Proceedings of the 7th International Conference on Civil Structural and Transportation Engineering (ICCSTE'22)

Niagara Falls, Canada – June 05-07, 2022

Paper No. 192
DOI: 10.11159/iccste22.192

Evaluation of Pre-tensioned Prestressed Concrete Beam Behavior by

Finite Element Analysis Using ATENA 3D
Zaher Alkurdi 1
1
Budapest University of Technology and Economics
Műegyetem rkp. 3, 1111, Budapest, Hungary
zaherkurdi@edu.bme.hu

Abstract - This research involves a numerical study of the nonlinear behavior of pre-tensioned prestressed concrete beams with
multiple layers of strand and shear reinforcements under flexural loading. First, a validation of the model was performed based on the
results of the standardized four-point static bending test reported in previous experimental research. Then, the influence of the
maximum aggregate size on the ultimate load of the pre-tensioned prestressed beam with reduced shear reinforcement was studied. The
ATENA 3D software could simulate the experimental tests and provide meaningful results. Moreover, a slight increase in the ultimate
load of the member was observed with a larger maximum aggregate.

Keywords: Prestressed concrete, numerical analysis, aggregate size, four-point static bending test

1. Introduction
Reinforced concrete is a structural material that is extensively utilized in civil engineering. A reinforced-concrete
building system is the most lasting and low-maintenance solution. To achieve equilibrium and control of deformation
following cracking, reinforcement is inserted in the concrete [1]. Concrete production contributes significantly to the
climate catastrophe by emitting massive volumes of carbon dioxide into the atmosphere. Carbon dioxide and methane are
the two greenhouse gases mainly responsible for global warming. Furthermore, concrete can harm the earth's most fertile
layer, the topsoil [2–3].
Prestressing is an important stage in reinforcement since it reduces the quantity of concrete required, lowering material
usage and transportation costs while increasing durability and service life [4]. A prestressing system is used to take
advantage of its compressive strength. Prestressing compresses the concrete structure in places where load causes tensile
stress. Before cracking in the concrete develops, the tension created by applied loads must first relieve the compression
caused by the prestressing, which is performed with tendons [5]. When reinforced with steel, concrete can bear significant
tensile loads while also exhibiting excellent ductility, allowing structural elements to fail in a ductile manner with
extensive plastic deformation when overloaded. This type of failure is known as a flexural failure. With the capacity to
provide sufficient warning before failure, this type of failure is commonly used in the design of a structural element. When
specific design standards are followed, this behavioral response can be correctly predicted. The four-point static bending
test is used to investigate the properties and behavior of materials with structural applications, thus establishing design
specifications.
Numerical modeling has made considerable progress in this area. Furthermore, numerical modeling is being
implemented in design because experiments can be challenging, costly, and time-consuming [6]. In addition, model
predictions closely align with experimental testing. As for modeling a pre-tensioned prestressed concrete beam, the most
common user-friendly software packages for the nonlinear analysis of reinforced-concrete structures are ATENA 3D,
ANSYS, and ABAQUS. In particular, ATENA 3D, based on advanced constitutive models, can effectively support and
extend experimental investigations for innovative solutions in the field of prestressed concrete members. The crack-band
approach employed for tensile and compressive softening avoids the finite element (FE) mesh sensitivity of the solution.
Tavares et al. [7] studied bond behavior for different bar diameters with simulated pull-out tests using ATENA
software. Their numerical results were in the range of experimental results. Furthermore, Yapar et al. [8] presented a
nonlinear FE model for pre-tensioned prestressed concrete beams using ABAQUS software. Up to the collapse load, the

192-1
modeling and simulation results for the test problem were quite close to the test results. The simulation results
provided a good explanation of how such beams behave.
Similarly, Hoang et al. [9] developed a FE model in ATENA 3D to examine the compressive behavior of circular
steel tube–confined concrete stub columns using various concrete compressive strengths, including normal, high, and
ultra-high strengths. The FE model predicted curves of ultimate and axial loads versus vertical strain that aligned well
with those of previous test results. Additionally, Abed et al. [10] simulated pull-out tests using ATENA 3D software.
The numerical results were comparable to those obtained experimentally, which expressed real behavior.
Daluga [11] studied the influence of the maximum aggregate size on the shear strength of reinforced concrete
beams without web reinforcement. A 20 percent increase in mean shear strength was observed between beams with a
maximum aggregate size of 19 mm (3/8 inch) and 25.4 mm (1 inch). Ismael et al. [12] evaluated eight reinforced
concrete beams with self-compacting concrete and conventional concrete using different aggregate sizes (10 mm and
20 mm). The results showed that increasing the coarse aggregate maximum size increases the diagonal cracking load
and ultimate shear strength of self-compacting concrete beams slightly, but the effect was more significant for
conventional concrete beams.
As discussed, several experiments have been simulated using various software where the numerical results align
with those of experimental tests. Although several reinforced-concrete member models have been simulated, few
numerical models are available in the literature on the flexural behavior of the pre-tensioned prestressed members.
Therefore, more research is needed to numerically model the flexural behavior of a pre-tensioned prestressed concrete
beam accurately and understand the important parameters involved.

