Available online at www.sciencedirect.

com

ScienceDirect
Procedia Engineering 95 (2014) 252 – 259

2nd International Conference on Sustainable Civil Engineering Structures and Construction

Materials 2014 (SCESCM 2014)

Finite element modeling of concrete fracture in tension with the

Brazilian splitting test on the case of plane-stress and plane-strain
Bobby Rio Indriyanthoa,*, Nurojib
a
Master Program in Civil Engineering, Diponegoro University, Semarang 50275, Indonesia
b
Department of Civil Engineering, Diponegoro University, Semarang 50275, Indonesia

Abstract

Most of tensile tests by researchers were inconclusive, due to differentiation in results and variation in testing methods. The
available test methods provide only a limited amount of information on this tensile behavior, producing the properload-
displacement response. The most common testing method is the Brazilian splitting test according to ASTM C496. In the
numerical model the question as whether the case of splitting test should be considered as a plane-stress or plane-
strainphenomena, has surfaced. A finite element model to produce a more sophisticated explanation of the concrete behavior in
tension,is constructed. The experimental tests were performed to validate the developed program based on the load-displacement
curves. The developed program also resulted the stress distribution of cylinder splitting test.
© 2014The
© 2014 TheAuthors.
Authors. Published
Published by Elsevier
by Elsevier Ltd.
Ltd. This is an open access article under the CC BY-NC-ND license
Peer-review under responsibility of organizing committee of the 2nd International Conference on Sustainable Civil Engineering
(http://creativecommons.org/licenses/by-nc-nd/3.0/).
Peer-review and
Structures underConstruction
responsibility of organizing
Materials committee of the 2nd International Conference on Sustainable Civil Engineering
2014.
Structures and Construction Materials 2014
Keywords:Finite element model; splitting test; plane-stress; plane-strain; stress distribution

* Corresponding author. Tel.: +628-564-372-5591; fax: +62-24-7648-0727.

E-mail address:bobbyrio77@yahoo.com

1877-7058 © 2014 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/3.0/).
Peer-review under responsibility of organizing committee of the 2nd International Conference on Sustainable Civil Engineering
Structures and Construction Materials 2014
doi:10.1016/j.proeng.2014.12.185
 Bobby Rio Indriyantho and Nuroji / Procedia Engineering 95 (2014) 252 – 259 253

Introduction

1.1. Concrete behavior in tension

The failure mechanism of concrete can be distinguished in two types of phenomena, i.e. crushing and cracking.
Crushing is a specific compression failure case and cracking takes place when an element experiences a state of
tension strains. The configuration of principal stresses and strains in the two-dimensional analysis, results in three
possible combinations [1]. Firstly when all the principal stresses are in tension, secondly when both these stresses
are in compression and finally the third, when a combination of tension and compression exists. The tension-tension
and compression-tension case are the most vulnerable in the bi-axial condition. The failure state is evaluated based
on a failure envelope. Previous studies showed that the most powerful approach to these yielding criteria are the
failure envelopes as proposed by Möhr-Coulumb and Kupfer-Hildorf-Rusch [2-4].
In the past, researchers [5-7] had studied the phenomenon of concrete cracking based on the fracture energy
analysis. While the results are promising, the proper load-displacement behavior could not be generated based on
this theory alone.

1.2. State of the art on concrete tensile test

Tensile testing of concrete are categorized into direct and indirect tensile tests. A direct tensile test is conducted
by applying an axial tension force to the concrete specimens. Among these are the methods developed by [8-10].
The latest used a molded specimen with embedded steel bars, the bars were used as media to apply the tensile force
to the concrete. This method resulted in a large differentiation in test results since the difficulties to avoid the
eccentricity. Further, the bond stress developed in the vicinity of the embedded steel bars resulted in stress
concentrations that initiated cracks and affected the tensile strength negatively. Additionally, the method developed
by [11] used a “dog-bone” specimen as un-notched element and prism specimen as notched one. Comparable
experiments with notched specimens were also conducted by the other researchers [12-14].
Due to the constrains as mentioned, most of the researchers shifted to the indirect tensile test methods i.e. flexural
testing (rupture analysis) and the Brazilian splitting test. These tests are easier to be undertaken since the test set-up
procedure has been standardized in many codes, ASTM and the Euro Code among others. The flexural test is
conducted by loading a simply supported prismatic beam subjected to a one-point loading or two-point loading
system [15-17]. The result from this test is the concrete modulus of rupture. The Brazilian splitting test was firstly
proposed by [18] in Brazil, explaining its name, the Brazilian splitting test. A concrete cylinder is placed
horizontally and loaded in compression by the loading platens of the compression testing machine along the length
of the cylinder height (Fig. 1). Extensive papers were published based on this test method, among which [19-24].

