Smeared Crack Models for Reinforced Concrete Beams by Finite Element Method

© 2009 IBRACON a Grupo de Modelagem de Estruturas de Concreto / GMEC, Departamento de Engenharia de Estruturas e Geotécnica, Escola Politécnica da Universidade de São Paulo / EPUSP, renatomenin@gmail.com, Av. Prof. Almeida Prado tv.2, n.83 Cidade Universitária CEP 05508-900, São Paulo, Brasil; b Grupo de Modelagem de Estruturas de Concreto / GMEC, Departamento de Engenharia de Estruturas e Geotécnica, Escola Politécnica da Universidade de São Paulo / EPUSP, leandromt@gmail.com, Av. Prof. Almeida Prado tv.2, n.83 Cidade Universitária CEP 05508-900, São Paulo, Brasil; c Grupo de Modelagem de Estruturas de Concreto / GMEC, Departamento de Engenharia de Estruturas e Geotécnica, Escola Politécnica da Universidade de São Paulo / EPUSP, tbitten@gmail.com, Av. Prof. Almeida Prado tv.2, n.83, Cidade Universitária CEP 05508-900, São Paulo, Brasil.


Introduction
The reinforced concrete became one of the most important structural materials in the last century, being widely used in different areas of civil engineering.Due its low resistance to tensile stresses, the concrete structures suffer cracks even when they are subjected to low levels of load, which causes a reduction on their stiffness and modify the internal stress distribution, and as a result the structure starts to present a non-linear behavior.The crack modeling is a quite complex and intriguing topic for being associated to the discontinuity study in the displacement field.In the numerical simulation of reinforced concrete structures by Finite Element Method, three different tendencies have been developed to represent the cracks: discrete, smeared and incorporated.The discrete crack models are based on the idea of always working with the portion of the solid which remains continuous and without any damages, so that during the formation or progression of an existing crack, its sides incorporate to the outline of the solid.The cracks are modeled as displacement discontinuities among the finite elements and they must develop across their outlines, what ends up generating a restriction in its propagation direction.The application of these models is limited to the study of the problems involving the progression of few cracks, what hinders its use in most of the reinforced concrete structures.In the smeared crack models, the cracked material is treated as continuous, and the discontinuity of the displacement field caused by the crack is spread across the element by changing the constitutive equation.Despite the successful study of most of the reinforced concrete structures, among the deficiencies of these models, it can be pointed out the difficulty in studying the localized cracking situations, as well as the sensitivity problems concerning the finite element mesh.Later on, the incorporated crack models appeared which gather the favorable aspects of the previous two tendencies: there is no need of a mesh redefinition (deficiency observed in the discrete model) and the results obtained are independent from the finite element mesh (deficiency of the smeared model).The models are based on the concept of discontinuities incorporated in standard finite elements.

Smeared crack models
Since it has begun to be used in reinforced concrete structures, the Finite Element Method has shown more advantages by representing the cracks through changes of the constitutive equations (smeared models) instead of changes on the finite element mesh (discrete models).The first analyses were based on the idea of fragile failure, that is, to make the material stiffness null in the direction of the maximum tensile stress when it exceeded the tensile strength.Later on, it was observed that better results in the post-peak phase could be achieved by adopting a gradual reduction in the stress.In order to represent this behavior, the stress-strain diagram started to be defined by experimental values and, thus, several models were proposed (Cedolin e Dei Poli [1], Bazant and Gambarova [2], Gupta and Maestrini [3], and Vecchio [4]).Though successfully used to represent the behavior of the reinforced concrete structures which presented a crack pattern well distributed, when used to simulate the behavior of structures in which a crack is predominant (plain concrete or deep beams), these models presented some deficiencies.One of the deficiencies is the sensitivity regarding the mesh, caused by the non-consideration of concepts associated to Fracture Mechanics.So, it was necessary to apply the Fracture Mechanics concepts directly over the concrete structures analysis, creating a series of new models (Bazant and Cedolin [5], Bazant and Oh [6], Feestra and de Borst [7]).Rots and Blaauwendraad [8] presented a comparative study between the discrete and smeared crack models.This study introduced the idea of dividing the smeared crack models into fixed and rotating.In the fixed model, the crack orientation is kept constant during the whole computational process, while in the rotating model the crack orientation may change, following the main directions.There is also an intermediary option which is the concept of multidirectional fixed smeared crack.There are also the models based on the plasticity theory, which are able to represent well the pre-peak and post-peak phases, consisting of a constitutive model and a failure criterion.In this line it can be pointed out the models by Ottosen [9] and Pramono and Willam [10].