2. Methodology
Based on an experimental investigation from the literature review, this study consists of numerical modeling
efforts targeted at understanding and validating the structural behavior of a pre-tensioned prestressed concrete
member. In addition, studying the influence of the coarse aggregate size on the behavior of the member under the
bending test. A four-point static bending test is used to perform the flexural test on the pre-tensioned prestressed
concrete beam. Cowen and VanHorn [13] carried out an experimental study on a simply supported T-beam. As shown
in Figure 5, the specimen was a rectangular prestressed concrete beam with a cast-in-place slab. In this study, the
ATENA 3D software was used to simulate the actual behavior of concrete structures, such as reinforcement yielding,
concrete cracking, and crushing [14]. Furthermore, ATENA 3D was specially designed for concrete, making it more
user-friendly with appropriate default values.

Fig. 1: Four-point static bending test.

The concrete mix was produced with Type III (high-early strength) Portland cement, sand, and crushed limestone
coarse aggregate (3/4 inch, or 19 mm, maximum). The proportions of the mix by weight (cement-to-sand-to-coarse
aggregate) were 1.00:2.64:2.98. The cylindric concrete compressive strengths at the time of the test were 41.16 MPa
and 46.56 MPa for the rectangular beam and slab, respectively. In addition, Young's modulus of the concrete was
28 GPa.

192-2
 In this study, stress-relieved type seven-wire 270 K strands with a diameter of 12.8 mm were used as longitudinal
reinforcements. The elastic modulus of the strand is 200.595 GPa, as determined by Cowen and VanHorn [13]. These
strands were located in three layers within the cross-section, as shown in Fig. 2. The rebars for the shear reinforcements
were fabricated from No. 3 deformed bars with a nominal yield stress of 345 MPa and a diameter of 10 mm.

2.1. Four-point static bending test

In this section, the laboratory experiment of the four-point static bending test related to Cowen and VanHorn [13] is
presented and validated using the nonlinear FE software ATENA 3D. The test was carried out to determine to expected
ultimate load and the associate failure characteristics.
The pre-tensioned prestressed concrete beam was tested in the Fritz Engineering Laboratory's 300 kips (1334.5 kN)
hydraulic testing machine [13]. As illustrated in Fig. 1, the loading beam and associated apparatus were arranged to
provide symmetric two-point loading for the spacemen. The pre-tensioned prestressed concrete beam used in this study had
a span of 2743.2 mm. Figure 2 shows the geometry, beam cross-section, and position of the loading spins. The test
specimen was initially loaded in increments of 44.5 kN, which approximated to around 8% of the computed ultimate load.
When cracking of the concrete became visible, the loading increment was reduced to 22.25 kN until failure occurred. The
strands were initially tensioned to a stress level of 70% of the specified ultimate stress [13]. Table 1 presents the theoretical
and measured losses in the strand.

Fig. 2: Schematic and cross-section of the pre-tensioned prestressed concrete beam.

Table 1: Losses in prestressing strand.

Elastic Stress Percent Creep, Total The

Initial Total Effective
loss at after loss after Shrinkage steel and loss finite
stress loss stress
release release release (kN) concrete present element
(kN) (kN) (kN)
(kN) (kN) % (kN) % (FE)
model
Theoretical 128.55 6.81 121.88 5.19 5.60 22.33 34.74 27.02 93.9
was
Measured 128.55 15.03 113.52 11.70 19.88 34.91 27.20 93.64 upgrade
d to
study the influence of the coarse aggregate size on the behavior of the pre-tensioned prestressed concrete beam under the
bending test with maximum aggregate sizes equal to 13, 21, 25, and 37 mm, as carried out by Ajamu et al. [15]. The
upgraded model is called "n-beam". Since the aggregate size affects the shear strength, the shear reinforcement ratio of the
beam was decreased to half to fail in shear and not in flexure or bond. The arrangement of shear reinforcement of the n-
beam was Φ10@304.8 mm, as shown in Fig. 3.

