Engineering Structures 92 (2015) 114

Contents lists available at ScienceDirect

Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct

Testing and nite element analysis of reinforced concrete column

footings failing by punching shear
Todor Vacev a,, Zoran Bonic a,, Verka Prolovic a, Neboja Davidovic a, Dragan Lukic b
a
b

Faculty of Civil Engineering and Architecture, University of Ni, Aleksandra Medvedeva 14, Ni, Serbia
Faculty of Civil Engineering Subotica, University of Novi Sad, Kozaracka 2, Subotica, Serbia

a r t i c l e

i n f o

Article history:
Received 22 June 2014
Revised 19 February 2015
Accepted 20 February 2015
Available online 17 March 2015
Keywords:
Reinforced concrete
Column footing
Punching
Test
FEA
ANSYS

a b s t r a c t
Finite element analysis (FEA) and engineering software is increasingly used in modelling of different
structures and in analyzing their behaviour. The subject of this paper was an analysis of behaviour of
reinforced concrete column footing laid on deformable subgrade and loaded by concentrated load until
failure. The modelling and 3D nonlinear analysis were implemented by applying nite element method
(FEM) and using software package ANSYS 14.5. Field test data were used for calibration of the FE model
and validation of the adopted parameters for all materials. The comparison of the eld test results
and FEA results showed good agreement, but also revealed some questions regarding FEA, and especially
concrete crushing.
2015 Elsevier Ltd. All rights reserved.

1. Introduction
Design of shallow foundations, especially column footings, in
most cases involves calculations in which the footings are considered as stiff, and distribution of contact pressures is adopted as linear. In such calculations, the important part represents checking of
punching of the column through the footing, in cases with, and
without reinforcement that accepts shear in the column zone.
Such control, as well as the punching control of reinforced concrete
at slabs, is mostly based on experimental research. Behaviour of
column footings laid on deformable subgrade was, as a rule, investigated using laboratory tests, which caused difculties in simulation of real conditions. Due to the complexity of the problem, both
in mechanical and in organization aspects, such experiments on
real footing structures were very rare hitherto. Finite element analysis (FEA) of 3D models using some of the available software packages represents a good way to overcome those problems. Besides
the signicant saving of time, material, and labour, this method
enables unobstructed tracking of behaviour of all materials in the
foundation structure under load (soil, concrete, and reinforcement). In this paper a FEA of an isolated reinforced concrete (RC)
Corresponding authors at: Faculty of Civil Engineering and Architecture,
University of Ni, A. Medvedeva 14, 18000 Ni, Serbia. Tel.: +381 631045368
(T. Vacev), +381 631048217 (Z. Bonic).
E-mail addresses: ttomas@eunet.rs (T. Vacev), zokibon@yahoo.com (Z. Bonic).
http://dx.doi.org/10.1016/j.engstruct.2015.02.027
0141-0296/ 2015 Elsevier Ltd. All rights reserved.

column footing on gravel ground is shown. FEA was done using

ANSYS software, which allows successful simulation of behaving
of materials such as concrete, steel, and soil.

2. Experimental analysis of the column footing

2.1. Aim of the experimental analysis
The aim of the experimental analysis was to determine
behaviour of isolated RC column footings rested on subgrade soil
and loaded by proposed external load until their failure. Thereat,
it is important to determine the occurrence and mechanism
of cracking and crushing development, and the way of stress
distribution in reinforcement and in concrete of the footing. Also,
it would be important to analyze inuence of the factors like: concrete strength at compression, applied percentage of reinforcing,
emplacement and diameter of the reinforcement, type and characteristics of the soil, unevenness of the contact pressures, etc. The
key result of the performed parametric study should be determining of the factors whose inuence in column footing calculation is
dominant.
The experimental research was performed by the authors of this
paper, and it was conducted from 2009 to 2014. The experimental
analysis programme included construction of complete foundation
structure in situ, consisting of prepared subgrade soil with

T. Vacev et al. / Engineering Structures 92 (2015) 114

2.2. Experimental setup and material data

2.3. Testing procedure

The experimental analysis was conducted by placing the footing
on the soil surface and by loading it using vertical centric force,
which was applied by a hydraulic jack having capacity of
1000 kN and monitored at every second (dynamometer HBM
RTN, Fig. 5). The hydraulic jack and dynamometer were positioned
between the steel crossbeam and the footing (Fig. 4b and 5). The
load was applied in the load steps of approximately 20 kN, and
kept constant at every load step until total consolidation of the
subgrade soil for that load step. The consolidation was veried
by observed stagnating of the vertical displacements of the footing.

175

3..6O8
2..8O8

25

1..7O8

1..7O8
120

120

120

120

120

120

65

SG5 SG6

120

120

175

120

SG4

120

337

120

65

65

25

100

For the purpose of the experiment, a testing steel frame was

made, and its role was to accept the reactive force of the hydraulic
jack that was used to load the footing.
The test frame (Fig. 1) was laid down in the foundation pit measuring 4.0  5.0 m in layout, and 3.0 m in depth. The structure of
the frame and its dimensions allow for unimpeded formation of
sliding surfaces in the soil underneath the footing, in case sufciently high failure load is reached. In this way, we made a step further in respect to the earlier laboratory experiments in the
literature, because the footing testing is conducted within the completely realistic boundary conditions in terms of the soil.
Simultaneously, comparison and verication of earlier tests in
laboratories (from the literature) with the in situ tests is provided.
The excavated material was replaced by river aggregate with controlled density and granulometric composition (Fig. 3a). The prepared mixture was laid in layers 30 cm thick, with compacting of
every layer using vibrating plate (Fig. 3b). After compacting of
every layer, the mixture density was checked using plate bearing
test (Fig. 3c). The measured mean values of the modulus of compressibility (Ms) in the layers were in range of 43.366.7 MPa, thus
corresponding to normal compaction of subsoil. The compaction of
the subgrade soil was controlled before testing of each footing. It
ranged between 39.5 MPa and 76.7 MPa, and the corresponding
modulus of elasticity of the subgrade soil was calculated.
The tested footing (Fig. 2) was loaded at its column top by vertical axial load using the hydraulic jack (Figs. 4b and 5). The load
was increased until the punching of column through the footing
occurred. The dimensions of the specimens for the proposed