Smeared crack models in DIANA
DIANA code -version 9.1 was used to evaluate the performance of different smeared crack models.DIANA is a finite element software based on the displacement method for non-linear analysis of concrete structures, which has been developed since 1972 by Delft University of Technology in Netherlands.The smeared crack models are defined through the combination of three factors: a failure criterion (constant or linear), the cut transfer through the crack (total, constant or variable) and the material softening behavior (brittle, linear, multilinear or non-linear).In order to enable the combination of the crack model with a plastic behavior of the material, the total strain, ε, is decomposed in two parts, one is the elastic strain, ε e , and the other is the crack strain, ε cr .In the definition of the constitutive model, a criterion for the beginning of a new crack and its stress-strain relationship must still be established.For the beginning of a new crack, two pre-requirements are necessary: the principal stress must exceed a limit stress; and, in case there is a previous crack, the angle between the existing crack and the principal tensile stress must exceed the initial crack angle.Regarding the constitutive model, the stress-strain relationship in the normal direction of the crack may be expressed by: where, f t is the concrete tensile strength and the variable y is the function that represents the material softening diagram.The ultimate strain ε ult is assumed to be constant, and it may be obtained from the concrete tensile strength f t , the fracture energy G f , and the element area (represented by its equivalent length h):

Smeared Crack Models for Reinforced Concrete Beams by Finite Element Method
• Constant: it admits a constant value of the tensile strength and considers the beginning of a crack if the maximum principal tensile stress exceeds the concrete tensile strength.

• Linear:
it admits a linear variation and considers the beginning of new cracks, if the principal tensile stress exceeds the minimum value between f t and f t (1+ σ lateral / f c ), in which σ lateral is the principal lateral stress and f c is the concrete compressive strength.In the smeared crack models, it is adopted a reduction parameter β for the transverse stiffness of the material known as "shear retention factor".Thus, the shear stiffness may be expressed by: The parameter β may range from 0 to 1.For values next to zero, it is achieved a very small concrete shear stiffness and in the opposite case, for the unitary value it is obtained an infinite shear stiffness that will prevent the formation of cracks in this direction.Following the recommendations found in the literature, in the numerical examples analyzed in this work it was adopted β = 0.2.In the multidirectional smeared crack model it is possible to adopt a plasticity criterion for the material: Tresca, von Mises, Mohr-Coulomb or Drucker-Prager, and to choose the concrete tensile behavior in the softening branch.Among the different softening rules of the material available in DIANA it can be pointed out: brittle, linear, and non-linear by Moelands-Reinhardt and Hordijk, according to Figure 1.The brittle behavior is characterized by the total reduction of stress after the failure criterion has been achieved.This behavior may be described by the following expression: In the case of the linear softening model, the stress-strain relationship in the crack is defined by the following expression: Two other non-linear models may still be adopted.The softening model by Moelands-Reinhardt uses a non-linear relationship between stresses and strains expressed in the following equation, with the coefficient c 1 taking the constant value of 0.31: The other non-linear model present in DIANA is the model by Hordijk which uses an exponential relationship between the normal tensile where: The fracture energy G f may be calculated in two ways: the first one based on the model code CEB-FIP [11] and the second based on the laws of the softening and hardening branches of concrete.Another possibility is to use the hardening-softening laws with rheological models originated from the plasticity theory.Among the smeared crack models available in DIANA, it may be pointed out: fixed, multidirectional and rotating models.In the present work just the multidirectional smeared crack models were used.

