You are on page 1of 6

MODELLING THE NONLINEAR BEHAVIOUR OF R/C BEAMS WITH

MODERATE SHEAR SPAN AND WITHOUT STIRRUPS USING ANSYS

Sonia L. Parvanova1, Konstantin S. Kazakov2, Irina G. Kerelezova1,


Gospodin K. Gospodinov1 , Mogens P. Nielsen3

University of Architecture, Civil Engineering and Geodesy


Faculty of Civil Engineering, 1 Hr. Smirnenski blv., 1046 Sofia

Abstract: A special purpose 3D concrete finite element is employed in this research to


model the complex nonlinear behaviour of reinforced concrete beams with no shear
reinforcement. The theory is based on the 5 parameter plasticity failure surface due to
Willam-Warnke and takes into account the tensile cracking in three orthogonal directions
plus crushing status of certain Gaussian integration point. A comparison is made with
experimental and numerical results taken from the literature. The load deflection curve
analysis indicates a big sensitivity of the solution to small changes of the material
parameters which shows that a further research for their adjustment is needed. Although in
a smeared manner the shear cracking is available, the preliminary conclusions suggest
that the typical big sliding accompanying the development of the critical shear crack is not
fully handled by the present model.

Key words: Fracture mechanics, flexural and shear cracks, reinforced concrete, ANSYS

1. Introduction
The experimental data of R/C beams working in flexure suggests a close agreement
between the test results and those from numerical simulations. For deeper beams working in
flexure and shear, on the other hand, there is much larger difference between predictions and test
results, indicating that the problem is very complex and not fully understood. It is however well
known fact that the flexural failure of R/C beams is generally ductile, whereas the shear failure is
usually brittle, so the shear transfer mechanism is depending of many different geometric and
material parameters.
Beam data
F F L=1800 mm
loading plates a= 600 mm
b= 100 mm
d= 240 mm
h= 200 mm
d concrete beam h a/h=3
concrete:
Ec=35500 N/mm2
bearing plate ft=2.88 N/mm2
2Φ16 40 mm fc=26.7 N/mm2
b а L-2a а b
steel:
Es=206000 N/mm2
L fy=396 N/mm2
Figure 1. Geometric and material data for the reinforced concrete beam with no shear reinforce-
ment analyzed in paper [3]. The shear span ratio in this case a/h=3 is considered moderate

1
University of Architecture, Civil Engineering and Geodesy, Sofia, Bulgaria
2
Lozenetz-consult Ltd, Sofia, Bulgaria
3
Technical University of Denmark, Lyngby, Kopenhagen, Denmark
To understand the complex nature of the interaction that takes place between the cracked
concrete and reinforcement, rectangular cross section beams with flexural reinforcement, but
without shear reinforcement (stirrups) will be considered in this numerical research. Many test of
simply supported R/C beams with no stirrups under two symmetrical concentrated forces are
reported all over the world - see references [1] and [2]. In this paper we use the test results of
Hibino et al. [3], where along with the experimental results a nonlinear numerical model is
proposed and some useful comparisons and conclusions are made – see Fig. 1 for the particular
beam geometric and material data details.
2. The 3D ANYS concrete finite element and its options
The implementation of nonlinear material laws in finite element analysis codes is usually
tackled by software developers by programming the material behaviour independently of the
elements to which it may be specified. Another idea is employed in ANSYS, where a special
three-dimensional eight noded solid isoparametric element, Solid65 is developed. It models the
nonlinear response of brittle materials and is based on a constitutive model for the triaxial
behaviour of concrete after Williams and Warnke - see Fig. 2. (c) and [4] for details.