192-3
 Fig. 3: The schematic view of the FEM model with the reduced amount of shear reinforcement "n-beam".

After establishing the model geometry and the flexural bond test procedures used in the experiment, an inverse
analysis was executed by ATENA 3D. To simplify the calculations, half of the pre-tensioned prestressed concrete
beam was modeled, owing to symmetricity. Steel plates 3 cm thick were placed at the bottom of the beam for support
and at the top of the beam for loading. To avoid early cracking, the displacement stresses were applied at the upper
steel plate rather than directly on the beam, as indicated by the software manual [16]. An elastic tetrahedral element
with an elastic modulus of 2.1 × 105 MPa was used to model the steel plates. For the boundary conditions, since
analyzing only half of the beam, the axis of symmetry along the right side of the beam was enforced. Therefore, the
horizontal y-displacements along this side should be equal to zero. The transition of the left support was fixed in the
vertical and lateral directions, as illustrated in Fig. 4. The FE mesh type for the model was brick with a size of 0.05 m.
The loading history for the analysis was defined, consisting of load steps, and each load step included a combination
of load cases. In the software, the Newton-Raphson solution method was chosen. The simulation's aim was to increase
the load to failure. Forces, displacements, strains, and stresses in the model are typically monitored during nonlinear
analysis. The monitored data can provide information about the model's status and produce the load-deflection curve.
Therefore, two monitoring points were selected in the numerical model for this purpose (see Fig. 4). The first point
was for the monitoring of the force applied at the top steel plate. The second point was located at the middle of the
beam near its bottom surface, where the largest vertical displacements could be expected. Thus, the required load-
deflection curve could be obtained. In addition, other monitor points were defined to record the stresses, strains, and
crack widths of the member.
When defining the materials, the compressive strength of concrete was entered into the program for the slab
(flange) and the beam (web). The strand properties were defined according to the experiments, as shown in Fig. 5. In
addition, prestressing forces were applied to the strands in the software in the form of an effective stress of 93.64 MPa.

Fig. 4: Program displays of the four-point bending model and the location of the monitor points.

192-4
 Fig. 5: Stress-strain curve for the seven-wire strand used in ATENA 3D.

2.2. Nonlinear Finite Element Analysis Using ATENA 3D

Fig. 6 summarizes the compression, tension, and shear transfer models implemented in the ATENA 3D software
package. In the compression model, the compression hardening function is defined using an elliptical relation through the
following Equation 1 [17]–[18]. Beyond the peak, the softening relation is determined through an analogical model
commonly known as the crush band model (to ensure mesh objectivity). The assumption was made on the localization of
post-peak compressive stress and energy dissipation in a plane normal to the direction of principal stress. A 𝑤d value of
0.5 mm was used in this study, following the recommended value for normal concrete [19]. In the tension model, the
exponential softening law was adopted following the formulation of a fictitious crack model based on the crack-opening
law and fracture energy, as expressed in Equation 2 [17].

√( ) (1)

where is the normal compressive stress (MPa), is the mean cylinder compressive strength (MPa), is the
compressive stress at the post-elastic point (MPa), is the concrete strain (mm/mm), and is the plastic strain at the
peak point (mm/mm).
𝑤 𝑤 𝑤
( ( ) ) ( ) (2)
𝑤 𝑤 𝑤

where is the normal tensile stress (MPa), is the tensile cracking strength (MPa), 𝑤 is the crack opening (mm),
and 𝑤 is the crack opening at the complete release of stress deriving from fracture energy (mm). Empirical constants =3
and =6.93 are considered.

Fig. 6: Compression, tension, and transfer implemented models in the ATENA 3D software package.

ATENA 3D software uses the Modified Compression Field Theory of Vecchio and Collins to calculate the shear
strength of cracked concrete, as shown in Eqs (3) [20]. Cracking occurs for most concretes along the interface between the

192-5
aggregate and paste of cement. As illustrated in Fig. 7, shear stresses can transmit with the resulting rough cracks by
aggregate interlock.
√
𝑤 (3)

where is the shear strength of a cracked concrete, is the compressive strength of the concrete (MPa), 𝑤 is
the maximum crack width (mm), and is the maximum aggregate size (mm).

Fig. 7: Transmitting shear stresses across crack by aggregate interlock [20].

