 0.74 and 0.59, respectively. Under the same
conditions, RF values for RDFS are 0.87, 0.84 and 0.81, respectively.
It can also be inferred from Fig. 5 that as the loose sand is replaced
with compacted RDFS, there is a slight decrease in RF which clearly
shows the effect of reinforcement.
respectively. Similarly, at the load inclinations of 5
, 10

and 15
, the ultimate loads Qu for loose sand are 3.6 kN, 3.4 kN
and 2.7 kN, respectively, and those for compacted RDFS are 37.5 kN,
36.2 kN and 34.8 kN, respectively. It can be observed from these
values and Fig. 6 that when the load inclination is negative, the
ultimate load of footing is larger as compared to its respective
positive load inclination (Similar behavior can be inferred from
Fig. 5 when RF is concerned). This behavior is due to the fact that as
-10 -15 -20
the load inclination is positive, the load is acting away from the
center of the footing. Similarly, when the load inclination is nega-
tive, the load is acting towards the center. This behavior is more
pronounced in loose sand which has been reduced to some extent
by the compacted RDFS.
Fig. 7 shows the load-tilt relationship of ring footing for loose
sand and compacted RDFS. It can be observed from Fig. 7a and c
that as the positive load inclination increases, both for loose sand
and compacted RDFS, there is a significant increase in tilt of footing
at different load eccentricities. In Fig. 7b, it can be seen that the tilt
of footing at load inclination of 5
is smaller than that at 0
. This is
due to the same reason that has been explained in the previous
paragraph, i.e. when the load inclination is negative, the load is
acting towards the center of the footing as compared to the positive
load inclination where the load is acting away from the center of
the footing. Fig. 7d has not shown this phenomenon which signifies
the effect of compacted RDFS.
Figs. 8 and 9 show the horizontal displacement vectors at ulti-
mate load for loose sand and compacted RDFS, respectively, when
e/B ¼ 0.1 and i ¼ 0
, 5
, 10
, 15
, 5
, 10
and 15
. It can be
inferred from Figs. 8 and 9 that as the load inclination increases,
there is a significant increase in horizontal displacement vector at
ultimate load. For loose sand at i ¼ 5
, 10
and 15
(e/B ¼ 0.1), the
horizontal displacement vectors at ultimate load are 116.2 mm
and 1.09 mm, 140.6 mm and 3.55 mm, and 134.6 mm
and 4.02 mm, respectively (positive and negative values show the
direction of movement of soil particles). Similarly, when the load
inclinations are 5
, 10
and 15
, their values are 98.22 mm
and 49.74 mm, 88.98 mm and 107.9 mm, and 42.6 mm

5 45
(a) Loose sand (b) Compacted RDFS

4
40
Ulmate load, Qu (kN)

3
Ulmate load, Qu (kN)

35
2

30
e/B = 0.00 1 e/B = 0.00
e/B = 0.05 e/B = 0.05
e/B = 0.10 e/B = 0.10
0 25
20 15 10 5 0 -5 -10 -15 -20 20 15 10 5 0 -5 -10 -15 -20
i (°) i (°)

Fig. 6. Effects of load inclination on ultimate load for (a) loose sand and (b) compacted RDFS.
 V. Sharma, A. Kumar / Journal of Rock Mechanics and Geotechnical Engineering 10 (2018) 347e357 351

Load (kN) Load (kN)

0 1 2 3 4 5 0 1 2 3 4 5
0.0 0.0

(a) (b)
-0.5 -0.5
Loose sand d
Loose sand
-1.0 -1.0

Tilt
Tilt

-1.5 i = 0° -1.5
i = 0°
i = 5° i = -5°
-2.0 i = 10° -2.0
i = -10°
i = 15° i = -15°
-2.5 e/B =0.005 -2.5
e/BB =0.05
e/B = 0.10
e/BB = 0.10
-3.0 -3.0

Load (kN) Load (kN))

0 10 20 30 40 50
5 0 10 20 30 40 50
0 0
(c) (d)
-5 -5
Compacted RDFS ed RDFS
Compacte
-10 -10
Tilt

