You are on page 1of 11

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/285833414

Compression-Tension Strength of Reinforced and Fiber-Reinforced Concrete

Article  in  Aci Structural Journal · May 2011

CITATIONS READS
6 245

3 authors, including:

Ekkehard Fehling Prof. Dr.-Ing Friedrich‐Karl Röder


Universität Kassel Universität Kassel
78 PUBLICATIONS   572 CITATIONS    13 PUBLICATIONS   35 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Druck-Zug-Festigkeit von Stahlbeton View project

Kippstabilität View project

All content following this page was uploaded by Friedrich‐Karl Röder on 02 January 2019.

The user has requested enhancement of the downloaded file.


ACI STRUCTURAL JOURNAL TECHNICAL PAPER
Title no. 108-S34

Compression-Tension Strength of Reinforced


and Fiber-Reinforced Concrete
by Ekkehard Fehling, Torsten Leutbecher, and Friedrich-Karl Roeder

The compressive strength of concrete can substantially decrease in influence of transverse tension and cracking on the
relation to the uniaxial compressive strength by transverse tension compressive strain at peak load and on the compressive
and cracking. This holds true for plain and reinforced concrete. stiffness. Based on the authors’ own experimental studies9
The question of the biaxial compression-tension strength of reinforced and a critical review and re-evaluation of similar tests done
concrete has been examined over the past 40 years by numerous
by other researchers, a design proposal has been developed
scientists. Their results, however, vary considerably and lead to
contradictory conclusions. Accordingly, in national standards, that enables the determination of the compressive strength
very different calculation rules can be found on this subject, reduction of RC with and without additional steel fibers as a
whereby the provided reductions differ up to a factor of 2 for the function of the applied transverse tensile strain.
same application. Based on the authors’ own experimental
investigations and a critical review and classification of former test EXPERIMENTAL PROGRAM
series that are regarded as trendsetting, a proposal for the Within four test series, a total of 56 panel-shaped specimens
reduction of the compressive strength of cracked reinforced were tested (Table 1). Of these specimens, seven panels were
concrete is developed. For the first time, the influence of fibers in unreinforced (plain concrete panels), three panels were
addition to bar reinforcement is also considered. reinforced with steel fibers only (fiber-reinforced concrete
[FRC] panels), and 23 panels were reinforced with reinforcing
Keywords: biaxial test; compression; concrete panel; fiber-reinforced
concrete; material model; reinforced concrete; tension.
bars only (RC panels). An additional 23 test specimens were
provided with a mixed reinforcement of reinforcing bars and
INTRODUCTION steel fibers (mixed reinforced concrete [MRC] panels).
The realistic computational modeling and design of All plain concrete and FRC panels (a total of 10 specimens) as
reinforced concrete (RC) structures has to consider multi- well as some of the RC and MRC panels (a total of 13 specimens)
axial stress and strain states. Special attention has to be paid were loaded uniaxially in compression only (reference panels).
to compression stress states with simultaneous transverse Within these tests, the influences of the type of test specimen
tensile strain. In particular, panel and shell structures are (cylinders versus panels) and the local stress concentrations
frequently subject to such biaxial stress states. With regard caused by the reinforcing bars (called the “detrimental effect” in
to the design and analysis of strut-and-tie models as well as the following) were determined. The additional 33 panels were
to the shear design of beams, transverse tension and cracking loaded biaxially in tension and compression. The applied tensile
also have a substantial influence on the strength of the strain values varied between 0.0005 and 0.030.
compression struts. The test specimens had a length of 1000 mm (39.4 in.) in
Over the past 40 years, tests on the bearing capacity of the horizontal (tensile) direction, a height of 500 mm (19.7 in.)
cracked RC panels (without fibers) have been carried out by in the vertical (compression) direction, and a thickness of
numerous scientists. Because of partially different objectives 100 mm (3.94 in.) (Fig. 1). The RC and MRC panels were
and due to the highly differing test executions (test setup, orthogonally reinforced with two reinforcement layers of
load application, and so on), dimensions of the test specimens, BSt 500 reinforcing bars with a yield stress of 500 N/mm2
and reinforcement configurations, the test results vary (72 ksi) and a ds of 10 mm (0.394 in.) in each loading direction.
considerably. Vecchio and Collins1 obtained a decrease of The spacing of the horizontal (tensile) reinforcing bars was
the compressive strength after cracking to a minimum value 100 mm (3.94 in.) (reinforcement ratio ρs = 1.57%) and
of 20% of the uniaxial compressive strength at high transverse 225 mm (8.87 in.) for the vertical (compression) reinforcing
strains. Belarbi and Hsu2 also observed very large decreases bars (reinforcement ratio ρs = 0.79%). The concrete cover
to a minimum value of 25% of the uniaxial compressive was 15 mm (0.591 in.) regarding the vertical reinforcement
strength. In contrast, a maximum decrease to between 80 and and 25 mm (0.985 in.) regarding the horizontal reinforcement.
85% of the uniaxial compressive strength was noticed by In the tensile direction, along with the continuous straight
Schlaich and Schaefer,3 Kollegger and Mehlhorn,4 and Eibl reinforcing bars, a U-bent reinforcement (refer to Item 2 in
and Neuroth.5 Based on the different test results, the pertinent Fig. 1) was placed in a 120 mm (4.73 in.) long load transmission
rules in the current design codes6-8 differ significantly. zone to avoid yielding of the continuous reinforcement
outside the specimen.
RESEARCH SIGNIFICANCE
The objective of this paper is to clarify the existing
inconsistencies regarding the proposed strength reductions ACI Structural Journal, V. 108, No. 3, May-June 2011.
MS No. S-2010-135 received April 29, 2010, and reviewed under Institute
for RC due to biaxial compression-tension loading and publication policies. Copyright © 2011, American Concrete Institute. All rights reserved,
moreover to determine if the behavior can be favorably including the making of copies unless permission is obtained from the copyright proprietors.
Pertinent discussion including author’s closure, if any, will be published in the March-
influenced by fiber addition. Attention was also paid to the April 2012 ACI Structural Journal if the discussion is received by November 1, 2011.