Fig. 1.The Brazilian splitting test.

254 Bobby Rio Indriyantho and Nuroji / Procedia Engineering 95 (2014) 252 – 259

From [25], the formula for calculating the Brazilian splitting strength is expressed by Eq. (1).
2P
f sp (1)
SdL
fsp is the splitting tensile strength at failure due to the ultimate load (P), and d is the diameter of the cylinder with
L as the length of the specimen.
Considering the simplicity of testing method and the extensive research, the Brazilian splitting test model was
chosen for this study. The numerical studies on tensile behavior of concrete are very limited, therefore, a nonlinear
Finite Element Model (FEM) program can be developed to estimate the tensile behavior of concrete.
The developed program was written in the Visual Basic language, this language was favored for its ease in use
and its excellent visual representation. The program was further validated to laboratory tested cylinders having exact
similar material properties and dimensions as the FEM.

2. Finite element modeling

2.1. Material behavior

In the two-dimensional analysis, a different opinion about the approach to the material behavior obtained from
the Brazilian splitting test method exists. The issue is as whether the material constitutive matrix should be assumed
as a plane-stress, or as a plane-strain state. A plane-stress state is defined as a state of stress that prevails in a thin
plate, loaded only in its own, or x,y plane. In this case, all of the stresses in the z-direction are always zero.
However, the strain in this direction is non-zero (ız = IJyz = IJzx = 0, İz  0). The governing equation of the material
constitutive matrix for a plane-stress condition is shown in Eq. (2) as adapted from [26].
ª E1 X 0 º
E2
E1 « E1
»
>C @ E1
« X E2
0 » (2)
X 2 « »
E2
«¬ 0 0 1
2
 E1
E2

X »
¼
Where E1 and E2 are respectively modulus of elasticity in the major and minor principal strain direction, ȣ is
Poisson’s ratio of concrete. In the formula, the two principal modulus of elasticity E1 and E2 govern isotropic or
anisotropic behavior. At initial loading, the material can be assumed isotropic (E1 = E2 = E0 = E), the behavior
changes to anisotropy after cracking takes place.
A plane-strain state is used for analysis of long-body elements with a constant cross section subjected to line
pressure loading along the cylinder. All strains in the z-direction are always zero, with a non-zero stress (İz = Ȗyz = Ȗzx
= 0, ız  0). Eq. (3) exhibits the governing equation for a plane-strain condition.

ª E1  X X 0 º
« E2 »
E1 E1
>C@ « X X 0 » (3)
 E1
E2
X  E1
E2
 2X « 
« 0
E2

0 1
 E1
 X
»
»
¬ 2 E2 ¼
This research work governed the two states of stress-strains i.e. plane-stress and plane-strain. The results were
compared and evaluated. An isoparametric quadrilateral element with four nodes (2×2 Gauss point) was used for
this FEM. The stress-strain relationship was generated based on [27]. This code has been studied and was proven
accurate in representing the plain concrete behavior [2-4].
In this study the Möhr failure criteria was used to set the boundaries for the Gauss-point failure. In the Möhr
failure criteria the failure of an element is evaluated based on either the shear (IJ’) and normal stresses (ın) or the
principal stresses ı1 and ı2. The program was terminated when the value of the structural stiffness approached zero.
The failure of a Gauss point was marked by the absence of the material stiffness modulus, thus approaching a value
 Bobby Rio Indriyantho and Nuroji / Procedia Engineering 95 (2014) 252 – 259 255

of zero. In programming, the error of a zero determinant due to a zero in the E was accommodated by assigning a
substantial small value for E.