Tilt
-15 i = 0° -15 i = 0°
i = 5° i = -5°
-20 i = 10°° -20 i = -10°
i = 15°° i = -15°
-25 e/B =0 0.05 -25 e//B =0.05
e/B = 0.10 e//B = 0.10
-30 -30

Fig. 7. Effects of load inclination on tilt of footing for (a, b) loose sand and (c, d) compacted RDFS.

and 155.9 mm, respectively. For the compacted RDFS at i ¼ 5
, 10

and 15
(e/B ¼ 0.1), the horizontal displacement vectors at ultimate
load are 463.6 mm and 182.5 mm, 697 mm and 105.4 mm, and
1094 mm and 32.13 mm, respectively. Similarly, when the load
inclinations are 5
, 10
and 15
, their values are 236.2 mm
and 307.7 mm, 229.7 mm and 454.7 mm, and 118.9 mm
and 418.2 mm, respectively. It can be interpreted from these
values that as the load inclination changes from positive to nega-
tive, there is a decrease in horizontal displacement of sand in
positive X direction and minor increase in negative X direction. This
effect is more significant when sand is loose and unreinforced, and
it has been reduced to some extent when sand is replaced with
compacted RDFS.
4.2. Effect of load eccentricity
Fig. 10a and b shows the effect of load eccentricity on RF for
loose sand and compacted RDFS, respectively. It can be inferred
from Fig. 10 that as the load eccentricity increases, RF decreases
significantly. As e/B changes from 0 to 0.05 and then to 0.1, RF values
for loose sand are obtained as follows: 1, 0.93 and 0.76 (for i ¼ 0
);
0.96, 0.85 and 0.7 (for i ¼ 5
); and 0.99, 0.92 and 0.78 (for i ¼ 5
),
respectively. Similarly, RF values for compacted RDFS are obtained
as follows: 1, 0.95 and 0.89 (for i ¼ 0
); 0.97, 0.94 and 0.88 (for
i ¼ 5
); and 0.97, 0.93 and 0.87 (for i ¼ 5
), respectively. Similar
decreasing phenomenon has been observed for other load in-
clinations considered in this study. One interesting behavior can be
observed from comparison of Fig. 10a with Fig. 10b that as the load
eccentricity increases, the decrease in RF is more pronounced in
loose sand (Fig. 10a) than that in compacted RDFS (Fig. 10b). This
behavior completely justifies the effect of reinforcement on the
behavior of eccentrically loaded footing.
Fig. 11a and b shows the effect of load eccentricity on ultimate
load of footing resting on loose sand and compacted RDFS,
respectively. It can be seen from Fig. 11 that as the load eccentricity
increases, there is a significant decrease in Qu. When e/B changes
from 0 to 0.05 and then to 0.1, Qu values for loose sand are obtained
as follows: 4.6 kN, 4.3 kN and 3.5 kN (for i ¼ 0
); 4.4 kN, 3.92 kN and
3.2 kN (for i ¼ 5
); and 4.55 kN, 4.25 kN and 3.6 kN (for i ¼ 5
),
respectively. Similarly, Qu values for compacted RDFS are obtained
as follows: 43 kN, 40.9 kN and 38.1 kN (for i ¼ 0
); 41.9 kN, 40.5 kN
and 37.9 kN (for i ¼ 5
); and 41.8 kN, 39.9 kN and 37.5 kN (for
i ¼ 5
), respectively. For other load inclinations, similar decreasing
behavior in Qu has been observed. The effect of reinforcement has
also been observed by comparing the Qu values of Fig. 11a with
Fig. 11b.
Effect of load eccentricity on tilt of footing can be observed from
Fig. 7. One can see that the tilt of footing increases with the load
eccentricity. When comparing the load-tilt curves of loose sand
(Fig. 7a and b) with those of compacted RDFS (Fig. 7c and d),
reinforcement does not show much significant reduction effect on
tilt. It can be due to the model selection or input parameters used
for reinforcing the sand.
Figs. 12a and b, 8f and g show the effects of load eccentricity on
horizontal displacement vectors at ultimate load for loose sand.
Similarly, Figs. 12c and d, 9f and g show the effects of load eccen-
tricity on horizontal displacement vectors at ultimate load for
compacted RDFS. It can be inferred from these figures that as the
load eccentricity increases, the horizontal displacement vectors at
ultimate load increase. When i ¼ 15
for loose sand, the horizontal
displacement vectors at ultimate load for e/B ¼ 0.05 and 0.1 are
352 V. Sharma, A. Kumar / Journal of Rock Mechanics and Geotechnical Engineering 10 (2018) 347e357
Fig. 8. Horizontal displacement vectors at ultimate load for loose sand at e/B ¼ 0.1 and (a) i ¼ 0
, (b) i ¼ 5
, (c) i ¼ 5
, (d) i ¼ 10
, (e) i ¼ 10
, (f) i ¼ 15
, and (g) i ¼ 15
.
121.4 mm and 17.09 mm, and 134.6 mm and 4.02 mm, respec-
tively. When i ¼ 15
for loose sand, the horizontal displacement
vectors at ultimate load for e/B ¼ 0.05 and 0.1 are 44.32 mm
and 165.6 mm, and 42.6 mm and 155.9 mm, respectively. When
i ¼ 15
for compacted RDFS, the horizontal displacement vectors at
ultimate load for e/B ¼ 0.05 and 0.1 are 752.7 mm and 24.1 mm,
and 1094 mm and 32.13 mm, respectively. When i ¼ 15
for
compacted RDFS, the horizontal displacement vectors at ultimate
load for e/B ¼ 0.05 and 0.1 are 21.4 mm and 481.9 mm, and
118.9 mm and 418.2 mm, respectively.
4.3. Regression analysis
The load-settlement curves for footing under consideration can
be plotted using PLAXIS 3D. If the software is not available, the
load-settlement curves for the footing used in this study can be
plotted by using Eq. (2), which is developed using regression
analysis on the load-settlement data obtained from the FE analysis
of ring footing by considering all of the vital bearing capacity
parameters.
Q ¼ 0:1567Qs1:1457 ðS=BÞ0:0884 0:998i 0:1743e=B 1:0551RD1 (2)
where Q is the load corresponding to settlement (S), Qs and S are
respectively the load and settlement of loose sand under central
point loading without inclination. The coefficient of determination
(R2) is found to be 0.9779 for Eq. (2). Fig. 13 illustrates a typical
graph of observed versus predicted data of ultimate load of ring
footing resting on loose sand and compacted RDFS.
Regression analysis is one of the available empirical estimation
methods which are common in civil engineering (Bera et al., 2005).
In this technique, a function is assumed to estimate the relationship
between the independent and dependent variables. The co-
efficients of the assumed function are determined by the least
square method using observed data. A general linear model is
assumed in the following form:
y ¼ x1 x1 þ x2 x2 þ / þ xp xp (3)
 V. Sharma, A. Kumar / Journal of Rock Mechanics and Geotechnical Engineering 10 (2018) 347e357 353