350 ACI Structural Journal/May-June 2011


The reference cylinders (d/h = 150 mm/300 mm [5.91 in./
ACI member Ekkehard Fehling is a Professor of civil engineering at the Institute of
Structural Engineering in the Department of Civil Engineering at the University of 11.8 in.]) were fabricated to determine the modulus of elas-
Kassel, Kassel, Germany. He received his diploma in civil engineering and his ticity, the splitting tensile strength, and the compressive
doctorate from the Technical University of Darmstadt, Darmstadt, Germany, in 1983 strength. They were stored under the same conditions and
and 1990, respectively. His research interests include nonlinear modeling of reinforced
concrete structures in SLS and ULS, the safety of masonry structures under earth- tested on the same day as the panels. The influence of
quake loading, and the structural behavior of ultra high-performance concrete. shrinkage can be considered as marginal. Details regarding
Torsten Leutbecher is a Research Associate and Lecturer at the Institute of Structural
the casting procedure, concrete age at testing day, and
Engineering in the Department of Civil Engineering at the University of Kassel. He storage conditions can be found in the detailed test report.9
received two diplomas in civil engineering from the University of Applied Sciences,
Wuerzburg, Germany, and the University of Kassel in 1999 and 2001, respectively,
and his doctorate from the University of Kassel in 2007. His research interests include TEST SETUP AND EXECUTION
the structural behavior of ultra high-performance concrete, the biaxial compression- The compression-tension tests were conducted on the
tension strength of concrete, and reinforced concrete in the serviceability state. strong floor of the Institute of Structural Engineering at the
Friedrich-Karl Roeder is a former Senior Research Associate at the Institute of University of Kassel. The test setup is shown in Fig. 2.
Structural Engineering in the Department of Civil Engineering at the University of The horizontal tension forces were applied by two
Kassel. He received his diploma in civil engineering and his doctorate from the hydraulic jacks (No. 1 in Fig. 2), each with a nominal
Technical University of Darmstadt in 1970 and 1982, respectively. His research interests
include nonlinear modeling of reinforced concrete structures, especially lateral capacity of 400 kN (90 kips), and transmitted by a crosshead
buckling of slender reinforced and prestressed concrete girders. (No. 2 in Fig. 2) and via threaded bars and sleeves to the
reinforcement of the test specimen.
For the compression loading, two hydraulic jacks (No. 3 in
The maximum grain size of the concrete was 8 mm (0.31 in.). Fig. 2), each with a nominal capacity of 2.5 MN (563 kips),
For the FRC and MRC panels, the fiber content amounted to were provided. The compression forces were transmitted by
0.5 volume percent (40 kg steel fibers per cubic meter of two roller bearings (No. 4 in Fig. 2), which permitted a tilting
concrete [67.4 lb/yd3]), which is a very common fiber content in the plane of the panel, to two load distributing steel panels
in practice. In Series 1, normal-strength steel fibers with (No. 5 in Fig. 2) on top of each other into the specimen (No. 6
hooked ends (λ = lf /df = 50 mm/1.05 mm = 48 [lf /df = in Fig. 2). The steel panels were prevented from moving out of
1.97 in./0.041 in.]) with a tensile strength of 1000 N/mm2 the loading plane by horizontal struts. A block of ultra-high-
(144 ksi) were used. In Series 2, 3, and 4, high-strength steel performance concrete (UHPC) (No. 7 in Fig. 2) with an
fibers with hooked ends (λ = lf /df = 40 mm/0.50 mm = attached polished steel plate acted as support for the specimens.
80[lf /df = 1.58 in./0.020 in.]) with a tensile strength of The specimens were loaded under displacement control in
2300 N/mm2 (331 ksi) were used. both directions. The two compression jacks operated