Multidirectional smeared crack models
The multidirectional smeared crack models are known for presenting an intermediary behavior between the fixed and rotating models.These models allow the propagation of several cracks at the same point and according to the definition of the parameter called "threshold angle", a fixed or rotating model may be obtained.This parameter represents the angle between an existing crack and another one formed at the same point and its standard value is 60º.If this parameter is changed to 90º, a fixed model is obtained and if changed to 0º a rotating model is obtained.In the present work, it was used the standard value defined by DIANA.The multidirectional models present some differences in relation to the fixed models, among which it may be pointed out: different concrete models for tensile stresses, an option to decrease or not the transverse stiffness and the possibility to use constitutive models originated from the plasticity theory.The behavior of the concrete subjected to tensile stresses may be defined in two different forms: constant or linear: (3) ( ) stresses and strains, with the coefficients c 1 and c 2 taking, respectively, the values of 3.0 and 6.93, according to the following equation: The multidirectional model available in DIANA is characterized for modeling the material combining a smeared crack model (brittle, linear, exponential,...) with a plastic model for compression, in which it is possible to use the classical failure models by Tresca, Von Mises, Mohr-Coulomb and Drucker-Prager.The constitutive model for the reinforcement is the perfect elasto-plastic.By adopting this model, the reinforcement stress may be determined by the equations ( 9) and (10), where f y is the yield stress, E s is the elastic module and ε s is the reinforcement strain:

Smeared crack models in VecTor2
VecTor2 is a finite element software used in the analysis of twodimensional reinforced concrete structures which has been developed at the University of Toronto since 1990.The software development has been happening parallel to a series of experimental tests, in order to better predict the behavior of a variety of reinforced concrete structures.
The theoretical bases of VecTor2 are the MCFT -Modified Compression Field Theory, developed by Vecchio and Collins [12] and DSFM -Disturbed Stress Field Model, developed later by Vecchio [4].These formulations are analytical models to predict the response of reinforced concrete elements, representing the cracked concrete as an orthotropic material with rotating smeared cracks.The main concepts of the MCFT and DSFM formulations will be addressed in the next sections.

Modified Compression Field Theory (MCFT)
The MCFT is an analytical model to represent the behavior of two-dimensional reinforced concrete structures discretized by membrane elements subjected to normal and shear stresses, as shown in Figure 2.This formulation evaluates the average stresses and strains (in the region between the cracks) and the local stresses and strains of the concrete and reinforcement, as well as the widths and orientations of the cracks during the loading and, based on this information, the failure mode of the element can be determined.The modeling of the cracked concrete element is performed by using an orthotropic material and a rotating smeared crack model.In other words, the cracked concrete is treated like a continuous medium with cracks smeared across the membrane element.The smeared cracks may suffer changes in their directions always remaining linked to the direction changes in the principal stress field.The MCFT is based on three groups of relationships: compatibility relationships for the average strains in the concrete and reinforcement; equilibrium relationships for the average stresses in the concrete and reinforcement; and the constitutive relationships for the cracked concrete and reinforcement.Among the hypotheses admitted in the MCFT formulation, it may be pointed out: the uniformly distributed reinforcement; uniformly applied shear and normal stresses; perfect bond between the concrete and the reinforcement; uniformly distributed and rotating cracks; and the orientation of principal stress and strain are the same. (8) Smeared Crack Models for Reinforced Concrete Beams by Finite Element Method

Compatibility relationships in the MCFT
The compatibility relationships in the MCFT are achieved from the average strains in the components which represent the concrete and the reinforcement, according to Figure 3. Based on the perfect bond hypothesis, the average strains in the concrete (ε c ) are equal to the average strains in the reinforcement (ε s ).Although any number of components and reinforcement orientations may be considered in the MCFT formulation, considering the membrane element orthogonally reinforced of Figure 2, the compatibility equations may be expressed by: With a value for the shear strain γ xy , it is possible to obtain the average principal concrete tensile strain (ε c1 ) and the average principal concrete compressive (ε c2 ), as well as the orientations of the average principal tensile strain and stress axes using Mohr's circle:

Equilibrium relationships in the MCFT
Considering the free body diagram of the membrane element shown in Figure 4, through the balance of forces in the x and y directions, the results of the normal stresses applied in panel σ x e σ y must be balanced by the normal stresses in the concrete f cx and f cy and by the stresses in the reinforcement f sx e f sy .The balance of moments requires that the shear stress applied to the panel, represented by τ xy , to be totally resisted by the average shear stress in the concrete τ cxy , assuming that the reinforcement does not present any kind of dowel action.Thus, these equilibrium relationships for the average stresses may be expressed by the following equations, where ρ sx and ρ sy are the reinforcement ratios in the x and y directions: Since the cracked concrete is orthotropic with respect to the directions of the principal stress, Mohr's circle can be used to relate the average concrete stresses f cx and f cy , to the average principal concrete tensile stress f c1 :

Constitutive relationships in the MCFT
A set of thirty panels measuring 890x890x70 mm was tested experimentally and the results were analyzed to develop constitutive models able to represent the behavior of the cracked concrete in compression and tension.The constitutive relationship to describe the concrete in com- (11) pression relates the principal compressive stress, f c2 , to the principal compressive strain, ε c2 .The tests with the panels have indicated that the compressive strength and stiffness suffered reductions as the principal tensile strain, ε c1 increased.This phenomenon, known as compression softening is incorporated through the stiffness reduction in the stressstrain curve of the concrete: The term which appears in the numerator is the Hognestad's parabolic relationship, obtained for the concrete subjected to the uniaxial compression, generally used for normal strength concrete.The value ε 0 corresponds to the strain associated to the peak stress of the concrete f c ', as determined experimentally from uniaxial compression tests of concrete cylinders.The denominator reflects the softening effect.With regards to concrete in tension, it is first necessary to determine the values of the uniaxial cracking strength of the concrete, f t ', and its corresponding tensile strain, ε cr , which may be obtained through the following expressions: where E c is the initial tangent stiffness of the concrete, estimated by the following expression: It is admitted that before cracking, the concrete presents a linear-elastic behavior in tension.Therefore, after cracking, tensile stresses may continue to exist in the concrete between the cracks due to bond interactions between the concrete and reinforcement.To model this phenomenon known as tension stiffening, the MCFT proposed the following relationship: For the reinforcement in compression and tension, the MCFT uses a bilinear relationship to relate the average stress and strain, in the same way it had been done with the crack model in DIANA.
where E s is the elastic module and f s yield is the yield stress of the reinforcement. (20 ) Smeared Crack Models for Reinforced Concrete Beams by Finite Element Method

Local crack conditions in the MCFT
Given a compatible average strain condition, the constitutive relationships can be used to determine the average stresses in the concrete and reinforcement, as well as the normal and shear stresses acting on the structure.However, it would be unconservative to disregard the possibility that the response of the structure may be guided by the local yielding of the reinforcement at the crack or sliding shear failure along a crack.To cover these possibilities, the MCFT limits the local stresses at the crack and average tensile concrete stress.The stress field in the reinforced concrete suffers variations when evaluated according to the average values in the area between the cracks and when evaluated locally at the crack.This behavior can be better understood considering Figure 5(a), which depicts the average stresses in a section located between cracks, and Figure 5(b), which depicts the local stresses at the free surface of the crack.At a free surface of the crack, the average concrete tensile stresses are practically reduced to zero.Consequently, to transmit the average tensile stresses across the crack, the stresses and strains in the reinforcement must increase locally at the crack.Static equivalence between the average and local stresses in the normal direction to the crack surface, results in the following relationship: where f scrx and f scry are the local reinforcement stresses at a crack, and θ nx and θ ny the values of the angles between the normal to the crack and the reinforcement.Considering the equation (27), the average tensile concrete stress must be limited by the yielding of the reinforcement at the crack and consequently: As a principal plane, the shear stress is absent and thus, they do not appear in Figure 5(a).However, as the reinforcement generally crosses the cracks at a skew angle, the local shear stresses, τ ci will be present on the crack surface.Consequently, through the static equivalence of the average and local stresses in the tangential direction to the surface of the crack, the local shear stresses can be determined by the following equation: However, the local shear stress is limited by the aggregate interlock mechanism, which decreases with the increase of the crack width (w) and with the reduction of the size of the aggregated (a).Based on the analysis of the aggregate interlock mechanisms developed by Walraven [13], the MCFT establishes a limit for the shear stress on the crack:

Disturbed Stress Field Model (DSFM)
The DSFM is an extension of the MCFT, with the purpose of treating the MCFT deficiencies.For example, in lightly reinforced elements, in which the shear slip along the crack is more expressive, the rotation of the stress field tends to present a certain discrepancy concerning the rotation of the strain field.In these cases, the stiffness and strength end up being overestimated by the MCFT, which assumes the orientations of the principal stress and strain are the same.On the other hand, in elements which present small rotations in the stress and strain fields, the MCFT generally underestimates the shear stiffness and strength.The DSFM is conceptually similar to the MCFT, however, it ends up extending the MCFT in several aspects.Primarily, the DSFM enlarges the compatibility relationships of the MCFT to include crack shear slip.Furthermore, the DSFM decouples the orientation of the principal stress and strain fields.By explicitly calculating crack slip deformations, the DSFM eliminates the need to check the shear stress.Modifications in the calculation of the constitutive relationships for the concrete and reinforcement are also presented.

Compatibility relationships in the DSFM
Although the MCFT admits that the orientations of the principal stresses and strains remain the same, experimental tests indicate that this hypothesis is not always valid after the crack.The experimental results show that the principal strain field generally suffers modifications in its orientation at a larger rate than in the principal stress field.
This phenomenon is attributable to the manner by which the stress and strain fields are determined.The measured strains are total strains, which are obtained by the sum of the average strains in cracked concrete (Figure 2) and the shear strains, caused by slipping across the crack, as shown in Figure 6.In order to represent this behavior, the DSFM defines the total strains, ε x , ε y and γ xy as the sum of the net concrete strains, ε cx , ε cy and γ cxy , and the strains due to shear slip, ε x s , ε y s and γ xy s .
The principal net concrete tensile strain (ε c1 ) and principal net concrete compressive strain (ε c2 ) may be evaluated using Mohr's circle: The crack slip shear components, ε x s , ε y s and γ xy s , can be calculated from the average crack slip shear strain γ s .This strain is equal to the slip in the crack, δ s , divided by the average space between the cracks, s, according to the following expression: The orientation of the principal net concrete strains, θ, and the orientation of the principal concrete stress, θ σ , with respect to the x axis can be determined from Mohr's circle with the components of the net concrete strain as follows: Likewise, the orientation of the principal total strain field, θ ε , can be determined from the components of the total strain: Although in the DSFM, it is possible to use any number of reinforcement components and orientations, considering the orthogonally reinforced membrane element and assuming perfect bond, the average strains in the components of the reinforcement in the x and y directions will be equal to the total strains:

Equilibrium relationships in the DSFM
Considering again an orthogonally reinforced membrane element, the equilibrium relationships of the DSFM are the same determined previously for the MCFT, that is:

Smeared Crack Models for Reinforced Concrete Beams by Finite Element Method
Similarly, the DSFM formulation also incorporates the equilibrium relationships for the local stresses at the crack: It is worthy pointing out that the average concrete tensile stress must be limited by the yielding of the reinforcement traversing the crack: However, unlike the MCFT, the tensile stress is not subjected to the limitation of the shear stresses at a crack, since the DSFM incorporates deformations due to shear slip, instead of imposing a limit stress corresponding to the shear slip failure.