testing programme were 85  85 cm in layout and correspond to

the earlier experiments of Hegger et al. and Kinnunen et al. [1
4]. The footing depths were 12.5 cm to 25 cm and the diameters
of applied steel reinforcing bars were 8 mm. The reinforcement
ratio (q) for all the footings was approximately 0.4%. The properties
of the used steel were determined on three samples. The obtained
values were as follows: tensile strength fsu = 653 MPa, yield point
fsy = 570 MPa, and corresponding yield strain e  2.7 .
Relevant characteristics of the footing that was tested and analyzed by FEA are given in Table 1. Concrete compressive strength
was obtained at the time of testing using three cube specimens
with edge length of 15 cm and one standard cylinder specimen,
and all averaged values were converted to a standard cylinder.

200

prescribed geotechnical characteristics and test specimens column footings with proposed dimensions and dened characteristics of the concrete and the reinforcement.

15
33 33 33 33 33
20

338

SG7

65

120

SG8
SG9

337

175

338

850

Fig. 1. Test frame and the foundation pit.

Fig. 2. Column footing (dimensions in mm); SG4. . .SG9 strain gages for
reinforcement strains.

T. Vacev et al. / Engineering Structures 92 (2015) 114

GRAIN SIZE DISTRIBUTION DIAGRAM
100,0

Gravel

Sand

Silt

Clay

90,0
80,0

Percent finer [%]

70,0
60,0
50,0
40,0
30,0
20,0
10,0
0,0
100

10

0,1

0,01

0,001

0,0001

Grain size D [mm]

(a)

(b)

(c)

Fig. 3. Soil at test site: (a) granulometric composition; (b) compacting of soil using vibrating plate; (c) density checking using plate bearing test.

Fig. 4. Experimental setup; (a) scheme (dimensions in mm); (b) setup in situ.

During the experiment, the following parameters were registered every one second: vertical displacements of the control
points, intensity of the applied force, and contact pressure values
beneath the footing measured by the soil pressure cells having
the diameter of 100 mm and capacity of 3.0 MPa.
3. Finite element analysis (FEA)
For the numerical analysis of the tested footing, the generalpurpose FEA software ANSYS was chosen. This software is used
both for linear and for nonlinear analysis of static and dynamic
problems. For solving of nonlinear problems, ANSYS uses the
NewtonRaphson method, in which the total load is divided into
series of load increments, until the nal convergence of solution.
For numerical modelling of the foundations laid on deformable
subgrade, nonlinear behaviour of the soil, reinforcement, and concrete were enabled, as well as concrete cracking at tension, and
crushing at compression.
3.1. Material model for concrete

Fig. 5. Experimental setup details: hydraulic jack with dynamometer and electronic
tranducers.

Two control points were observed: at the footing corner and on the
footing column top. The displacements of these points were monitored by the electronic transducer HBM-W50 (Fig. 5).

Concrete material shows quasi-brittle behaviour, different at

tension and compression. Thereat, the tension strength of concrete
is usually 815% of the compression strength [5,6].
Concrete behaves as linearly elastic material at low compression (until approx. 30% of maximal compression strength), and
across almost the whole range of tension. Considerable nonlinearity is noticed at modest and high compression levels, when concrete yields, that is, it behaves as ductile material. After achieving
the compression strength rcu, concrete enters into the softening

T. Vacev et al. / Engineering Structures 92 (2015) 114

Table 1
Characteristics of the tested footing sample.
Subgrade modulus of
compressibility Ms (MPa)

Subgrade modulus of
elasticity Es (MPa)

Footingdepth
h (cm)

Footing
effective depth
d (cm)

Concrete
strength fc
(MPa)

Reinforcement bar
diameter (mm)

Reinforcement
ratio q (%)

Reinforcement
yield stress fsy
(MPa)

44.5

40.0

12.5

10.0

16.85

0.40

570

zone, and eventual crushing fracture occurs at boundary strain ecu.

At tension, behaviour is approximately linearly elastic until the
maximal tension strength rtu, after which concrete cracks, and
strength gradually decreases to zero value.
For modelling of brittle materials like concrete, ANSYS proposes
the Willam-Warnke material model [7], using the isoparametric
nite element SOLID65 [8]. This element allows use of four different materials: one basic material of the structure (e.g. concrete)
and maximally three different materials for reinforcement, (e.g.,
steel and bre composites). The reinforcement may be modelled
as smeared or discrete, and in both cases total adhesion between
reinforcement and concrete is assumed. The element has eight
nodes and three degrees of freedom (DOF) per node translations
in x-, y-, and z-directions. SOLID65 allows cracking at tension, and
plastic behaviour, creep, and crushing at compression.
Cracking may occur in three orthogonal directions at every integration point, so the maximal number of cracks in one element is
8  3 = 24. Forming of cracks is enabled by modication of
stressstrain relations in the element, i.e., introducing the material
weakening in the direction perpendicular to the crack plane.
Possibility of shear stress transfer across the crack is dened by
coefcient bt, which may vary from total shear transfer (bt = 1) to
total absence of shear transfer (bt = 0) [9].
The algorithm of crushing satises the law of perfect plasticity,
where after exceeding the material strength and further increasing
of load, stresses remain constant and strains are increasing. At the
integration point in which the strength at uniaxial, biaxial, or triaxial compression is exceeded, crushing occurs, and it is manifested by complete loss of element stiffness. Once an element is
completely crushed (or cracked), load is entirely redistributed on
near, still undamaged elements.
The failure criterion of Willam and Warnke was developed in
1974 [7], and it represents a model of triaxial failure surface of
an unlimited concrete plane. In the elastic domain, the properties
are dened by the modulus of elasticity and Poissons ratio, where
the modulus of elasticity may be determined experimentally or
based on the existing relations. Advantages of this model are as
follows:
1. good agreement with experimental results in domain of working stresses;
2. simple determination of the model parameters from standard
test data.
In every concrete element, the cracking occurs when the principal tension stress in any direction is out of the failure surface.
Therefore, if the principal stresses r1 and r2 are compression
stresses, and r3 is tension stress, cracking will occur in direction
perpendicular to r3. In that case the modulus of elasticity of the
concrete element in direction parallel to the tension principal
stress direction attains zero value. Consequently, the cracking symbols in ANSYS take positions of three mutually orthogonal planes.
The symbols for cracking are circles (red for the 1st crack, green
for the 2nd crack, and blue for the 3rd crack). The circle symbols
are perpendicular to the direction of the corresponding principal
stress, and may represent the crack at the centroid of an element,
or at its integration points.