8 Gaussian integration points


(a) (c)

critical shear crack flexural cracks

Flexural and shear cracks development taken from the experiment


8 Nodal points of the brick finite element

(b) (d) y
300 concrete (only) elements Integration point

crack plane

60 concrete and smeared steel elements


z
Finite element mesh principal stress direction

Figure 2. (a) Crack development picture taken from the test [3]; (b)Finite element mesh; (c) The
concrete brick element – nodal and Gaussian integration points; (d) Crack plane at a given
Gaussian integration point is perpendicular to the principal tensile stress direction

The element includes a smeared crack idea for handling cracking in tension zones and a
plasticity algorithm to account for the possibility of concrete crushing in compression zones.
Each element has eight integration points (Fig. 2. (c) and (d)) at which cracking and crushing
checks are performed. The element behaves in a linear elastic manner until either of the specified
tensile f t or compressive f c strengths is exceeded. Cracking (or crushing) of an element is
initiated once one of the element principal stresses, at an element integration point, exceeds the
specified tensile or compressive concrete strength. The formation of a crack is achieved by the
modification of the stress-strain relationships of the element to introduce a plane of weakness in
the principal stress direction. Subsequent to the formation of an initial crack, stresses tangential
to the crack face may cause a second even third crack to develop at considered integration point.
The amount of shear transfer across a crack can be varied between full shear transfer and
no shear transfer at a cracked section. In ANSYS that option is controlled by two coefficients -
β1 for an open crack and β 2 for a closed crack. They are so called shear retention factors and
range between 0 (no aggregate interlock) to 1 (full aggregate interlock).
The reinforcement may be modeled as an additional smeared stiffness distributed through
the element in a specified orientation (as used in this paper) or alternatively by using discrete
strut or beam elements connected to the solid elements. The smeared stiffness which is the
option employed in this research, allows the elastic-plastic response of the reinforcement to be
included in the simulation. No shear stiffness of the reinforcing bars can be modeled, so an
important factor such as dowel action is excluded in the shear transfer mechanism.
As shown in Fig. 2.(b) half of the beam is analyzed and the finite element mesh consists of
300 concrete elements plus another 60 “mixed” elements put on the lower side of the beam. In
those elements the flexural steel bars are “smeared” over the concrete material representing the
required additional stiffness in x direction.
3. Solution strategy and some numerical results
Many numerical tests were performed for the beam described in Fig. 1 using different
constitutive options and parameters. It became clear that the flexural and shear crack
development is of a paramount importance for the beam carrying capacity and its failure mode.

(а) F Plane of symmetry


Load step 20 of 89:
F=10260 [N]
δ=0.756 [mm]
typical bending vertical cracks
flexural vertical cracks
no yielding in steel
no concrete plasticity

V
(b) 8- flexural-shear cracks development of the bending cracks

Load step 69 of 89:


F=36720 [N]
δ=3.781 [mm]
developed flex. cracks
developed flex/shear cr.
just before initiation of
the critical shear crack

V idea of the critical shear crack cracks due to compression (splitting cracks)
(c) 8-
Load step 87 of 89:
F=46440 [N]
δ=7.600 [mm]
full cracking state
yielding in steel
concrete plasticity
big displacements incre-
ment just before collapse