2.2. Meshing technique

The cylindrical concrete specimen for the FEM is manually modeled as a two-dimensional quarter of a circle
with restrains at the edge-nodes, corresponding to the boundaries of the actual specimen (Fig. 2). An algorithm to
produce the node coordinates was written, enabling the variation in node and element number. The choice of a
quarter-model was due to the geometric symmetry nature of the cross section and loading, and served mainly for
simplification purposes, and to reduce the running time of the program.
This model functioned as input to the geometric parameters of the developed program, and is substituted in terms
of their joints coordinate and element numberings.

Fig. 2.Cylinder meshed model.

3. Experimental Work

This research work was accompanied by experimental investigations, for validation purposes. The laboratory
produced specimens had three compressive strengths i.e. C30, C40, and C50. Cylinders sized 150×300 mm based on
[25] were used to carry out the Brazilian splitting tests. For each compressive strengthfourspecimens were casted,
and kept moist for 28 days. The mix proportion of concrete is presented by Table 1. The cement used was PPC type
1 whereas the sand was a local product from Muntilan, Central Java and. The maximum size of coarse aggregate
was 20 mm and Since the numerical program required the material properties as input, the compressive strength (f’c)
of the concrete was tested at the age of 28 days.

Table 1. The mix proportion of concrete.

Mix Proportions
Code f’c (MPa)
(Cement : Sand : Coarse Aggregate)
C30 31.0 1.00 : 1.75 : 2.75
C40 41.0 1.00 : 1.37 : 2.16
C50 50.6 1.00 : 1.04 : 1.64

The compressive strengthsresulted from the cylindrical compression tests were coded C30 for f’c= 31.0 MPa, C40
for f’c= 41.0 MPa, and C50 for f’c= 50.6 MPa. In preparation for the Brazilian test, each specimen was laid
horizontally on a splitting test apparatus. The incremental loading was applied by the Universal Testing Machine
256 Bobby Rio Indriyantho and Nuroji / Procedia Engineering 95 (2014) 252 – 259

(UTM) with a rate of 700 kPa/min based on [25]. The load was monitored by a load-cell and the displacements were
measured by twoLVDTs connected to a data logger. The test set-up can be seen in the Fig. 3.

Fig. 3.The test set-up of the Brazilian splitting test.

4. Result and validation

The developed FEM program was constructed to compare the plane-stress and plane-strain state of plain
concrete. Before doing so, this program was validated to evaluate its reliability and correctness. The load
displacement responses and the stress distributions were obtained from this numerical simulation for the
compressive strengths of 31.0 MPa, 41.0 MPa, and 50.6 MPa, and compared to the laboratory specimen results.

4.1. The load-displacement response

The FEM program resulted in the generation of the load-displacement curves simulated by plane-stress and
plane-strain model. The data from the laboratory specimens are shown dotted line, whiles the product of the FEM
program is presented in a solid line. The resulting data are plotted in one graph and shown in Fig. 4-6.The Fig. 4-6
respectively represents the specimens with cylindrical compression strength of 31.0 MPa, 41.0 MPa, and 50.6 MPa.
From these figures it can be seen that the proposed FEM produced a close approximation to the experimental curves.
The program could therefore be considered versatile and correct, in representing the concrete behavior under a state
of splitting tension loading.

200
Load (kN)

160

120

80
Experimental C30
FEAST 31 - Plane Strain
40
FEAST 31 - Plane Stress
0
0 0,04 0,08 0,12 0,16
Displacement (mm)

Fig. 4.The load-displacement validation for C30 [28].

 Bobby Rio Indriyantho and Nuroji / Procedia Engineering 95 (2014) 252 – 259 257

240

Load (kN)
200

160

120
Experimental C40
80
FEAST 41 - Plane Strain
40
FEAST 41 - Plane Stress

0
0 0,04 0,08 0,12 0,16 0,2
Displacement (mm)

Fig. 5.The load-displacement validation for C40 [28].