Crushing occurs when all three principal stresses are compressive (though stress r3 may be also equal to zero) and lay out of the
failure surface. Then the modulus of elasticity of the concrete element attains zero value in all directions and practically exists no
more. Crushing is represented in ANSYS by symbol of an octahedron. A totally crushed element manifests its behaviour bythe total
loss of its stiffness, i.e., the crushed element is excluded from further stress/strain distribution. In other words, the crushed element
vanishes from the structure. Stresses and strains are then redistributed to the elements in the vicinity. On the other hand, differently cracked elements may still accept stresses and strains.
The choice of the key material parameters was a direct consequence of several conditions: rst of all, of the concrete material
itself and of its behaviour described above, then choice of
ANSYS as the software for analysis, and consequently of the material model used by ANSYS, i.e., the WillamWarnke model. All of
this dictated the choice of the required material parameters.
In order to dene the WillamWarnke model, the following
properties of concrete must be entered:
1.
2.
3.
4.

modulus of elasticity of concrete Ec;

Poissons ratio m;
uniaxial compression strength f0c ;
uniaxial tension strength ft, where
0 0:5

f t 0:62  f c

5. shear transfer coefcient bt, which represents conditions on the

crack front; it has values ranging from 0.0 to 1.0, where 0.0 corresponds to the completely smooth crack (total loss of shear
transfer) and 1.0 corresponds to the completely rough crack
(no loss of shear transfer ability);
6. stressstrain relation for uniaxial pressure in concrete.
3.2. Material model for reinforcement
The element LINK180 was used to model steel reinforcement.
This is a line element with two nodes and three DOF per
node translations in x-, y-, and z-direction. It is supposed that
the relation between the concrete and the reinforcement is ideal.
This element has the ability of plastic deformation, that is, the reinforcement behaviour is modelled using bilinear-kinematic model
with stiffening [8]. Thereby, it is necessary to dene material
behaviour before and after yielding. Before yielding, in the linear
elastic domain, the reinforcement behaviour is described by the
elastic modulus Es, while in the plastic domain, its behaviour is
described by the yield stress fy and steel stiffening modulus Es0 .
Thus, the steel reinforcement behaviour may be represented by
the expressions:

rs Es  es ; es 6 ey

rs f y Es0  es  ey ; es > ey

where
rs denotes stress in steel reinforcement;
es denotes strain in steel reinforcement;
Es denotes modulus of elasticity for steel reinforcement;

T. Vacev et al. / Engineering Structures 92 (2015) 114

Es0 denotes tangent modulus of elasticity in steel reinforcement

after yielding;
fy and ey denote stress and strain at yielding, respectively.
3.3. Material model for soil
For modelling of soil in ANSYS, the DruckerPrager constitutive
law was adopted. It was successfully applied by other researchers,
e.g., [9]. It is represented in the principal stress space r1, r2, r3 by a
cone circumscribed around the 6-sided MohrCoulomb yield surface. For modelling of the soil block, element SOLID45 with eight
integration points and three DOF per node translations in x-, y-,
and z-direction, was adopted.
The yield surface is not changing during yielding, so no hardening law exists, and the material is perfectly elastoplastic. In order
to use this material model, it is necessary to dene the modulus
of elasticity, Poissons ratio, and the following additional properties: cohesion c, angle of internal friction u and dilatancy angle
W. When u = W the yielding law is associative, plastic deformations are perpendicular to the yielding surface, and volume expansion of material occurs during plastic deformation. If u < W volume
expansion will be lower, and if W = 0, there will be no volume
expansion at all.
3.4. FE model and mesh
FE model of the foundation structure is presented in detail in
Fig. 6. Regarding the symmetry of the structure and of the load,
only 1/4 of the structure (soil block and the foundation) was modelled and analyzed.
The FEA model comprised all relevant parts of the structure and
their material properties: soil, concrete footing with the column,
and reinforcement. The nodes of the reinforcement elements
matched completely with the concrete element nodes. Concrete
covering layer 25 mm thick was also included into the FE model.
Since our primary interest in this research was behaviour of the
reinforced concrete footing, and especially the phenomenon of
punching the column through the footing, the mesh density of
the footing FE was considerably higher compared with the mesh
density of the soil block. Consequently, these two meshes were
not conforming. However, such modelling is in fact more real,
because in reality lateral connection of the concrete and the soil
block in the contact surface plane is generated only by friction.
Perpendicularly to the contact surface plane, the soil produces
reactive force proportionally to its stiffness. Such mechanism of

load transfer from the footing to the soil led to the application of
contact analysis, introducing special ANSYS contact elements
(CONTA173 and TARGE170).
The TARGE170 element represents various 3-D target (passive) surfaces paired with the corresponding contact (active)
elements (CONTA173). The contact elements are surface elements
that overlay the solid elements which describe the boundary of a
deformable body [8].
The CONTA173 element is used to represent contact and sliding
between 3-D target surfaces (TARGE170) and a deformable surface, dened by this element. This element is located on the surfaces of 3-D solid or shell elements. It has the same geometric
characteristics as the solid or shell element face with which it is
connected. The contact occurs when the element surface penetrates one of the target segment elements (TARGE170) on a specied target surface [8].
The basic mechanism of contact implies presence or absence of
pressure stresses perpendicular to the contact surface between
two bodies, depending of load distribution and contact status.
Lateral (transverse) loads are transferred by friction between the
two materials (the concrete and the soil). In this research, all loads
were axial, so the only lateral forces were those caused by sliding
between the concrete footing and the soil.
The boundary conditions were as follows: the nodes on the
bottom of the soil block had all translations restrained (in x-, yand z-direction); the nodes on all vertical sides of the soil block
had restrained translations perpendicularly to their sides (in x-,
and z-direction); the nodes on the symmetry planes of concrete
had restrained translations perpendicularly to the symmetry
planes (in x-, and z-direction). The load consisted of the vertical
force directed downward, with maximal value of 500 kN. This force
was applied as surface load on the column top.