V
8- 3. Bending and shear crack development at the front side of the beam:(a) Purely flexural
Figure
(bending) vertical cracks at step 20; (b) Additional flexural - shear cracks at step 69; (c) Final
crack development at step 87 just before the collapse – the critical shear diagonal crack is clearly
identified and there are splitting cracks at the upper part of the beam due to compression
In Fig. 3 crack developments are shown for three of the most important load steps. As
reported in work [5] we found that ANSYS crushing option does not work properly for
simulation like that, because at the moment few integration points located at upper part of the
beam exceed the compressive concrete strength they do not contribute anymore to the stiffness
matrix. As a result the shear resistance capacity at the vertical cross section is very low and the
solution does not converge. In our case (see Fig. 4) the solution stops very soon after the
development of bending cracks (at about 3 mm deflection), so it was not able to reproduce the
shear cracks development, not to speak about yielding of the longitudinal tensile reinforcement
steel. Under those circumstances the “crushing” option was removed and the concrete plasticity
(crushing) in the compression zone was modeled using von Mises plasticity multilinear option.
The following well known nonlinear uniaxial stress-total strain relationship was used [5], being
discretized at 6 points of equal distance ranging between 0 and ε 0 :
E cε
σ (ε ) = 2
, (1)
ε 
1+  
 ε0 
where ε 0 = 2 f c / Ec , Ec = σ / ε , and σ is the stress at any strain ε , ε 0 is the strain at the
ultimate compressive strength f c and Ec is the Young’s modulus for the concrete.
For the nonlinear loading program the number of loading substeps was chosen 100 and the
beam was loaded up to failure using load control option. Activation of the above concrete
plasticity option led to completely new crack development picture – see Fig. 3 (a), (b) and (c).
The nonlinear numerical solution was able to completely reproduce the full range of cracking
including the pure flexural (Fig. 3.(a)), flexural-shear (Fig. 3.(b)), and the critical shear crack, see
Fig. 3.(c). There is a clear indication that the path of the shear cracks follows the trajectory of the
principal stresses.
It is instructive when discussing the nonlinear crack propagation and plasticity softening to
follow the force-midspan deflection response given in Fig.4. The bending cracks development in
the experiment is quite smooth whereas in the numerical solution curve it is flat and “sudden”.
The explanation is simple. The ANSYS cracking option does not include properly the tensile
stress relaxation, to say it in other words the fracture energy parameter G f is not included in the
model as an important material constant. That fact does not generally affect the solution since the
tensile steel capacity is available, therefore the sudden stress drop at the cracking points explains
the discrepancy between the two curves at the beginning of nonlinear process.
Once the tensile cracking phase is completed the beam develops some bending-shear type
of cracks which are shown in Fig. 3 (b). They initiate as bending cracks in the shear beam span
and later turn over to diagonal shear crack at upper part following the principal stress direction.
At the same time a growing of the bending cracks is observed, so the need of additional capacity
activates the main shear transfer mechanism. That involves the initiation and quick propagation
of the so-called critical shear crack which is happening immediately after load step 69 with
“jump” of deflection value from 3.78 mm to 4.02 mm. The propagation of the critical diagonal
crack provokes grow up of concrete plastic strain and relevant material softening, although the
aggregate interlock mechanism is on through beta factors (in our case β1 = β 2 = 0.6 ). The tensile
steel yielding begins at deflection value of about 5 mm and after that the solution needs more and
more iteration to converge. The numerical curve becomes flat which corresponds very well to the
experiment, so after sudden and big increasing of deflection of about 9 mm the solution stops.
It is interesting to note that just before the collapse few splitting cracks appear at the upper
part of the beam due to crushing failure of the concrete there. Probably that fact including the big
amount of plasticity deformations in the steel are the reason for failure of the numerical solution.
Figure 4. Force - deflection curves comparison between experimental test and numerical solution