280
Load (kN)

240

200

160

120
Experimental C50
80
FEAST 50,6 - Plane Strain
40 FEAST 50,6 - Plane Stress
0
0 0,04 0,08 0,12 0,16 0,2 0,24 0,28
Displacement (mm)

Fig. 6.The load-displacement validation for C50 [28].

4.2. The stress distribution

Beside the load-displacement response, this developed program is also able of producing the stress distribution.
The stress distribution predicted by this program was compared to the failure of experimental tested specimens.
Comparing the results of specimen C50, it can be seen that the stress distribution result of the numerical simulation
corresponds to the experiment (Fig. 7 and 8). From the program simulation, the cylinder C50 failed at the loading
stage of 257 kN. At the upper most fibers of the cylinder, crushing due to compressive stresses (compression-
compression) were detected, but along the center of the section cracking or compression-tension was noticed. This
conclusion also underlined the stress distribution of a cylinder under splitting load as mandated in [25] (Fig. 9).

Fig. 7.Nodal stresses at X-X and Y-Y direction [28].

258 Bobby Rio Indriyantho and Nuroji / Procedia Engineering 95 (2014) 252 – 259

Fig. 8.The crack pattern of the laboratory experiment [28].

Fig. 9.Stress distribution of the Brazilian splitting test [25].

5. Analysis and Discussion

As for the case study of the plane-stress and plane-strain state, the concrete behavior was analyzed using the
developed program. The plane-stress model is customary for analyzing thin plates, and the plane-strain model for
the calculation of dam bodies. In the ease of the splitting test analysis, the cylinder is modeled as two-dimensional or
plane element assuming a negligible thickness. In the model, the length of the cylinder is considered as element
thickness. In the actual case, the thickness or height of the cylinder is two times larger than its diameter. Therefore,
the approach to the model as a plane-strain case, might give a better representation to the actual behavior.
From the result of the numerical simulation, it was found that the load-displacement curve of the plane-strain
model nearly coincides with the plane-stress model (Fig. 4-6). In terms of the displacements, the plane-stress model
exceeded the plane-strain model. Comparing to the experimental data, the plane-strain model resulted the load-
displacement curve more accurately than the plane stress one. From the numerical analysis (Fig.7), the tensile stress
distribution at X-X direction was approximately 75% up to 79% of the diameter.

6. Conclusion

The developed program was proven to generate accurate load-displacement curves which were confirmed by the
laboratory tested specimens results. The FEM predicted that the compressive stresses occured at the top fibers of the
cylinder, and the combination of compression and tension that occur along its diameter. Cracking happens at the
center of the circle first and propagates then up till failure. The generated FEM program also confirmed the stress
distribution as dictated by [25]. From this study it was also found that the plane-strain model presented a more
accurate outcome as compared to the plane-stress state.
 Bobby Rio Indriyantho and Nuroji / Procedia Engineering 95 (2014) 252 – 259 259