3.5. FEA input parameters

All input parameters relevant for the FEA are presented in the
Tables 24. In order to focus the research to the footing bearing
capacity, the concrete class of the column was signicantly higher
than that of the footing, i.e., modulus of elasticity of the column
concrete was 35.25% higher. It was applied as in the eld test, as
well as in the FEA, too. All input data are according to [8].
Special attention was paid to determination of input parameters
for concrete. The material model used for concrete (Willam
Warnke model) applied in ANSYS software requires several input
parameters which concern mostly to the concrete strength in

Fig. 6. FE model and mesh (1/4 of the structure); (a) soil, footing, and column; (b) footing and column; (c) footing and column reinforcement.

T. Vacev et al. / Engineering Structures 92 (2015) 114

Table 2
Input parameters for reinforcement.

Table 4
Input parameters for soil.

Item

Value

Notice

Item

Value

Notice

Material model
No./TYPE
FE No./TYPE

1/Bilinear
kinematic
4/LINK180

Selected from [8] for soil material

11/8 mm
12/8/2 mm

On symmetry plane (Area = 50%)

E = 2.1 1011 Pa

Standard for steel material

Material model
No./TYPE
FE No./TYPE
REAL
CONSTANT
Modulus of
elasticity
Poissons ratio
Density

3/Drucker
Prager
2 (SOLID45)
Empty

REAL CONSTANT
No./SECTION
REAL CONSTANT
No./SECTION
Modulus of
elasticity
Poissons ratio
Yield point
Tangent modulus
of elasticity
Density

Selected from [8] for steel material

Selected from [8] for line elements
According to the test sample

NUXY = 0.30
fy = 570 MPa
Et = 2.1 1010 Pa

Standard for steel material

According to test sample data

Cohesion
Friction angle
Dilatancy angle

0.0001
40
20

q = 7850 kg/m3

Standard for steel material

different stress states, such are: ultimate uniaxial tensile strength,

ultimate uniaxial compressive strength, ultimate biaxial compressive strength, ambient hydrostatic stress state, ultimate compressive strength for a state of biaxial compression superimposed on
hydrostatic stress state, and ultimate compressive strength for a
state of uniaxial compression superimposed on hydrostatic stress
state, also given in Table 3.
4. FEA results and discussion
4.1. Structure displacements
The footing displacements of the FE model were observed in
two control points where the test data were available, that is, in
the middle of the column top and on the top corner of the footing.
The results of the FEA and of the performed experiment generally show very good agreement (Fig. 8a). It must be stressed here
that experimental load was gradually increased from zero value
to the maximum of P = 430 kN, when the foundation collapsed,
while on the other hand the FEA load ranged from zero to the maximum of P = 500 kN, when solution diverged. The question of FEA
failure load will be discussed later.
The displacement component of the utmost interest was of
course the vertical component, or settlement (UY). This component
consisted of four phases: 1. soil settlement, 2. footing bending, 3.
footing and column compression, 4. column punching into the
footing. The FE model enabled all of the four phases to be
developed.

E = 40 106 Pa
NUXY = 0.27

q = 1700 kg/m3

Selected from [8] for concrete solids

According to [8] for solid nite elements
Calculated based on reached modulus
of compressibility (Ms)
Adopted value for gravel
Measured on soil sample under the
footing
Value close to zero for gravel
Adopted value for gravel
Adopted value for gravel

For the footing corner the nal comparable data (at P = 430 kN)
were as follows: UY = 7.7 mm (test) and UY = 6.8 mm (FEA),
making a deviation D = 11.69%. Deformation progress during
the test was mostly constant, with several sudden drops, at
P = 60/200/290 kN.
For the column top, i.e., for the centre point of the column, the
corresponding results (at P = 430 kN) were: UY = 29.0 mm (test),
and UY = 25.1 mm (FEA), giving a deviation D = 13.45%. The test
curve shows noticeable drops at P = 200/290/400 kN.
Here it must be noticed that the load step increments for the
FEA were set at 10 kN, but, during the experiment they were not
as regular. Therefore, the data for the test curve had to be interpolated and adapted in order to be compared with FEA.
Along with the FEA, a calculation of the punching bearing capacity according to EUROCODE 2 [10,11] and to b Model Code 2010
[12,13], was done. In order to match the calculation results with
the experiment, the safety factor for concrete and reinforcement
steel was adopted as 1.0.
The bearing capacity regarding bending was calculated as
247 kN, while regarding punching the calculation was done for
several control sections. Namely, EUROCODE 2 takes the section
which is set at the distance 2d (d is static depth of the section) from
the column face as a control section, but recommends a control in
sections closer to the column. The basic control section shows a
punching cone with an angle to the horizontal of 26.6, while other
control sections show higher angles. The calculation was done
starting from the ratio a/d = 2.0 till a/d = 0.5, with an increment
of 0.05, where a is the distance of the control section from the column face, and d is the static depth of the section.