3. Results
In this section, the load-deflection curve of the beam (Φ10@152.4 mm) obtained from ATENA 3D is compared
with the experimental results. Fig. 8 illustrates the relationship between the load and deflection of the beam obtained
from the laboratory experiment and the numerical modeling. After the yielding of the strands, the beam reached
failure, as confirmed in Fig. 9. In other words, the failure type is flexural such that the pre-tensioned prestressed
concrete beam reached failure by the crushing of the concrete long after the yielding of the strands in the beam's mid-
span. The ultimate loads were 552.69 kN and 552.80 kN, and the corresponding mid-span deflections were 32.93 cm
and 33.52 cm for the laboratory experiment and FEM model, respectively.

Fig. 8: Experimental and FE results of the four-point static bending test.

192-6
 Fig. 9: Stresses along the strands of the beam at failure.

Fig. 10 presents the crack pattern of the beam obtained from the FE analysis and laboratory experiment. Flexural
cracking occurred in the high-moment region of the test specimens when the stresses in the bottom fibers reached values
normally associated with the tensile strength of the concrete. The development of vertical cracks to a height between the
lower and upper strands characterized the flexural cracking. Flexural-shear cracking followed flexural cracking in the
shear-span region of the test specimen. A flexural-shear crack is different from a diagonal crack in that a diagonal crack is
caused by primary tensile stresses that develop in the web of an I-beam or the stem of a T-beam.
The nonlinear FE analyses showed strong agreements on the overall load-deflections and failure modes of the beams
tested, validating the experimental results.

Fig. 10: Comparison of the crack pattern of the beam with the laboratory experiment.

After assessing the bending behavior of the beam and verifying the results with the laboratory experiment, the n-beam,
having reduced shear reinforcement (Φ10@304.8 mm), was modeled with maximum aggregate sizes of 13, 21, 25, and 37
mm. Four models failed in the same mode with different ultimate load capacities. The failures were unpredictable,
occurring in the mid–shear span by yielding the two bottom layers and crushing the concrete around the top unyielded
strand and specifically between two stirrups, as shown in Fig. 11. The results of FE models indicated that the ultimate load
capacity of the beam increased with increasing the maximum aggregate size. Similar results were found in the previous
studies by Murray et al. [21] and Chana [22]. Fig. 12 illustrates the influence of the maximum aggregate size on the
ultimate load of the n-beam. The ultimate loads at failure were 516.6 kN, 517.2 kN, 519.8 kN, and 520.4 kN for maximum
aggregate sizes of 13, 21, 25, and 37 mm, respectively. A linear regression shows the relationship between the ultimate
load and the maximum aggregate size with a coefficient of determination R2= 0.81. Although the relationship is positive,
the increase in the ultimate loads due to a larger maximum aggregate size was slight, as found by Ismael et al. [12]. This is
attributed to the failure mode, which was not shear, owing to the minimal shear reinforcement.

Fig. 11: Crack pattern and crack width of "n-beam" (Φ10@304.8 mm) at failure.

192-7
 Fig. 12: Relationship between the coarse aggregate size and the maximum load at the failure.

4. Conclusion
The four-point static bending test of a prestressed T-beam with three layers of strands was simulated with a
nonlinear FE analysis using ATENA 3D in this study, and the results were compared with existing experimental
results to validate the model. The specimen was a rectangular prestressed concrete beam with a cast-in-place slab with
cylindric concrete compressive strengths of 41.16 MPa and 46.56 MPa at the test, respectively. In this study, the
reinforcing steel was seven-wire strands as a longitudinal reinforcement and deformed bars as a transverse
reinforcement. The arrangement of shear reinforcement was Φ10@152.4 mm. Subsequently, the FE model was
upgraded to study the influence of the maximum coarse aggregate size on the behavior of a pre-tensioned prestressed
concrete beam under the bending test with maximum aggregate sizes of 13, 21, 25, and 37 mm. The arrangements of
shear reinforcements of these beams were Φ10@304.8 mm. After examining the analyses, the following points are put
forward:
• The failure mode of the beam (10@152.4 mm) exhibited an expected mode of failure, which was flexure
characterized by strand yielding and eventual concrete crushing. This behavior is desired by structural designers
because it provides ample notice before failure occurs.
• The experimental specimen and FE model results were nearly identical in terms of failure mode, crack pattern,
ultimate load, and corresponding mid-span deflection, with a variation of less than 2%, which ATENA 3D software
express the real behavior.
• The ultimate load of the beam with the reduced amount of shear reinforcement (Φ10@304.8 mm) at failure was
slightly influenced by the maximum aggregate size. The increase of the ultimate load was not significant, which is
mostly due to the failure mode, which was shear, owing to the minimal shear reinforcement.