Table 1—Experimental program


Concrete panels
Bar reinforced concrete panels (RC) with mixed reinforcement (MRC)
Loading Number of plain Number of fiber-reinforced
Series mode concrete panels concrete panels (FRC) Number Applied tensile strain Number Applied tensile strain
Uniaxial 1 1 1 — 1 —
1
Biaxial — — 6 0.0017 to 0.030 7 0.0017 to 0.025
Uniaxial 4 1 5 — 2 —
2
Biaxial — — 5 0.0005 to 0.027 3 0.0005 to 0.029
Uniaxial — — — — 2 —
3
Biaxial — — — — 4 0.0006 to 0.013
Uniaxial 2 1 1 — 1 —
4
Biaxial — — 5 0.0007 to 0.025 3 0.0016 to 0.0045

Fig. 1—Dimensions and reinforcement of test specimens.

ACI Structural Journal/May-June 2011 351


Fig. 2—Test setup for biaxial compression-tension tests.

Fig. 3—Crack patterns and failure of panels.

separately with the target that the displacements measured at applied after applying the appropriate tensile strain. Two
both ends of the concrete panel were equal. panels were loaded simultaneously at a strain rate of 1 με/s
The compressive and tensile forces were measured by the on the tension side and 4 με/s on the compression side
load cells of the four hydraulic jacks. Additionally, the (proportional load application). All other specimens were
horizontal tension forces were checked by two further load loaded sequentially. Thereby, the transverse load was
cells on the side opposite to the load application (abutment applied first deformation controlled. During the loading, the
side). To ensure the uniformity of the load application across cracks on the panel surface were marked. The loading rate
the height of the panel, the tension forces of all continuous was 2 με/s until the first crack, between 4 and 8 με/s until it
reinforcing bars were monitored by load cells on both the reached the yield stress of the reinforcement, and finally 50 με/s
side of the hydraulic jacks and the abutment side. up to the target tensile strain. Afterward, the compression
To measure displacements in the vertical and horizontal load was applied continuously at a rate of 4 με/s until failure.
directions, three inductive displacement transducers were At this point, the horizontal restraint between the specimen
placed in each direction at the front and back of the panel. The and the top and bottom supports was not prevented.
gauge lengths were 400 mm (15.8 in.) in the vertical direction
and 650 mm (25.6 in.) in the horizontal direction (refer to TEST RESULTS
Fig. 3(c)) because of the embedded loops at the ends. During the test series, the experimental setup, the test
The biaxially tested specimens were placed in the testing execution, and the concrete mixture had to be modified to mini-
device such that the compression load could be centrically mize and/or eliminate the influence of identified disturbances.

352 ACI Structural Journal/May-June 2011


Table 2—Test results classified to be of special significance and to be representative
Panels Cylinders
Nominal Concrete Relative
Transverse compressive compressive compres-
Fiber Applied strain at Compressive strength strength Modulus of Compressive Modulus of sive
Panel content, tensile peak load Number strain at peak σ2,max , σc2,max , elasticity Ec, strength fc,cyl , elasticity Ec, strength
(Test series) vol. % strain ε1 ε1,max of cracks load ε2,max N/mm2 N/mm2 N/mm2 N/mm2 N/mm2 σc2/fc,cyl
Plain concrete panel
VK6 (4) — — 0.00026 — –0.00284 –37.8 –37.8 35,300 –38.8 34,700 0.97
Fiber-reinforced concrete panel (FRC panel)
FVK2 (4) 0.5 — 0.00041 — –0.00313 –41.7 –41.7 33,700 –41.4 32,400 1.01
Reinforced concrete panels (RC panels)
000VK4 (4) — — 0.00033 — –0.00297 –43.7 –39.1 33,800 –44.7 33,900 0.88
005VK3 (4) — 0.00065 0.00076 5 –0.00243 –37.0 –33.2 29,500 –42.5 32,600 0.78
020VK2 (4) — 0.00200 0.00229 11 –0.00275 –36.0 –31.7 29,400 –42.3 33,200 0.75
050VK2 (4) — 0.00505 0.00501 12 –0.00189 –28.8 –25.8 27,500 –43.5 32,400 0.59
250VK2 (4) — 0.02511 0.02549 12 –0.00238 –28.6 –24.8 26,500 –46.4 34,500 0.53
Concrete panels with mixed reinforcement (MRC panels)
F000VK2 (3) 0.5 — 0.00022 — –0.00230 –43.3 –39.7 32,800 –39.7 29,400 1.00
F020VK2 (3) 0.5 0.00211 0.00226 13 –0.00303 –48.1 –43.4 30,000 –51.4 33,700 0.84
F025VK2 (4) 0.5 0.00448 0.00461 12 –0.00286 –36.2 –31.8 29,000 –44.0 33,300 0.72
F050VK2 (3) 0.5 0.00547 0.00550 13 –0.00269 –45.9 –41.7 26,800 –57.1 33,000 0.73
F250VK2 (3) 0.5 0.01304 0.01332 14 –0.00301 –45.4 –40.7 25,900 –55.5 34,200 0.73
Note: 1 N/mm2 = 0.1438 ksi.