References

[1] H. Kupfer, H.K. Hilsdorf, K. Rusch, Behavior of concrete under biaxial stresses, ACI J. Proc. 66 (1969) 656–666.
[2] B.R. Indriyantho, A.S. Pamungkas, Study of rupture by numerical simulation, Thesis, Dept. of Civ. Eng., Diponegoro University, Indonesia,
2013, unpublished.
[3] A.L. Han, Sukamta, B.R. Indriyantho, The analysis of concrete fracture by the numerical method, in: Proc. of 2th Annu. Int. Conf. of Arch.
and Civ. Eng., Singapore, 2014, pp. 116-121.
[4] A.L. Han, Sukamta, B.R. Indriyantho, A finite element approach to the concrete failure criteria of Möhr-Coulomb and Kupfer-Hilsdorf-Rusch,
J. Eng. Tech. 3 (2014).
[5] J.G.M. Van Mier, Fracture Processes of Concrete, CRC Press, 1996.
[6] A. Hillerborg, M. Modeer, P.E. Petersson, Analysis of crack formation and crack growth in concrete by means of fracture mechanic and finite
elements, Cement and Concrete Research 6 (1976) 773–782.
[7] Z.P. Bazant, B.H. Oh, Deformation of progressively cracking reinforced concrete beams, ACI J. Proc. 81 (1984) 268–278.
[8] R.H. Evans, M.S. Marathe, Microcracking and stress-strain curves for concrete in tension, Mat. andStruct. 1 (1968) 61–64.
[9] H.W. Reinhardt, T. Rinder, High strength concrete under sustained tensile loading, Otto-Graf J. 9 (1998) 123–134.
[10] V.M.C.F. Cunha, J.A.O. Barros, J.M. Senna-Cruz, An integrated approach for modeling the tensile behavior of steel reinforced self-
compacting concrete, Cement and Concrete Research41 (2011) 64–76.
[11] V. Mechtcherine, H.S. Müller, Effect of test set-up on fracture mechanical parameters of concrete, in: Proc. FRAMCOS-3 Fracture Mech
and Concrete Struct., Germany, 1998, pp. 377-386.
[12] Q. Li, F. Ansari, High-strength concrete in uniaxial tension, ACI Mat. J. 97 (2000) 49–55.
[13] J.G. Rots, R. de Borst, Analysis of concrete fracture in direct tension, Int. J. Solids Structures 25 (1989) 1381–1394.
[14] L. Cedolin, S.D. Poli, I. Iori, Tensile behavior of concrete, ASCE J. Eng. Mech.113 (1987) 431–449.
[15] M. Gencoglu, B. Mobasher, Monotonic and cyclic flexural behavior of plain concrete beams strengthened by fabric-cement based
composites, ACI J. Proc. 81 (1984) 268–278.
[16] S. Wu, X. Chen, J. Zhou, Tensile strength of concrete under static and intermediate strain rates: correlated results from different testing
methods, Nuclear Eng. and Design250 (2012) 173–183.
[17] H.D. Yun, Flexural behavior crack-damage mitigation of plain concrete beam with a strain-hardening cement composites (SHCC) layer at
tensile region, Composites: Part B 45 (2013) 377–387.
[18] F.L.L.B. Carneiro, A. Barcellos, Concrete tensile strength, Bull. No.13, Int. Assoc. of Testing and Research Laboratories for Mat.andStruct.,
1953.
[19] Z.P. Bazant, M.T. Kazemi, T. Hasegawa, J. Mazars, Size effect in Brazilian-split cylinder test: measurements and fracture analysis, ACI
Mat. J. 88 (1991) 325–332.
[20] G. Khalaj, A. Nazari, Modeling split tensile strength of high strength self compacting concrete incorporating randomly oriented steel
fibersand SiO2 nanoparticles, Composites: Part B 43 (2012) 1887–1892.
[21] Z. Lin, L. Wood, Concrete uniaxial strength and cylinder splitting test, ASCE J. Struct. Eng. 129 (2003) 692–698.
[22] O.K. Mahabadi, G. Grasselli, A. Munjiza, Numerical modeling of Brazilian disc test of layered rocks using a combined finite-discrete
element method, in: Proc. of the 3rd CANUS Rock Mech. Symposium, Toronto, 2012.
[23] C. Rocco, J.V. Guinea, J. Planas, M. Elices, Size effect and boundary conditions in the Brazilian test: theoretical analysis, Mat. andStruct.
32 (1999) 437–444.
[24] M. Saridemir, Empirical modeling of splitting tensile strength from cylinder compressive strength of concrete by genetic programming,
Expert Systems with Application 38 (2011) 14257–14268.
[25] ASTM C496, Test Method for Splitting Tensile Strength of Cylindrical Concrete Specimens, ASTM Book of Standards.
[26] W.F. Chen, A.F. Saleeb, Constitutive Equations for Engineering Material Vol. 1: Elasticity and Modeling, John Wiley and Sons, 1982.
[27] FIB Bulletin No. 55 and 56, Model Code 2010 First Complete Draft Vol. 1 and 2, Federal Institute of Technology, Lausanne, Switzerland,
2010.
[28] B.R. Indriyantho, The numerical analysis of concrete behavior in tension with Brazilian splitting test method, Master Thesis,Master
Program in Struct. Eng., Diponegoro University, Indonesia, 2014, unpublished.