4. Discussion
It is well known that the good correlation of test and numerical solutions depends on the
assignment of accurate linear and non-linear material properties. In general given the
compressive strength f c , taken from the experiment of the concrete, it is thus usually possible to
arrive at a sensible set of material data for inclusion in the nonlinear numerical model. The other
material parameters such as Young’s modulus Ec and tensile strength f t are treated as generic
data and could be assessed by the existing rules of thumb. For example the BS code for
reinforced concrete estimates the tensile strength of concrete from its known compression
strength by the formula: ft = 0.36 (fc)1/2, whereas for the Young’s modulus the following equation
could be used: Ec = 9100 (fc)1/3. In our case the material parameters f c , f t and Ec are explicitly
given in paper [3] without comments how they are received. Taking as a basis the value of
f c = 26.7 N / mm 2 and exploring the above equations we get f t = 1.86 N / mm 2 (against 2.88 in
[3]), and Ec = 27195 N / mm 2 (against 35500 in [3]). The differences are quite big and that could
affect the numerical solution accordingly. The situation is much clear in the context of the
reinforcing bars. Generally the nominal strength of the reinforcement is specified and it is
assumed in design that it behaves in an elastic-perfectly plastic manner.
Without going into details we shall mention that a kind of a parametric study was
performed and a number of numerical solutions were analyzed in order to better estimate the
sensitivity of different factors. For example calculating a different value for the tensile strength
f t having in mind the size of concrete finite element and value of fracture energy
GF = 0.1 N / mm , we get very good fit between numerical and experimental curves at the first
nonlinear phase up to deflection of 2 mm. After that no influence of f t factor was observed. It
was also established that the numerical solution is very sensitive to the steel constitutive
nonlinear modeling and the small changes of steel material parameters, such as the value of
Young’s modulus Es or yield stress fy. For example employing a bilinear elastic - perfectly
plastic model we get a stiffer model. When using a multilinear stress-total strain relationship
(similar to equation (1)) for the steel reinforcement bar, the numerical solution gets the type
given in Fig. 4. The discrepancy for the value of ultimate load obtained by means of the
numerical solution and experiment is about 4.2 %. We shall also mention that in the numerical
simulation the value of steel yielding stress is 11 % less then the one given in paper [3].
5. Conclusions
This work should be considered as part of a research project related to development of new
numerical models for nonlinear analysis of structures made from quasibrittle materials like
concrete using fracture mechanics approach.
Based on the points raised in numerical results and discussion sections, we draw the
following main conclusions from this numerical research:
ØThe general conclusion is that using 3D ANSYS modeling we are able to properly
simulate the nonlinear behaviour of R/C beams without shear reinforcement having a
moderate shear span size (a/h=3 for the beam under consideration);
ØANSYS 3D concrete element is very good concerning flexural and shear crack
development but poor concerning the crushing state. However this deficiency could be
easier removed by employing a certain multilinear plasticity model available in ANSYS;
ØThe particular concrete finite element does not consider one of the most important
fracture mechanics parameters – the fracture energy GF. That means that in the case of
concrete beam with no reinforcement we will be not able to get a solution;
ØBy using ANSYS smeared approach for beams with moderate shear span we are not able
to reproduce satisfactory the softening due to big sliding emerging at the critical shear
crack. That is likely to be more realistically achieved by 2D discrete crack approach;
ØThe results and the parametric study (not given in the paper) suggest that we need some
correction factors to adjust the values of material parameters available from experiment
and convert them to effective parameters related to the particular modeling;
ØTherefore much more research is needed in order to: (1) develop a similar simulation for
R/C deeper beams; (2) suggest reliable methods for adjusting the experimental material
data to effective parameter data suitable to particular finite element models.

Literature
1. Nielsen, M. P., Limit analysis and concrete plasticity, second edition, CRC press, 1999.
2. Leonhardt, F., R. Walter, Schubversuche an einfeldrigen Stahlbetonbalken mit und
ohne Schubbewehrung, Deutscher Ausschuss für Stahlbeton, Heft 151, Berlin,1962.
3. Hibino, K., T. Kojima, N. Takagi, FEM study on the shear behaviour of R/C beam
by the use of discrete model, Proceedings of the DIANA Conference - Constitutive
models for quasi-brittle materials, October 2002.
4. ANSYS , ANSYS 9.0 Manual Set, ANSYS Inc., Southpoint, 275 Technology Drive,
Canonsburg, PA 15317, USA.
5. Kachlakev, D., T. Miller, FE Modeling of Reinfoced Concrete Structures
Strengthened with FRP Lamiates, Final Report SPR 316, Oregon State Uiversity,
Department of Transportation, May 2001.

Acknowledgements
Funding for this project and the research described herein was supplied by the National
Science Fund under the contract № TH – 1406/04. Their support is gratefully acknowledged.

You might also like