Table 3
Input parameters for concrete.
Item

Value (footing)

Value (column)

Notice

Material model No./TYPE

FE No./TYPE
REAL CONSTANT No.
Modulus of elasticity
Poissons ratio
Open shear transfer coefcient
Closed shear transfer coefcient
Uniaxial cracking stress
Uniaxial crushing stress
Biaxial crushing stress
Hydrostatic pressure
Hydrostatic biaxial crushing stress

2/CONCRETE
3/SOLID65
21
E = 28.91 109 Pa
NUXY = 0.20
0.30
1.00
2.545 106 Pa
1.685 107 Pa
2.022 107 Pa
1
2.4433 107 Pa

4/CONCRETE
3/SOLID65
21
E = 39.10 109 Pa
NUXY = 0.20
0.30
1.00
4.8025 106 Pa
6 107 Pa
7.2 107 Pa
1
8.7 107 Pa

Hydrostatic uniaxial crushing stress

2.9066 107 Pa

1.035 108 Pa

Tensile crack factor

Density

0.60

0.60

q = 2500 kg/m3

q = 2500 kg/m3

Selected from [8] for concrete material

Selected from [8] for concrete solids
Arbitrary input
Ec = 22 ((fc + 8)/10)0.3, [10]
Standard for concrete material
bc (Recommended from [8])
bt (Recommended from [8])
Ultimate uniaxial tensile strength: ft = 0.62 (fc0 )0.5 [5]
Ultimate uniaxial compressive strength: fc (see Table 1)
Ultimate biaxial compressive strength: fcb = 1.2 fc [8]
Ambient hydrostatic stress state [8]
Ultimate compressive strength for a state of biaxial compression
superimposed on hydrostatic stress state: f1 = 1.45 fc [8]
Ultimate compressive strength for a state of uniaxial compression
superimposed on hydrostatic stress state: f2 = 1.725 fc [8]
(Recommended from [8])
Standard for concrete material

T. Vacev et al. / Engineering Structures 92 (2015) 114

(failure test load), and P = 460 kN (proposed failure FEA load). In

other words, at the moment of collapse, both methods show identical deection values.
The test curve shape is not as smooth as the FEA curve, and the
main reason for this is holding the load at some levels for a prolonged time, thus giving a dynamic character to the experiment.
On the other hand, the dynamic, or time effect, was not included
in the FEA.
4.3. Development of cracking and crushing
Fig. 7. Ultimate punching force depending on selected cross section, dened by
ratio a/d.

The part of the soil reactive pressure within the control section
was subtracted in the calculation. The calculation result is presented in Fig. 7. For the basic control section (a = 2.0d), an ultimate
punching load of 225 kN was obtained, while the governing minimal ultimate load of 208 kN was obtained for the control section
with ratio a/d = 1.4, which completely corresponds to the recommendations given in [11]. For this section an angle of the punching
cone of 35 is matching, which is approximate to the value
obtained by the experiment. The b Model Code 2010 gives the
ultimate punching load of 163 kN, for the control section at the distance a = d/2 from the column edge.
The results obtained using EUROCODE 2 and b Model Code
2010 show difference, which was expected, considering that the
different standards were used. Thereat, EUROCODE 2 gives results
closer to the experiment, with the safety factor related to the
experiment of cca. 2.0. The obtained punching load according to
EUROCODE 2 corresponds to the end of the 2nd and to the start
of the 3rd stage given in Fig. 8, that is, to the start of the punching
process, which in experimental and numerical analysis is not an
instantaneous event, but rather has its course.
A calculation of the bearing capacity of the soil was also performed, using EUROCODE 7. The calculation of the ultimate axial
force in the column was performed using the actual characteristics
of the soil (internal friction angle u0 and cohesion c0 ) in order to
match the calculation results with the test results and it was
assumed that the safety factors regarding the soil material, resistance, and actions are equal to one. The internal friction angle of
the soil u0 was not measured (tested), but adopted based on recommendations from the scientic literature. As the internal friction angle of the soil is increasing with the increase of the soil
compaction value, and considering that the compaction value
increased during the footing loading, the calculation of the ultimate bearing capacity of the soil was performed for values
u0 = 3640. For these values of internal friction angle u0 and cohesion c0 , the ultimate axial forces in the column were in the range of
215.8429.0 kN. Considering that the calculation provided lower
values of the failure load of the foundation soil (for u0 = 3639)
in comparison to the test values, one may conclude that EC7
provides conservative results. For the value of internal friction
angle of the soil u0 = 40, the calculation value obtained by EC7 is
approaching the failure load value obtained by the test
(F = 430 kN).
4.2. Footing deection
Footing deection here represents the difference between vertical displacements of the two control points of the foundation
structure: midpoint of the column top and the top corner of the
footing. The Fig. 8b shows comparison between the test and the
FEA data regarding deection of the structure, and it proves a very
good agreement. The two methods show almost equal value of
deection (21.3 mm) for the following load levels: P = 430 kN

General explanations about cracking and crushing capabilities

for concrete material in ANSYS are given in Section 3.1. Here the
overall course of development of cracks and crushings at several
typical stages of loading will be elaborated. Remark: cracking symbols are given as circles (red for the 1st crack, green for the 2nd
crack, and blue for the 3rd crack); crushing symbols are given as
red octahedrons.
Cracking arises at relatively low load level, P = 70 kN (Fig. 9).
The 1st cracks dominate, and they are placed near the bottom
of the footing (front view, Fig. 9b) and in the central part of the
footing, under, and around the column (top view, Fig. 9a). The
horizontal pattern of the cracks has a nearly radial shape, spreading from the column towards the diagonal of the footing, and it
is completely symmetric. Position of the 1st crack symbols indicates that the cracks lay primarily in the horizontal plane. The distribution of the cracks is as expected, showing that the source of
the cracks is right under the column, and that they are caused by
bending of the footing.
The distribution of the principal stresses in the footing correlates with the cracking pattern. Fig. 10 represents a criterion contour image with a contour legend starting from zero value to the
positive, i.e., tensile stresses, and with grey zones representing
compression.
At the load level P = 100 kN the cracks progress further, following the mentioned radial path, but being more progressive towards
the transversal directions, x- and z- (Fig. 11a). Now one may
observe the 1st and the 2nd cracks, all indicating cracking in
horizontal plane, again due to bending of the footing. Besides this,
the 3rd cracks are also present, located in the zone around the column outline (top view, Fig. 11a). Regarding vertical disposition, the
3rd cracks are found at the footing bottom and at the footing top
(Fig. 11b), i.e., where footing meets the column. The 3rd cracks
are mostly oriented vertically, indicating the source of the punching that arises. Conrmation for this is emerging of the rst crushings, located mainly at the same places.
Distribution of the principal stresses in the footing is now changed. Tensile stresses r1 and r2 are spreading to the side anks of
the footing, corresponding to the cracks in that region (Fig. 12a).
Compression still dominates (Fig. 12b and c) at the corner edge
of the footing. The reason for this stress redistribution is the 3-dimensional character of the FE model, i.e., the footing is not behaving as a plate, but as a 3D body. Symmetry in the crack and stress
distribution is still present.
At P = 200 kN the cracks are spread almost all over the footing
bottom, and through the footing depth, too. The crushings are concentrated under the column base, as expected, and they propagate
radially towards the peripheral parts of the footing (Fig. 13a). Some
cracks and crushings may be now observed at the lower part of the
column, and even near top surface of the footing (Fig. 13b). All this
announces occurrence of punching.
Tensile principal stresses, primarily r1, now occupy regions
away from the footing centre, and move to the corner and along
the side anks (Fig. 14a, and, to the less extent, Fig. 14b). The principal stress r3 still remains in compression range. The central part
of the footing bottom changes its stress state from tension to