In Series 1, a polytetrafluoroethylene (PTFE) foil used for was computationally determined on the basis of the panel’s
load introduction led to an overproportional decrease of the displacements at small loads. For the calculation, the net
compressive strength. In Series 2, and for the fiber-reinforced concrete compressive stress σc2 was used.
specimens in Series 4, a segregation of the concrete was
detected (settlement of coarse aggregates and steel fibers). Uniaxially loaded panels
These tests are not discussed in this paper. In the following, Panels VK6 (plain concrete panel) and FVK2 (FRC panel)
only tests of Series 3 and 4 (12 panels from a total of 56) are failed in shear compression (Fig. 3(a)). With nondimen-
presented, which were considered to be of special significance sional concrete compressive strengths σc2/fc,cyl = σ2/fc,cyl of
after careful and critical evaluation by the authors (Table 2). 0.97 and 1.01, they reached approximately the strength of the
Please refer to the detailed test report9 for the evaluation companion uniaxial cylinder results. The moduli of elasticity
and discussion of all test results. also differ only within the scope of the measuring accuracy.
The strains in the tensile and compressive directions, ε1 Thus, an influence of the different specimen geometries
and ε2, respectively, were obtained from the average (cylinders and panels) on the compressive strengths, as
displacement values at the front and back of the panel. achieved in further tests, can be widely excluded.
The net concrete compressive stress σc2 of the panel was The uniaxially tested Panel 000VK4 (RC panel)
determined from the measured forces of the hydraulic jacks, achieved a ratio σ2/fc,cyl of 0.98 and a ratio σc2/fc,cyl of
subtracting the load theoretically carried by the reinforcement 0.88. This means that the load-bearing capacity of an RC
panel cannot be simply gathered as the sum of the load-
bearing capacity of the concrete determined from the
σc2 = σ2 – σs2(ε2) · ρ2 (1)
uniaxial compressive strength and the load carried by the
reinforcement theoretically. Rather, even under uniaxial
where σ2 is the nominal compressive stress determined from loading, values of σc2/fc,cyl less than 1 can be achieved.
the forces of the hydraulic jacks referred to the gross cross- Depending on the reinforcement configuration (for
sectional area Ac = 1000 x 100 mm2 (39.4 x 3.94 in.2); example, hooks, lap splices, and transverse
σs2(ε2) is the stress of the reinforcing bars arranged in the reinforcement), the difference of stiffness between the
compressive direction determined as a function of the concrete structure and the reinforcement embedded
measured compressive strain ε2 (mean value from six therein causes local stress concentrations, which can
sensors); and ρ2 is the geometrical reinforcement ratio reduce the load-bearing capacity. Premature damage may
(reinforcing bars) in the compressive direction. This procedure also result from the shrinkage shortening of concrete that
was also used for the calculation of the maximum concrete is restrained by the reinforcement.
compressive stress σc2,max. Based on the result of Panel 000VK4, the reduction of the
Both the transverse strain of the test specimens and the load-carrying capacity of RC panels due to the detrimental
applied tensile load changed during the compressive loading, effect of the reinforcement can be estimated to be approx-
but only marginally. Along with the applied tensile strain ε1, imately 10%. On the other hand, MRC Panel F000VK2
the transverse strain at peak load ε1,max is shown in Table 2. reached the cylinder compressive strength (σc2/fc,cyl = 1.00).
To give an indication of the degradation of the compressive Thus, the crack-bridging effect of the fibers is obviously able
stiffness due to cracking, the tangent modulus of elasticity to delay an early failure caused by local stress concentrations.