Constitutive relationships in the DSFM
The experimental analysis of a series of reinforced concrete panels led Vecchio and Collins [12] to propose a reduction factor, β d , in order to reflect the softening effect of the concrete and its connection to the principal tensile strain.
In this equation, the softening effect of transverse tensile strains is taken into account by factor C d , defined by: The factor C s determines if the analysis of the element will take into account slip deformations.If the analysis does not consider the slip deformations, as in the MCFT, then factor C s = 1.0.If the analysis considers element slip distortions, as in the DSFM, then the softening effect appears to be less for the same value of the relationship ε c1 /ε c2 since the softening effect is attributed only to the tensile strains and in this case, C s = 0.55.In order to consider the concrete softening effect, the concrete cylinder strength f c ', and its corresponding strain peak, ε c ', are both reduced: With regards to the constitutive models to simulate the behavior of the concrete in tension, it is admitted that the response before crack is linear-elastic and for cracked concrete the average concrete tensile stresses, f c1 a , due to tension stiffening, may be modeled by a nonlinearly decaying relationship: The coefficient c t incorporates the influence of the reinforcement bond characteristics and can be determined by the following equation: where d bi is the bar diameter and ρ i is the reinforcement ratio of each of the n reinforcement components.In the DSFM, a trilinear constitutive model for reinforcement in tension or compression is used to account for strain-hardening phenomenon: where E s is the elastic module; f syield is the yield strength of the reinforcement; ε syield is the yield strain of the reinforcement; ε sh is the strain at the onset of strain hardening; and ε u is the ultimate strain of the reinforcement.

Composite material stiffness matrix
The composite material stiffness matrix is the sum of the concrete material stiffness matrix D c and the reinforcement component material stiffness matrices D si : As the MCFT and DSFM model the reinforced concrete element as an orthotropic material in the principal stress direction, it is necessary to formulate the concrete material stiffness matrix relative to these directions, D c '. Assuming that the Poisson's effect is negligible, then D c ' can be calculated by: The secant modules are computed from the current values of the principal stresses f c1 and f c2 and the corresponding net concrete strains ε c1 and ε c2 : Likewise, the reinforcement component matrices D si ' must be determined initially relative to their longitudinal axes.Assuming that the reinforcement only resists to uniaxial stresses, the matrix D si ' can be determined by: Afterwards, the material stiffness matrices D c ' and D si ' initially evaluated relative to the principal axes are transformed to the x and y axes by means of the transformation matrix:

Simply supported reinforced concrete beam -Leonhardt and Walther [14]
A set of four reinforced concrete beams (ET1, ET2, ET3 e ET4) experimentally analyzed by Leonhardt and Walther [14] will be studied numerically using different smeared crack models in DIANA and VecTor2.The beams are simply supported and submitted to vertical loads as shown in Figure 7.
The beams have the same amount of reinforcement but they have different core widths, what implies in different reinforcement ratios.
(57)  The concrete cube strength, achieved experimentally by Leonhardt and Walther [14], was f cub = 28.5 MPa.The corresponding values for the concrete cylinder strength (f c = 22.8 MPa), the concrete cracking strength (f t = 1.57MPa) and the initial elastic module (E co = 23874.67MPa) were determined from the concrete cube strength, according to Vecchio and Collins [12].The set of reinforced concrete beams was analyzed with smeared crack models, employing different softening rules from DIANA (brittle, linear, non-linear by Moelands-Reinhardt and Hordijk) and the plasticity criterion by von Mises, as well as using the DSFM from VecTor2.The parameters used in the softening models are showed in Table 1.The perfect elasto-plastic behavior and elastic module E s = 210000 MPa were considered for the reinforcements.The finite element mesh used for modeling the concrete structure with 341 nodes and 300 rectangular elements is shown in Figure 8. Incorporated models were used in DIANA for the discretization of the reinforcement.On the other hand, VecTor2 used discrete models for the longitudinal reinforcement and smeared models for the stirrups.The vertical displacements of the beams at Point A (midspan) are shown in Figure 9.It can be observed that the load x displacement curves achieved with the different models present a great concordance with the experimental results.In general, the computational models presented a little higher stiffness than the experimental results.
The additional stiffness associated to the consideration of perfect bond and the contribution of concrete between cracks is more significant for smaller reinforcement ratios.The beam ET1 (with the smaller reinforcement ratio) presented the biggest discrepancy to the experimental results.The discrepancy between numerical and experimental models was less significant in the beams with higher reinforcement ratios (ET2, ET3 and ET4), as according to d'Avila [15].To better understand the behavior of the beams and their collapse mechanisms, it is shown in Figure 10, the crack pattern and the orientations of the principal tensile strains ε 1 and principal compressive strain ε 2 of the beam ET4 for a load P=76.8 KN.