T. Vacev et al. / Engineering Structures 92 (2015) 114

Fig. 8. Test data vs. FEA data; (a) vertical displacement of the foundation structure; (b) footing deection.

Fig. 9. Cracking and crushing distribution, P = 70 kN; (a) top view; (b) front view.

compression (Fig. 14a), showing that a dominant inuence is no

longer bending, but pressure. The symmetry in the crack and stress
distribution begins to fade.
The proposed failure load, P = 460 kN, brings crack pattern in
which complete footing volume is covered with cracks and crushings (Fig. 15). The cracks and crushings now climb even up the
column, indicating that the higher-class concrete of the column

is close to its strength limits, too. Thus, the footing structure is

practically completely deteriorated.
The principal stresses continue transferring to compression,
particularly at the footing, but also at the side anks (Fig. 16a).
Such alteration matches the massive crushing manifestation in
the footing, which is now the dominant factor. The symmetry in
the crack and stress distribution faded almost completely.

T. Vacev et al. / Engineering Structures 92 (2015) 114

Fig. 10. Principal stresses distribution (Pa), P = 70 kN, view from below; (a) r1; (b) r2; (c) r3.

Fig. 11. Cracking and crushing distribution, P = 100 kN; (a) top view; (b) front view.

Fig. 12. Principal stresses distribution (Pa), P = 100 kN, view from below; (a) r1; (b) r2; (c) r3.

4.4. The mechanism of punching the column into the footing

The very process of footing punching is manifested by the column intruding into the footing volume, and according to Fig. 8a
and b, can be divided into four stages.
The 1st stage lasts from the loading start until the load reaches
level P  100 kN, during which the footing behaves as stiff, there
are almost no cracks, and no deections.

The 2nd stage starts from the load P  100 kN when footing
deection begins, and when also the massive occurence of cracks
in the lower zone of the footing is noticed (Fig. 11). The footing
deection and crack development continues until the load value
reaches P  220 kN. Crushing of some elements also occurs at the
beginning of this stage (Fig. 13).
At the load level P  220 kN, the 3rd stage starts, and the deection progresses more intensely as the load level increases. From the

10

T. Vacev et al. / Engineering Structures 92 (2015) 114

Fig. 13. Cracking and crushing distribution, P = 200 kN; (a) top view; (b) front view.

Fig. 14. Principal stresses distribution (Pa), P = 200 kN, view from below; (a) r1; (b) r2; (c) r3.

Fig. 15. Cracking and crushing distribution, P = 460 kN; (a) top view; (b) front view.

T. Vacev et al. / Engineering Structures 92 (2015) 114

11

Fig. 16. Principal stresses distribution (Pa), P = 460 kN, view from below; (a) r1; (b) r2; (c) r3.

Fig. 8 one may notice that the column settlement progressively

increases, and the footing corner settlement stagnates. Such trend
continues until the load value P = 460 kN.
At the load level P = 460 kN, starts the 4th, and the last stage of
punching, when deection grows faster, the cracks and crushings
are spread through the whole footing volume (Fig. 15), and the
nal punching occurs.
The punching mechanism of the footing can be also tracked
using the concrete and reinforcement strains. At the 1st stage of
the footing punching the strain distribution is as expected, i.e.,
maximal compressive strains are in the upper zone of the footing,
immediately against the column (Fig. 17a), and maximal tension
strains are in the reinforcing bars under the foundation column
(Fig. 17b).
At the 2nd stage the compressive strains in the concrete and
tension strains in the reinforcement (at the same locations) continue to grow, so that at the end of the 2nd stage, at P = 220 kN, maximal concrete strains are e = 6.146 (Fig. 18), and strains in the
reinforcement reach yielding point, e = 2.739, (Fig. 18b).
Transition from the 2nd into the 3rd loading stage at the same
time represents the beginning of the footing punching.
The proof for the achieved yielding point in reinforcement is
given by the following equation:

where fy = 570 MPa (see Table 2).

In the 3rd stage, the column continues to dive into the footing, which is accompanied with the decrease of compressive
strains in the concrete of the footing next to the column, and with
their transition into tension. This transition process is illustrated
by Figs. 17a and 18a, and it continues further. On the other hand,
strains in the reinforcement increase faster and faster.
The process of diving of the column into the footing may be
also noticed in Fig. 19, which shows vertical displacements. At
the beginning of the 3rd stage, there are no deformed elements,
while at its end one may clearly observe the elements that are
deformed due to punching.
The 4th and the last stage of punching starts at the load value of
P = 460 kN. It is characterized by an even faster increase of the
reinforcement tension strains practically its yielding, and with
fast growth of tension strains in the footing concrete next to the
column. Conrmation of the punching is also the reinforcement
strain (Fig. 20). At the end of this stage (Fig. 20b), the reinforcement strains rise abruptly, even in the column zone, which corresponds to the crack and crushing development (Fig. 15).

r eE 0:002739  2:1E11 575:19 MPa  f y

The problem of determining the exact failure, or collapse load

for a column footing structure using FEA, is multiply complex. In

4.5. Failure load

Compression

Fig. 17. Strains at P = 100 kN; (a) concrete, x-direction component (grey contours denote tension); (b) reinforcement (elastic component).