ACI Structural Journal/May-June 2011 353


capacity, however, is not sufficient to increase the width of
the other smaller cracks because in these cracks, the tensile
force is still carried by the combined action of the reinforcing
bars and fibers. In this respect, the variation of strain values
ε1,max specified in Table 2 only depends on the actual crack
width of the crack with yielding reinforcement for either
panel. This had no considerable influence, however, on the
compressive load-bearing capacity of the test specimens.
The failure of the biaxially loaded panels usually occurred
suddenly in shear compression (Fig. 3(b)) or by vertical
splitting of the cross section in the planes of the longitudinal
reinforcing bars (Fig. 3(d)). Buckling of the vertical
(compressive) reinforcement, connected with spalling of the
concrete cover, was also observed (Fig. 3(e)).
The decrease of the maximum concrete compressive stress
of the biaxially tested panels was accompanied by a reduction
of the compressive stiffness. Compared to the moduli of
elasticity of the cylinders and the uniaxially tested reference
panels, this reduction does not exceed approximately 20 to
25%, independent of the addition of fibers. Hypothetically,
the decrease of stiffness can be explained by the loss of parts
of the cross-sectional area due to cracking (irregularly bound
Fig. 4—Reduction of concrete compressive strength due to crack surfaces) and the gradual failure of aggregate interlock
cracking and detrimental effect of reinforcement (exemplified with increasing crack opening.
for RC with βs,lim = 0.5). Due to the comparatively large scatter, the development of
the compressive strain ε2,max at peak load does not show a
significant dependency on the tensile strain and/or the reduction
of the concrete compressive strength. The still-quite-high
strains at failure of the two panels with the largest strains in the
Biaxially tested panels tensile direction instead show no influence in this context.
Under tensile loading, the cracks of the panels run perpendicular
to the main tensile direction, as expected (Fig. 3(c)). PROPOSED MATERIAL MODEL
Depending on the applied tensile strain, the test specimens Reduction of compressive strength
showed from five to 14 cracks, whereas the crack spacings and
For the RC and MRC panels, the compressive strength
crack widths were somewhat smaller for the MRC panels.
decreases considerably and almost linearly for small tensile
The results of the following compressive loading until strains. For average tensile strains of approximately 0.004 to
failure show a significant influence of the transverse strain 0.006, the normalized compressive strength stabilizes. At
on the compressive strength and the compressive stiffness. larger tensile strains, hardly any change in the load-bearing
Thus, even for tensile strains below 0.002, the concrete capacity with increasing tensile strain is recognizable.
compressive strength of the RC panels showed a maximum
decrease of approximately 25% compared with the cylinder For a functional relationship of these characteristics, a
compressive strength. For very large average steel strains ε1, simple bilinear or trilinear approach is appropriate. Thus, the
the maximum reduction of the compressive strength effective compressive strength of the concrete under short-
amounted to approximately 50%. term loading is proposed to be calculated as follows (Fig. 4)
The MRC panels (with a maximum reduction of the
compressive strength of approximately 28%) show a higher fce = βs · fc′ (2)
relative compressive strength σc2/fc,cyl when compared to
RC panels with a similar ε1. The ratio between the relative where βs is the reduction factor according to Eq. (3) to
strengths of Panels 020VK2 (0.75) and F020VK2 (0.84, with account for the effect of transverse tension (expressed by the
fibers) amounts to 0.89 and thus reflects the coefficient of 0.9 as tensile strain ε1) on the concrete compressive strength, and
derived earlier to account for the detrimental effect of bar fc′ is the specified compressive strength of concrete.
reinforcement for panels reinforced with reinforcing bars only.
For the three MRC panels with large strains in the tensile βs,lim ≤ βs = βs,det (1.0 – 80ε1) ≤ βs,det · βs,0 (3)
direction (Panels F025VK2, F050VK2, and F250VK2), no
further influence could be found with the increasing transverse
strain. This can be explained by the fact that the reinforcing where ε1 is the tensile strain acting perpendicularly to the
steel reached the yield stress within these tests only in one compressive direction; βs,0 is the basic value of the reduction
crack. The further deformations localized in this crack. This referred to uniaxial concrete compressive strength due to
behavior is caused by the contribution of the fibers to the transverse tensile loading; βs,det is the reduction factor to
load-bearing in the crack for small tensile strains and thus account for the detrimental effect of the reinforcing bars; and
small crack widths. With increasing crack width, the fibers βs,lim is the maximum reduction referred to the uniaxial
are pulled out of the concrete matrix and/or rupture. Hence, concrete compressive strength due to transverse loading.
the load-carrying capacity in this crack gradually decreases For the terms βs,0, βs,det , and βs,lim, the following values
to the load-carrying capacity of the bar reinforcement. This are proposed on the basis of the experimental results.