Fig. 9. Horizontal displacement vectors at ultimate load for compacted RDFS at e/B ¼ 0.1 and (a) i ¼ 0
, (b) i ¼ 5
, (c) i ¼ 5
, (d) i ¼ 10
, (e) i ¼ 10
, (f) i ¼ 15
, and (g) i ¼ 15
.

where y is the dependent variable; x1, x2, ., xp are the independent X

p
 
variables; x1, x2, ., xp are the unknown parameters; and p is the yk ¼ xi xij ðk ¼ 1; 2; .; nÞ (5)
number of independent variables. l¼1

In real practice, n observations would be available on y with

Eq. (5) can be written in the matrix form as
corresponding n observations on each of the p independent vari-
ables. The n numbers of equations are written as
9
y1 ¼ x1 x11 þ x2 x12 þ . þ xp x1p >
> 8 9 8 98 9
=
y2 ¼ x1 x21 þ x2 x22 þ . þ xp x2p >
> y1 >
> >
> x11 x12 x13 / x1p >
>> x1 >
(4) >
> > > >>
>x > >
« >
> < y2 >
= >
< x21 x22 x23 / =>
x2p >< 2> =
; x3
yn ¼ x1 xn1 þ x2 xn2 þ . þ xp xnp y3 ¼ x31 x32 x33 / x3p (6)
>
> > > >> >
> « >
> >
>
>
>
> « « « / « >>>
>> « >>
: ; : ;>
: >
where yk is the kth observation on y and xkl is the kth observation on yn xn1 xn2 xn3 / xnp xp ;
the lth independent variable. Eq. (4) can be rewritten as
354 V. Sharma, A. Kumar / Journal of Rock Mechanics and Geotechnical Engineering 10 (2018) 347e357