References
[1] E. Association, Syrian Arab code for the design and implementation of reinforced concrete structures. Damascus,
Syrian Engineers Union, 2012.
[2] R. Rehan, M. Nehdi, "Carbon dioxide emissions and climate change: policy implications for the cement industry,"
Environmental Science & Policy, vol. 8, no. 2, pp. 105-114, 2005.
[3] Z. Alkurdi, "Influence of Concrete Compressive Strength on Transfer Length in Pretensioned Concrete Members
Using 3D Nonlinear FEM Analysis," 6th International Conference On Civil Structural and Transportation Engineering
(ICCSTE'21), Virtually, 2021.
[4] D. P. Billington, "Historical perspective on prestressed concrete," PCI Journal, vol. 21, no. 5, pp. 48-71, 1976.
[5] M. L. Arnold, "Un-tensioned pullout tests to predict the bond quality of different prestressing reinforcements used in
concrete railroad ties," Doctoral dissertation, Kansas State University, Manhattan, 2013.
[6] M. A. Abed, Z. Alkurdi, J. Fořt, R. Černý, S. Solyom, "Bond Behavior of FRP Bars in Lightweight SCC under Direct
Pull-Out Conditions: Experimental and Numerical Investigation," Materials, vol. 15, no. 10: 3555, 2022.

192-8
[7] A. J. Tavares, M. P. Barbosa, T. N. Bittencourt, M. Lorrain, “Aderência aço-concreto: simulação numéricadosensaios
de arran-chamento pull-out e APULOT usando o programa ATENA (Bond steel-concrete: simulation analysis of the
pull-out tests and APULOT using the program ATENA),” Revista IBRACON de Estruturas e Materiais, vol. 7, no. 1,
pp. 138–157, 2014.
[8] O. Yapar, P. K. Basu, N. Nordendale, "Accurate finite element modeling of pretensioned prestressed concrete beams,"
Engineering Structures, vol. 101, pp. 163-178, 2015.
[9] A. Le Hoang, E. Fehling, "Numerical study of circular steel tube confined concrete (STCC) stub columns," Journal of
Constructional Steel Research, vol. 136, pp. 238-255, 2017.
[10] M. A. Abed, Z. Alkurdi, A. Kheshfeh, T. Kovács, and S. Nehme, "Numerical Evaluation of Bond Behavior of Ribbed
Steel Bars or Seven-wire Strands Embedded in Lightweight Concrete," Period. Polytech. Civil Eng., vol. 65, no. 2, pp.
385-396, 2021.
[11] D. R. Daluga, "The effect of maximum aggregate size on the shear strength of geometrically scaled reinforced
concrete beams," Doctoral dissertation, Purdue University, West Lafayette, 2015.
[12] M. A. Ismael, H. J. Abd, A. A. Hameed, "Effect of coarse aggregate size on shear behavior of self-compacting
concrete and conventional concrete beams," International Journal of Engineering & Technology, vol. 7, no. 4.20, pp.
359-363, 2018.
[13] S. H. Cowen, D. A. Van Horn, "End support effects on ultimate flexural bond in pre-tensioned beams," Doctoral
dissertation, Lehigh University, Bethlehem, 1967.
[14] V. Červenka, L. Jendele, J. Cervenka, Atena theory. Prague , 2007.
[15] S. O. Ajamu, J. A. Ige, "Effect of coarse aggregate size on the compressive strength and the flexural strength of
concrete beam," Journal of Engineering Research and Applications, vol. 5, no. 4, pp. 2248-2267, 2015.
[16] J. Červenka, Z. Procházková, T. Sajdlová, ATENA Program Documentation, Part 4-2, Tutorial for Program ATENA
3D. Cervenka Consulting, Prague, Czech Republic, 2017.
[17] V. Červenka, L. Jendele, J. Červenka, Program documentation ATENA theory. Czech Republic, Červenka
Consulting, 2018.
[18] P. Menetrey, K. J. Willam, "Triaxial failure criterion for concrete and its generalization," Structural Journal, vol. 92,
no. 3, pp. 311-318.
[19] J. Červenka, V. Červenka, S. Laserna, "On crack band model in finite element analysis of concrete fracture in
engineering practice," Engineering Fracture Mechanics, vol. 197, pp. 27-47, 2018.
[20] F.J. Vecchio, M. P. Collins, "The modified compression-field theory for reinforced concrete elements subjected to
shear," ACI J., vol. 83, no. 2, pp. 219-231, 1986.
[21] M. R. Murray, "An investigation of the unit shear strength of geometrically scaled reinforced concrete beams," 2010.
[22] P. S. Chana, Some aspects of modelling the behaviour of reinforced concrete under shear loading (No. Monograph).
Cement and Concrete Association, 1981.

192-9