354 ACI Structural Journal/May-June 2011


⎧ acc. to Kupfer 10 or conservatively

⎪ 5/6 for uncracked RC members (relevant is maximum value)

β s, 0 = ⎨ 5/6 for cracked RC members

⎪ 1.0 for MRC members (fibers with hooked ends,
⎪ λ = 80, fiber content 0.5 vol. %)

⎧ 0.9 for RC members



β s, det = ⎨ 1.0 for MRC members (fibers with hooked ends,

⎩ λ = 80, fiber content 0.5 vol. %)

⎧ 2
c′ [N/mm ]
⎪ 0.8 – f----------------------------
- ≥ 0.5 for RC members (4)
⎪ 140
β s, lim = ⎨
⎪ 0.7 for MRC members (fibers with hooked ends, Fig. 5—Compressive stress-strain relationship of RC and

⎩ λ = 80, fiber content 0.5 vol. %) MRC members under biaxial compression-tension loading.

The proposal will be verified against various experimental from the compression direction. With lower-strength
results and discussed in the following paragraphs. concretes, failure of the complete member does not occur as
Because transverse tensile loadings lower than the cross- soon as the capacity of a single compression strut is reached
sectional cracking load were not investigated, the relationships because their higher ductility compared to higher-strength
valid for plain concrete, according to Kupfer,10 are concretes enables load redistributions between parallel
suggested to be applied conservatively also to uncracked RC compression struts. The addition of fibers shows a favorable
members. Thus, the reduction of the concrete compressive influence regarding the crack widths and thus regarding the
strength can be determined in the uncracked state as a function effect of aggregate interlock.
of the transverse tensile stress. Accordingly, for RC members, the maximum reduction of
In the cracked state, the effective compressive strength fce the concrete compressive strength βs,lim subject to high
of RC members is limited conservatively by a basic value of transverse strains is defined depending on the specified
βs,0 = 5/6 due to the fact that the concrete compressive compressive strength of concrete according to Eq. (4). Thus,
strength decreases in the phase of crack formation to a good agreement with both the authors’ own test results9
approximately 80% of the uniaxial compressive strength. and comparable investigations of other researchers can be
Considering the detrimental effect of the reinforcing bars observed (refer to the following section).
(βs,det = 0.9), the effective compressive strength of a panel For the MRC members with a fiber content of 0.5 volume
amounts to fce = 0.75fc′ . As the comparison with experi- percent, the maximum reduction of concrete compressive
mental results (refer to the following section) shows, in some strength can be determined on the basis of the authors’ own
cases, quite conservative results are achieved by this experimental results9 to be approximately βs,lim = 0.7. This
approach. Based on the presented test results and to reduce limit could also be observed within a test series on the biaxial
complexity, a basic value of βs,0 = 1.0 is also reasonable for behavior of UHPC with mixed reinforcement (fc′ ≈ 160 N/mm2
RC members. This would result in a linear course between [23,200 psi], fiber content 1.0 volume percent).11
βs,det and βs,lim (thick dashed line in the phase of crack
formation in Fig. 4). Stress-strain relationship under biaxial loading
The latter is also suggested for MRC members. Due to the Along with a reduction of the compressive strength, a
ability of the fibers to effectively transfer tensile forces in the reduction of the compressive stiffness could be observed in
state of microcracking, the proposed material model the biaxial tests. On the other hand, a dependency of the
provides no reduction of the compressive strength of MRC compressive strain at peak load on the applied transverse
members caused by tensile loads lower than the cross- tensile strain could not be observed. Therefore, it is
sectional cracking load and/or by the detrimental effect of suggested to reduce the modulus of elasticity Ec by the same
the reinforcing bars (βs,0 = 1.0, βs,det = 1.0). Thus, for MRC factor βs as the compressive strength and to assume the strain
members, a significantly lower reduction than for RC at peak load ε2,max independently of the transverse tension as
members is achieved at small transverse strains. constant (Fig. 5).
In the phase of stabilized cracking, the crack widths The stress-strain relationship of biaxially loaded RC and
significantly increase with the increasing load. Therefore, MRC members may then be calculated according to Eq. (5)
depending on the grain size, the aggregate interlock effect (modified approach of DIN 1045-18)
gradually decreases with the increasing tensile strain. In the
presence of inclined cracks, a load transfer from compression 2
k⋅η–η
strut to compression strut, as well as their mutual support, is σ c = – f ce ⎛ ----------------------------------⎞ (5)
thus marginal or impossible. The maximum reduction βs,lim ⎝ 1 + ( k – 2 ) ⋅ η⎠
is then determined primarily by the geometry of the
remaining compression struts. The strength reduction of the where η = εc /ε2,max ; ε2,max is the compressive strain at peak
compression struts is expected to be larger the more slender load under uniaxial/biaxial loading; k = –Ec /fc′ · ε2,max; and
they are, the more irregular their boundaries are, the more Ec is the tangent modulus of elasticity at the origin of the
irregular in their widths they are, and the more they deviate stress-strain-curve.