1.5 1.5
i = -15° i = -10° i = -5° i = 0° i = -15° i = -10° i = -5° i = 0°
i = 5° i = 10° i = 15° i = 5° i = 10° i = 15°
(a) Loose sand (b) Compacted RDFS
1.0 1.0
RF

RF
0.5 0.5

0.0 0.0
0.00 0.05 0.10 0.00 0.05 0.10
e/B e/B

Fig. 10. Effects of load eccentricity on RF for (a) loose sand and (b) compacted RDFS.

6 60
i = -15° i = -10° i = -5° i = 0° i = -15° i = -10° i = -5° i = 0°
5 i = 5° i = 10° i = 15° 50 i = 5° i = 10° i = 15°
UlƟmate load, Qu (kN)

(a) Loose sand UlƟmate load, Qu (kN) (b) Compacted RDFS

4 40

3 30

2 20

1 10

0 0
0.00 0.05 0.10 0.00 0.05 0.10
e/B e/B

Fig. 11. Effects of load eccentricity on ultimate load of footing resting on (a) loose sand and (b) compacted RDFS.

(a) (b)

(c) (d)

Fig. 12. Effects of load eccentricity on horizontal displacement vectors at ultimate load for (a, c) loose sand with i ¼ 15
and 15
, respectively, and (b, d) compacted RDFS with
i ¼ 15
and 15
, respectively (e/B ¼ 0.05).
 V. Sharma, A. Kumar / Journal of Rock Mechanics and Geotechnical Engineering 10 (2018) 347e357 355

carried out on both unreinforced and fiber-reinforced sands with

eccentric loading (e/B ¼ 0, 0.05 and 0.1). The testing tank was 1.5 m
wide, 1.5 m long and 1 m deep. Model footing was made up of mild
steel with internal and external diameters of 0.12 m and 0.3 m,
respectively. Detailed information about the testing assembly and
sample preparation for model plate load test can refer to Sharma
and Kumar (2017b). The load-settlement curves obtained from
model plate load tests are shown in Fig. 14. Fig. 15 shows the
comparison of numerical findings with model plate load test re-
sults. It can be inferred from Fig. 15 that there is a good agreement
between the numerical and model test results which validates the
numerical findings of this study.

6. Scale effect

PLAXIS 3D software has been used to study the scale effect of

Fig. 13. Observed load versus predicted load using Eq. (2). model footing resting on fiber-reinforced sand. Model footings,
with various sizes (outer diameter of model footing, Bp), used in the
i.e.
analysis are 30 cm, 45 cm, 60 cm and 75 cm. Similarly, prototype
y x x foundations, with various sizes (outer diameter of prototype
¼ foundation, Bf), used in the analysis are 100 cm, 210 cm, 300 cm and
n1 np p1
400 cm. Depending on the sizes of model/prototype footings, the
where y is the n  1 vector of observations, x is a matrix (n  p) size of soil mass has been modified accordingly. Thicknesses of
made up of observations on each of p independent variables, and x model footings and prototype foundations, and tank dimensions
is the p  1 vector of unknown parameters. To have an intercept have also been modified on the basis of their scale modification.
term in Eq. (6), it is necessary that xi1 ¼ 1, while x1 becomes the The horizontal and vertical extents of the tank are taken as 5B and
intercept.
The selection of an appropriate model highly depends on the
judgment of the researchers, including incorporation of relevant
theory and knowledge of the subject under consideration, and
careful examination of scatter plots, residual plots and regression
diagnostics (Dielman, 2001). Partial information about the rela-
tionship between dependent and independent variables can be
obtained by forming a scatter plots matrix (Dielman, 2001). In the
present study, a nonlinear power model is chosen, on the basis of
the scatter plots matrix, and can be written as

x x x
y ¼ x0 x11 x22 /xpp (7)

This nonlinear regression model can be transformed into linear

model by using a logarithmic transformation as

log10 y ¼ log10 x0 þ x1 log10 x1 þ . þ xp log10 xp (8)

Analyzing the scatter plot matrix obtained from the FE data, a

nonlinear power model has been chosen as Fig. 14. Load-settlement curves for model plate load tests.