12

T. Vacev et al. / Engineering Structures 92 (2015) 114

Turned into tension

Fig. 18. Strain at P = 220 kN; (a) concrete, x-direction component (grey contours denote tension); (b) reinforcement (elastic component).

Fig. 19. Vertical displacements of concrete elements (m); (a) at the beginning of the 3rd stage (P = 220 kN); (b) at the end of the 3rd stage (P = 460 kN).

the bearing capacity of the structure three materials take part, all
three with non-linear and unique behaviour, especially at high load
levels. The material nonlinearity is not the only one that emerges
in this soil-structure composition. A special kind of nonlinearity
is the contact status between the soil and the footing, combined
with friction between the two materials. Such complexity of the
FEA model made the analysis extremely sensitive regarding the
numerical convergence.
Every test model is, on the other hand, 100% real, and each one
is unique, considering all the material, structural and dimensional
imperfections. Also, there are numerous experimental setup
imperfections that are practically unavoidable, and may include
accidental eccentricity of load application, parasite deformations
of the test frame, imperfections in the subgrade soil preparation,
and so on.
The eld test of the footing specimen, designated as T9, showed
that the structure collapsed, i.e., the concrete experienced fracture
at load value P = 430 kN. This value was the aim for the FEA of the
T9 footing model. In order to achieve the test failure load as close

as possible, calibration of the FEA model and of the solution parameters were done. Some of the parameters whose exact value was
not available in advance were Open and Closed Shear Transfer
Coefcients (bt, bc). Some recommendations are given by other
researchers [1416], but the nal values here had to be determined
by numerous analyses, considering the convergence problems. The
coefcients bt, bc, are predened by the chosen software in the
range between 0.0 and 1.0. Specic values chosen in this paper
were selected based on the criteria for solution convergence. The
value for bc was taken as 0.0 for start value, and was subsequently
increased until the converged solution was achieved, so that nal
value was established as bc = 0.3.
The main question about determining the failure load in FEA is
the criterion on which we will base our decision. Careful revision of
all FEA data gave several distinctive features that led to a single
conclusion.
The rst criterion is the moment of emerging of divergence, or,
in other words, the moment when equilibrium of the internal
forces is no longer achievable due to the excessive deformations.

T. Vacev et al. / Engineering Structures 92 (2015) 114

13

Fig. 20. Strains in reinforcement (plastic component); (a) at the beginning of the 4th stage (P = 460 kN); (b) at the end of the 4th stage (P  500 kN).

Applying this criterion, we would get an appearance of the structure presented in Fig. 21. This is the moment of solution divergence
accompanied by software message indicating abrupt change of
contact status, i.e., exaggerated deformations of the model caused
setting some nodes into space and loosing the initial contact setup. All these phenomena arise at load level near to P = 500 kN.
The second criterion is represented by vertical displacements of
the column top (Fig. 8, FEA diagram). Here one may notice sudden
settlement at P = 460 kN, indicating punching process, and implying the failure load value of P = 460 kN.
The third criterion is the reinforcement strain (Fig. 20). Here one
may see sudden increase of the strains during the load range
P = 460500 kN, as in the lower zone (footing), as well as in the column reinforcement. This phenomenon is an obvious sign of punching of column through the footing, which means collapse of the
structure.
Considering all three of the criteria explained above, we nally
adopted the load value of P = 460 kN as a failure load.
Determining of the failure surface and its angle to the horizontal, regarding the experimental investigation, can be done using

the documenting photo from the test (Fig. 22). Here one may notice
the failure surfaces leading from the edge column-footing down to
the footing bottom surface. The angle to the horizontal can be
assessed as near 40 (right side), and near 30 (left side). It is obvious that failure surfaces are not symmetrical, which indicates some
eccentricity of loading.
In order to dene the FEA failure surface we observed a group of
footing concrete elements with vertical displacements exceeding
certain value, for the proposed failure load, P = 460 kN (Fig. 23).

Fig. 22. Experimental analysis photo of the test specimen at failure (vertical
section obtained by halving).

Fig. 21. Concrete foundation vertical displacements at the end of solution

(P  500 kN).

Fig. 23. FE analysis footing elements with vertical displacements UY > 21 mm, at
failure load, P = 460 kN.