Continuous reinforced concrete beam -Leonhardt and Walther [16]
A   the initial elastic module were determined from the concrete cube strength according to Vecchio and Collins [12].The parameters adopted for the Hordijk's model are showed in Table 3.The perfect elasto-plastic behavior and elastic module E s = 210000 MPa were adopted for the reinforcement.
The finite element mesh used for modeling the concrete structure with 369 nodes and 320 rectangular elements is shown in Figure 12.Incorporated models were used in DIANA for the discretization of the reinforcement.On the other hand, VecTor2 used discrete models for the longitudinal reinforcement and smeared models for the stirrups.
The vertical displacements of the beams at Point A are shown in Figure 13.The load x displacement curves achieved with the different models present a great concordance with the experimental results.At the beginning of the loading process the computational models presented a smaller stiffness than the experimental results, however, for the final phase, this situation inverts and the computational models start to present a little higher stiffness than the experimental results.
The additional stiffness in the computational models is caused by two factors: the consideration of the perfect bond and the contribution of concrete between cracks, which was also observed in the previous example.
For this series of continuous beams, it is also showed a comparison of the reinforcement stresses for the different models.The analyzed points are: under the load for the positive reinforcement (Point C) and at the central support for the negative reinforcement (Point B), as shown in Figure 14, indicating that the computational models were able to evaluate properly the reinforcement stresses during the loading process.

Conclusions
The smeared crack models of DIANA and VecTor2 proved to be very efficient to analyze the reinforced concrete beams subjected to bending.The load x displacement curves obtained with the different models presented a great concordance with the experimental values.In general, the computational models presented a higher stiffness than the experimental results.Among the causes of the additional stiffness, it can be pointed out the concrete contribution between cracks, which proved to be more significant for smaller reinforcement rates and the hypothesis of perfect bond, since due to the non-consideration of the bond loss, it is natural that the cracked structure becomes stiffer.The selection of the material softening rule showed to be more significant for smaller reinforcement ratios, in which the brittle model presented the highest discrepancy with respect to the experimental values.
The smeared crack models permitted to simulate the post-crack behavior of the reinforced concrete allowing to follow up the evolution of stresses and strains during the loading process and to obtain a crack pattern which enabled to better understand its be-havior, as well as to follow up the stress evolutions and the yielding of the reinforcements.

Figure 5 -
Figure 5 -(a) Average stresses between cracks and (b) local stresses at crack free surface

Figure 6 -
Figure 6 -Deformations due to crack shear slip

Figure 7 -
Figure 7 -Reinforced concrete beams (ET1, ET2, ET3 and ET4) for Reinforced Concrete Beams by Finite Element Method The longitudinal reinforcement at the bottom of the cross section consists of four bars with diameter of 20 mm and yield stress f y = 428 MPa.The longitudinal reinforcement at the top consists of two bars with diameter of 8 mm and yield stress f y = 465 MPa.The transverse reinforcement is represented by stirrups of 6 mm and yield stress f y = 320 MPa.
set of three reinforced concrete beams (HH3, HH4, and HH5) tested experimentally by Leonhardt and Walther [16] will be analyzed using a smeared crack model with the softening rule by Hordijk and the plasticity criterion by von Mises in DIANA and also with the DSFM in VecTor2.The beams have rectangular cross sections (25 x 32 cm 2 ), presenting different lengths and reinforcement ratios, as shown in Figure 11.The positive and negative longitudinal reinforcements are represented by bars of 14 mm with yield stress f y = 417 MPa.The transverse reinforcements consist of stirrups of 8 mm with yield stress f y = 371 MPa.The length of the beams and their respective mechanical properties are shown in Table 2. Likewise the previous example, the concrete cylinder strength, the concrete cracking strength and

Table 1 -
Mechanical properties for cracked concrete

Table 2 -
Mechanical properties of concrete

Table 3 -
Parameters for the Hordijk model