Q ¼ x0 ðQs Þx1 ðS=BÞx2 ðx3 Þ ðx4 Þ ðx5 Þ 1 ðx6 Þ

i e=B RD f
(9)
1.00
Eq. (9) can be transformed to a linear equation by taking loga-
rithm on both sides so that standard regression techniques can be 0.95
used. The transformed linear equation can be expressed as
0.90
log10 Q ¼ log10 x0 þ x1 log10 Qs þ x2 log10 ðs=BÞ þ i log10 x3
þ ðe=BÞlog10 x4 þ RD1 log10 x5 þ f log10 x6
RF

0.85
(10)
After applying standard regression techniques, all the unknown 0.80
Loose sand
parameters (x0, x1, x3, ., x6) can be obtained and can be referred to
0.75 Reinforced sand
Eq. (2).
Numerical study
Model test
0.70
5. Comparison with the model plate load tests
0.00 0.05 0.10 0.15
e/B
In order to validate the numerical findings, some laboratory
model plate load tests were performed. The model tests were Fig. 15. Comparison of numerical and experimental results.
356 V. Sharma, A. Kumar / Journal of Rock Mechanics and Geotechnical Engineering 10 (2018) 347e357
3.5B, respectively. Other parameters such as unit weight, Poisson’s
ratio and Young’s modulus of model footings and prototype foun-
dations are kept constant, as listed in Table 1.
Two dimensionless factors have been developed using regres-
sion analysis and are described along with their equations as
 0:559
Qf B (6) An empirical expression obtained using regression analysis
¼ 0:796 f 1:004RD1 0:9976RD2 0:945h1 =B 1:1507Fc
Qp Bp
(11)
 1:5476
Sf B
¼ 0:7327 f 1:006RD1 0:9965RD2 1:051h1 =B 1:083Fc
Sp Bp
(12)
where Qf/Qp is the load factor defined as the ratio of load
acting on prototype foundation to that acting on model footing,
Sf/Sp is the settlement factor defined as the ratio of settlement
corresponding to the load acting on prototype foundation to
that corresponding to load acting on model footing, h1 is the
thickness of top fiber-reinforced sand layer, and Fc is the per-
centage of fibers in the sand (1% in this study). The coefficients
of determination (R2) for Eqs. (11) and (12) are 0.971 and 0.965,
respectively.
In this study, RD1 will be the same as RD2 when studying footing
resting on loose unreinforced sand. RD1 will become 70% and RD2
will be 30% when studying footing resting on RDFS layer overlying
unreinforced sand. Both h1/B and Fc will be taken as 0 for footing
resting on loose unreinforced sand and 1 for footing resting on
RDFS layer overlying unreinforced sand.
7. Conclusions
An effort has been made to study the effects of load inclination
and eccentricity on the behavior of ring footing resting on unre-
inforced and/or fiber-reinforced sands. A numerical approach has
been advised to model fiber-reinforced sand. Moreover, numerical
results of this study have been compared with the model plate load
test results and the following conclusions can be drawn from this
study:
(1) As the load inclination increases from 0
to 15
, the ultimate
load of footing resting on loose sand is reduced by 41%e52%.
Similarly, this reduction is in the range of 19%e22% when
footing is resting on RDFS layer overlying unreinforced sand.
(2) As the load eccentricity (e/B) increases from 0 to 0.1, the ul-
timate load of footing resting on loose sand is reduced by
36%e52%. Similarly, this reduction is in the range of 14%e22%
when footing is resting on RDFS layer overlying unreinforced
sand.
(3) It has been interpreted from the horizontal displacement
vectors at ultimate load that as the load inclination changes
from positive to negative, there is a decrease in horizontal
displacement of sand in positive X direction and minor in-
crease in negative X direction. This effect is more significant
when sand is loose and unreinforced, and it has been
reduced to some extent when sand is replaced with com-
pacted RDFS.
(4) When the load inclination is negative, the RF and Qu of
footing are larger as compared to those at its respective
positive load inclination, especially when the load inclination
changes from 5
to 5
. This behavior is due to the fact that
as the load inclination is positive, the load is acting away
from the center of the footing.
(5) Comparing the effect of load eccentricity on RF for loose sand
and compacted RDFS shows that as the load eccentricity
increases, the decrease in RF is more pronounced in loose
sand than that in compacted RDFS. This behavior completely