ACI Structural Journal/May-June 2011 355


COMPARISON WITH STUDIES compression; the other panels were loaded sequentially in
OF OTHER RESEARCHERS tension and compression.
Test programs and setups In Karlsruhe, Germany, Eibl and Neuroth5 conducted 15
A direct comparison of the individual test data of previous panel tests. Two of these were unreinforced and 13 were
studies is very difficult because the test executions and reinforced with one or two layers in the tension direction
reinforcing configurations vary widely. Therefore, the following only. The test specimens had a width of 1000 mm (39.4 in.)
discussion is focused on those tests that involve sequential or
(tension direction), a height of 800 mm (31.5 in.) (compression
proportional compression-tension loadings and where the
direction), and a thickness of 80, 160, or 200 mm (3.15, 6.30,
cracks occur regularly parallel to the direction of the
compressive stress, as they did in the authors’ own tests.9 or 7.88 in.). The reinforcing bar diameter ds amounted to 10,
16, or 20 mm (0.39, 0.63, or 0.79 in.) and the reinforcement
On this basis, the trendsetting results of the 1980s from
ratio ρs was between 1.0 and 2.5%. The cylinder compressive
Bhide and Collins,12 Belarbi and Hsu,2 Schlaich and
Schaefer,3 Kollegger and Mehlhorn,4 and Eibl and Neuroth,5 strength varied between 31 and 37 N/mm2 (4495 and 5365 psi).
as well as the investigations published by Schiessl13 in 2005 Nine of the RC panels were loaded sequentially in tension
are selected. and compression; the other panels were tested uniaxially in
Bhide and Collins12 conducted a total of 31 panel tests in compression only.
Toronto, ON, Canada, in 1986, including five normal- The experimental tests conducted by Schiessl13 in
strength concrete (NSC) panels under biaxial compression- Munich, Germany, covered 58 panels under uniaxial
tension. The square test specimens had a length of 890 mm compression and biaxial compression-tension. The square
(35.1 in.) and a thickness of 70 mm (2.76 in.). They were test specimens had a length of 1000 mm (39.4 in.) and a
reinforced in two layers exclusively in the tensile direction. thickness of 100 mm (3.94 in.). Seven panels were unre-
The cylinder compressive strength varied between 20 and inforced; the other panels were reinforced with a single layer
28 N/mm2 (2900 and 4060 psi). Further variable parameters in the tensile direction only, consisting of 12 reinforcing bars.
were the reinforcement ratio (1.1 and 2.2%) and the type of The reinforcing bar diameter ds was 12 mm (0.47 in.) and the
reinforcing steel (S400 and S500). reinforcement ratio ρs was 1.4%. Along with an NSC with a
Belarbi and Hsu2 performed a total of 22 panel tests in cylinder compressive strength of 38 to 44 N/mm2 (5510 to
Houston, TX, in 1991. The square test specimens had a 6380 psi), a high-strength concrete (HSC) with a cylinder
length of 1397 mm (55 in.) and a thickness of 178 mm (7 in.). compressive strength of 73 to 82 N/mm2 (10,585 to 11,890
They were reinforced in the tensile and compressive directions psi) and a self-consolidating concrete (SCC) were also
with one reinforcing layer each. The reinforcement ratio examined. The inclination of the cracks regarding the main
amounted to 0.3 up to 2.1% and 0.5% in the tensile and compression stress direction was varied. Altogether, five
compressive directions, respectively. The cylinder compressive unreinforced and four reinforced NSC and HSC panels were
strength varied between 37 and 48 N/mm2 (5365 and 6960 psi). tested uniaxially. Five reinforced panels made of NSC and
With the exception of two panels, the specimens were loaded HSC—each with cracks running parallel to the compressive
sequentially or proportionally in compression and tension.
direction—were loaded biaxially. In general, the load was
In Stuttgart, Germany, Schlaich and Schaefer3 conducted applied manually force controlled in both the tensile and
tests on two plain concrete and eight RC panels in 1982. The
compression directions.
test specimens held a single layer of orthogonal reinforcement
ds = 10 mm (0.39 in.), which was oriented either parallel to
the compression/tension direction or at a 45-degree inclination Comparison of test results and material models
(four panels each). The reinforcement ratio was varied by To ensure comparability of the test results, a preparation of
modifying the bar spacings (ρs = 0.8 and 1.6%). The test the test data analogously to the authors’ own test evaluation9
specimens had a height of 400 mm (15.8 in.) (tension direction), became necessary.
a length of 1300 mm (51.2 in.) (compression direction), and a In Fig. 6, the relative concrete compressive strengths
thickness of 100 mm (3.94 in.). The cylinder compressive σc2/fc,cyl achieved in the different test series are depicted in
strength varied between 22 and 24 N/mm2 (3190 and 3480 psi). dependency of the main tensile strain ε1 (applied tensile
The plain concrete panels and four RC panels were tested strain or transverse strain at peak load). They are compared
uniaxially in compression. The remaining four panels were to the material models of Vecchio and Collins1 according to
loaded deformation controlled in tension and compression. Eq. (6a) and (6b), Belarbi and Hsu2 according to Eq. (7a) and
Kollegger and Mehlhorn4 performed a total of 47 tests in (7b), and the authors’ own proposal9 according to Eq. (3). At
Kassel, Germany, between 1984 and 1987, using a similar this point, the reduction factor βs,det that considers the influence
test setup as illustrated in Fig. 2. Four test specimens were of the detrimental effect of the reinforcing bars was determined
unreinforced and 31 had two layers of orthogonal reinforcement for the individual test series comparing the load-bearing
in the tension and compression directions each. In 12 further capacities of the uniaxially loaded reference panels (plain
tests, the reinforcing mesh was rotated by 45 degrees relative
concrete and RC panels) or was assumed to be 0.9 in the
to the main loading directions. Either deformed bars or wire
absence of the necessary data. The maximum reduction
mesh were used as reinforcement, consisting of smooth
wires or deformed bars with different diameters ds of 6.5, βs,lim was calculated according to Eq. (4), depending on the
8.5, or 10 mm (0.26, 0.33, or 0.39 in.). Thus, the reinforce- specified concrete compressive strength applied in the tests.
ment ratio ρs varied between 0.7 and 1.7%. The dimensions If βs,0 is assumed to be 1.0, the fat dashed line is achieved in
of the test specimens were equal to those shown in Fig. 1. the range of small transverse strains.
The cylinder compressive strength varied between 9 and The following is the reduction of the concrete compressive
21 N/mm2 (1305 and 3045 psi). The plain concrete panels strength due to tensile strain and cracking according to
and six of the RC panels were loaded uniaxially in Vecchio and Collins1