14

T. Vacev et al. / Engineering Structures 92 (2015) 114

The vertical displacement is chosen as a criterion because it is an

essential phenomenon that accompanies failure. The exact criterion value for the vertical displacement was set to 21 mm, which
was the maximal deection value of the footing, whereby the soil
settlement inuence was eliminated, and only the punching component of displacement remained.
Based on this, one may notice that failure surfaces obtained by
FEA resemble the experimental failure surface. Of course, in case of
a signicantly denser FE mesh, the failure surface would be represented in a more rened manner, but with tremendous load for
hardware resources.
5. Final conclusions
Considering all of the quantitative and qualitative data obtained
in this research, the following conclusions can be drawn:
1. Finite element method is a powerful tool for solving even multiply complex structural problems, such is behaviour of reinforced column footings laid on deformable subgrade.
2. ANSYS software with its FE library, material models, and solution capabilities enables implementation of diverse nonlinearities (reinforcement plasticization, concrete cracking and
crushing, soil elastoplasticity, and changing of contact status,
including friction). Based on these capabilities, one may build
a veritable mathematical model of a compound and materially
heterogeneous structure.
3. Simultaneous including of cracking and crushing capabilities in
the concrete material of the FE model generated problems well
known from similar research projects, [1521]. In this research,
those problems were manifested by difculties in achieving
convergence, and by asymmetric development of crushing, concrete strains, and reinforcement strains. This asymmetry was
registered in horizontal plane, comparing x- and z-direction,
and emerges at medium load levels, at about P = 220 kN, when
the reinforcement reaches the plasticity threshold, and it continues to grow until the failure. In Fig. 18b, at P = 220 kN, one
may see that the reinforcement strains are obviously symmetrical, and in Fig. 20a and b, at P = 460 kN and P = 500 kN, those
strains are clearly unsymmetrical.
4. Despite the difculties mentioned previously, the solution of
punching of the RC column footings applying FEA and using
ANSYS software gave results with very good accuracy, compared with the eld test results. Taking into account all the failure criteria given in Section 4.5, we adopted the failure load
value of 460 kN as a reliable reference. Based on this, a comparison of characteristical deformations obtained from the eld
test and FEA was made, giving very high agreement (Table 5).
This comparison shows that prediction of the failure load for
such structures using FEA is possible, and highly reliable.
5. The further improvement of the material model for concrete in
ANSYS software would be desirable and useful, and would contribute to the reliability of solving complex problems with multiple nonlinearities.

Table 5
Comparison of results from test and FEA.
Parameter

Failure load
(kN)

Footing vertical
displacements UY (mm)

Footing deections
DY (mm)

Test
FEA
D (%)

P = 430
P = 460
6.98

29
28
3.45

21
21
0

6. The signicance of this research lays in the attempt to use the

power of FEA and application of sophisticated software in
approaching to the reality of behaviour of materially complex
foundation structure, which includes concrete, reinforcement,
and soil. Such complex approach was, as a rule, avoided by
investigators. The numerical and test results obtained in this
research showed that this attempt is justied, and can impact
the further improvement of the building codes.

Acknowledgements
The paper is a result of the investigation within the scientic
projects TR 36028 and TR 36016 nanced by the Ministry of
Science and Technological Research of Republic of Serbia.
The sponsor nanced experimental research and supplied
needed software.
Special acknowledgement to Mr John Deibler, Pacic Northwest
National Lab., for valuable suggestions.
References
[1] Hegger J, Sherif AG, Ricker M. Experimental investigations on punching
behavior of reinforced concrete footings. ACI Struct J 2006:60413 [July
August].
[2] Hegger J, Ricker M, Ulke B, Ziegler M. Investigations on the punching behavior
of reinforced concrete footings. Eng Struct 2007;29:223341.
[3] Hallgren M, Bjerke M. Non-linear nite element analyses of punching shear
failure of column footings. Cement Concr Compos 2002;24:4916.
[4] Hallgren M, Kinnunen S, Nylander B. Punching shear tests on column footings.
Nordic Concr Res 1998;21(1):124.
[5] Shah SP, Swartz SE, Ouyang C. Fracture mechanics of concrete. New York: John
Wiley & Sons Inc.; 1995.
[6] Bangash MYH. Concrete and concrete structures, numerical modelling and
applications. London (England): Elsevier Science Publishers Ltd.; 1989.
[7] Willam KJ, Warnke EP. Constitutive model for the triaxial behaviour of
concrete, seminar on concrete structures subjected to triaxial stresses. In:
International association of bridge and structural engineering conference,
Bergamo, Italy; 1974. p. 213.
[8] ANSYS Software documentation. Release 14.5. ANSYS, Inc., Canonsburg, PA.
2012.
[9] Kazaz I, Yakut A, Glkan P. Numerical simulation of dynamic shear wall tests: a
benchmark study. Comput Struct 2006;84:54962.
[10] EN 1992-1-1:2004, Eurocode 2: Design of concrete structures Part 1-1:
General rules and rules for buildings.
[11] Eurocode 2 Commentary, Eurpoean Concrete Platform ASBL, Brussels; June
2008.
[12] b Model Code 2010. First complete draft, vol. 2. Lausanne, Switzerland; April
2010.
[13] Lips, S, Muttoni, A, Ruiz, MF, b Model Code 2010. Punching of at slabs:
design example, Ecole Polytechnique Fdrale de Lausanne, Switzerland; 2011.
[14] Kachlakev D, Miller T, Yim S, et al. Finite element modelling of concrete
structures strengthened with FRP laminates. Washington (DC): Oregon
Department of Transportation Research Group and Federal Highway
Administration; 2001.
[15] Wolanski AJ. Flexural behaviour of reinforced and prestressed concrete beams
using nite element analysis, A Thesis submitted to the Faculty of the
Graduate School, Marquette University in Partial Fullment of the
Requirements for the Degree of Master of Science, Milwaukee, Wisconsin;
2004. p. 38.
[16] Zhou S, Rizos DC, Petrou MF. Effects of superstructure exibility on strength of
reinforced concrete bridge decks. Comput Struct 2004;82:1323.
[17] Barbosa AF, Ribeiro GO. Analysis of reinforced concrete structures using ANSYS
nonlinear concrete model. In: Idelsohn S, Oate E, Dvorkin E, editors.
Computational mechanics, new trends and applications. CIMNE, Barcelona,
Spain; 1998.
[18] Dere Y, Dede FT. Nonlinear nite element analysis of an R/C frame under
lateral loading. Math Comput Appl 2011(16):94758.
[19] Kadhim MMA, Mohammed MJ, Abid AJ. Effect of prestressed CFRP plate
location on behaviour of RC beam strengthened with prestressed CFRP plate. J
Babylon Univ 2012;20(1):10513.
[20] Kaewunruen, S, Remennikov, A. Nonlinear nite element modelling of railway
prestressed concrete sleeper. In: The tenth East AsiaPacic conference on
structural engineering and construction, August 35, 2006, Bangkok, Thailand,
University of Wollongong, Research Online.
[21] Mohyeddin A, Goldsworthy HM, Gad EF. FE modelling of RC frames with
masonry inll panels under in-plane and out-of-plane loading. Eng Struct
2013;51(2013):7387.