justifies the effect of reinforcement on the behavior of
eccentrically loaded footing.
can be adopted to plot the load-settlement curve for ring
footing (for any condition used in the present study), if the
load-settlement data of ring footing resting on loose sand
with central point load are available.
Conflict of interest
The authors wish to confirm that there are no known conflicts of
interest associated with this publication and there has been no
significant financial support for this work that could have influ-
enced its outcome.
References
Al-Sanad HA, Ismael NF, Brenner RP. Settlement of circular and ring plates in very
dense calcareous sands. Journal of Geotechnical Engineering 1993;119(4):622e
38.
Badakhshan E, Noorzad A. Load eccentricity effects on behavior of circular footings
reinforced with geogrid sheets. Journal of Rock Mechanics and Geotechnical
Engineering 2015;7(6):691e9.
Badakhshan E, Noorzad A. Effect of footing shape and load eccentricity on behavior
of geosynthetic reinforced sand bed. Geotextiles and Geomembranes
2017a;45(2):58e67.
Badakhshan E, Noorzad A. A simplified method for prediction of ultimate bearing
capacity of eccentrically loaded foundation on geogrid reinforced sand bed. In-
ternational Journal of Geosynthetics and Ground Engineering 2017b;3(2):1e15.
Behera RN, Patra CR, Sivakugan N, Das BM. Prediction of ultimate bearing capacity
of eccentrically inclined loaded strip footing by ANN - part II. International
Journal of Geotechnical Engineering 2013;7(2):165e72.
Benmebarek S, Remadna MS, Benmebarek N, Belounar L. Numerical evaluation of
the bearing capacity factor N0 g of ring footings. Computers and Geotechnics
2012;44:132e8.
Bera AK, Ghosh A, Ghosh A. Regression model for bearing capacity of a square
footing on reinforced pond ash. Geotextiles and Geomembranes 2005;23(3):
261e85.
Boushehrian JH, Hataf N. Experimental and numerical investigation of the bearing
capacity of model circular and ring footings on reinforced sand. Geotextiles and
Geomembranes 2003;21(4):241e56.
Bowles JE. Elastic foundation settlements on sand deposits. Journal of Geotechnical
Engineering 1987;113(8):846e60.
Dielman TE. Applied regression analysis for business and economics. Duxbury, USA:
Thomson Learning, Inc.; 2001.
El Sawwaf MA. Behavior of strip footing on geogrid-reinforced sand over a soft clay
slope. Geotextiles and Geomembranes 2007;25(1):50e60.
El Sawwaf M. Experimental and numerical study of eccentrically loaded strip
footings resting on reinforced sand. Journal of Geotechnical and Geo-
environmental Engineering 2009;135(10):1509e18.
El Sawwaf M, Nazir A. Behavior of eccentrically loaded small-scale ring footings
resting on reinforced layered soil. Journal of Geotechnical and Geo-
environmental Engineering 2012;138(3):376e84.
Hjiaj M, Lyamin AV, Sloan SW. Bearing capacity of a cohesive-frictional soil under non-
eccentric inclined loading. Computers and Geotechnics 2004;31(6):491e516.
Ismael NF. Loading tests on circular and ring plates in very dense cemented sands.
Journal of Geotechnical Engineering 1996;122(4):281e7.
Jao M, Ahmed F, Muninarayana G, Wang MC. Stability of eccentrically loaded
footings on slopes. Geomechanics and Geoengineering 2008;3(2):107e11.
Keshavarz A, Kumar J. Bearing capacity computation for a ring foundation using the
stress characteristics method. Computers and Geotechnics 2017;89:33e42.
Kumar J, Chakraborty M. Bearing capacity factors for ring foundations. Journal of
Geotechnical and Geoenvironmental Engineering 2015;141(10). https://doi.org/
10.1061/(ASCE)GT.1943-5606.0001345.
Kumar J, Ghosh P. Bearing capacity factor Ng for ring footings using the method of
characteristics. Canadian Geotechnical Journal 2005;42(5):1474e84.
Laman M, Yildiz A. Model studies of ring foundations on geogrid-reinforced sand.
Geosynthetics International 2003;10(5):142e52.
Laman M, Yildiz A. Numerical studies of ring foundations on geogrid-reinforced
sand. Geosynthetics International 2007;14(2):52e64.
Lee J, Eun J. Estimation of bearing capacity for multiple footings in sand. Computers
and Geotechnics 2009;36(6):1000e8.
 V. Sharma, A. Kumar / Journal of Rock Mechanics and Geotechnical Engineering 10 (2018) 347e357 357

Lee JK, Jeong S, Lee S. Undrained bearing capacity factors for ring footings in het- Vaibhav Sharma obtained his PhD from Dr B.R. Ambedkar
erogeneous soil. Computers and Geotechnics 2016;75:103e11. National Institute of Technology, Jalandhar, Punjab, India
Meyerhof GG, Hanna AM. Ultimate bearing capacity of foundations on layered in 2018. He has published four research papers in reputed
soils under inclined load. Canadian Geotechnical Journal 1978;15(4): journals. His areas of interest cover ground improvement
565e72. techniques, numerical modeling in geotechnical engi-
Naderi E, Hataf N. Model testing and numerical investigation of interference effect neering, and environmental-geotechnology. He is pres-
of closely spaced ring and circular footings on reinforced sand. Geotextiles and ently working as assistant professor in Dr. B.R. Ambedkar
Geomembranes 2014;42(3):191e200. National Institute of Technology, Jalandhar, Punjab, India.
Naseri M, Hosseininia ES. Elastic settlement of ring foundations. Soils and Foun-
dations 2015;55(2):284e95.
Ornek M. Estimation of ultimate loads of eccentric-inclined loaded strip footings
rested on sandy soils. Neural Computing and Applications 2014;25(1):39e54.
Patra CR, Das BM, Bhoi M, Shin EC. Eccentrically loaded strip foundation on geogrid-
reinforced sand. Geotextiles and Geomembranes 2006;24(4):254e9.
Saran S, Agarwal RK. Eccentrically-obliquely loaded footing. Journal of Geotechnical
Engineering 1989;115(11):1673e80.
Saran S, Agarwal RK. Bearing capacity of eccentrically obliquely loaded footing.
Journal of Geotechnical Engineering 1991;117(11). 1669e9.
Saran S, Kumar S, Garg KG, Kumar A. Analysis of square and rectangular footings
Arvind Kumar obtained his PhD from University of Roor-
subjected to eccentric-inclined load resting on reinforced sand. Geotechnican
kee (now IIT Roorkee), India. He joined Dr. B.R. Ambedkar
and Geological Engineering 2007;25(1):123e37.
National Institute of Technology, Jalandhar, India in 1991
Sastry VVRN, Meyerhof GG. Inclination factors for strip footings. Journal of
and became professor in 2007. He is active consultant. He
Geotechnical Engineering 1987;113(5):524e7.
has published over 90 research papers in national and
Sharma V, Kumar A. Strength and bearing capacity of ring footings resting on fibre-
international journals and conferences. He is member of
reinforced sand. International Journal of Geosynthetics and Ground Engineering
Indian Geotechnical Society, International Society of Soil
2017a;3:1e17. https://doi.org/10.1007/s40891-017-0086-6.
Mechanics and Geotechnical Engineering, Indian Roads
Sharma V, Kumar A. Influence of relative density of soil on performance of fiber-
Congress, Institution of Engineers, and Indian Society of
reinforced soil foundations. Geotextiles and Geomembranes 2017b;45(5):
Technical Education.
499e507. https://doi.org/10.1016/j.geotexmem.2017.06.004.
Taiebat HA, Carter JP. Bearing capacity of strip and circular foundations on un-
drained clay subjected to eccentric loads. Géotechnique 2002;52(1):61e4.
Zhao L, Wang JH. Vertical bearing capacity for ring footings. Computers and Geo-
technics 2008;35(2):292e304.