356 ACI Structural Journal/May-June 2011


f ce 1
----- = ------------------------------------------- (6a)
f c′ ( 0.8 + 0.34ε 1 /ε c′ )

where εc′ is the strain in concrete cylinder at peak stress fc′ .


Assuming a usual value of εc′ = –0.002 leads to

f ce 1
----- = --------------------------------- (6b)
f c′ ( 0.8 + 170ε 1 )

The following is the reduction of the concrete compressive


strength due to tensile strain and cracking according to
Belarbi and Hsu2

f ce 0.9
----- = --------------------------- (sequential loading) (7a)
f c′ 1 + 250ε 1

f ce 0.9
----- = --------------------------- (proportional loading) (7b)
f c′ 1 + 400ε 1

Discussion of test results


For the evaluation and classification of the test results, it is
very important to consider the various problems that
occurred within the aforementioned test series.
Thus, within the tests conducted by Kollegger and
Mehlhorn,4 the transverse strain increased during the
compressive loading unintentionally by a factor of up to 4
compared to the applied tensile strain. This might be caused
by the elastomeric supports used for the load application.
Even those panels loaded uniaxially in compression only
showed cracks in the compressive direction and transverse
strains of up to 0.0035 immediately before failure. The plain
concrete panels loaded “uniaxially in compression” always
failed in transverse tension (cracks parallel to the direction of
compression loading). Likewise, Eibl and Neuroth5 report
the early failure of the corner regions and longitudinal cracks
at the face side of the specimens. All these observations
argue for deficiencies in the test setups.
Subsequently, Eibl and Neuroth5 consistently achieved
very low relative concrete compressive strengths σc2/fc,cyl of
the panels of 0.59 to 0.68 (refer to Fig. 6(e)). If the lowest
relative concrete compressive strength of all panels loaded Fig. 6—Comparison of test results of several researchers with
biaxially (σc2/fc,cyl = 0.59) is referred to the mean relative proposals of Vecchio/Collins, Belarbi/Hsu, and Fehling/
concrete compressive strength of the panels loaded uniaxially Leutbecher/Roeder for the reduction of concrete compressive
only, however, it results in approximately 0.85. Eibl and strength under biaxial compression-tension loading.
Neuroth5 therefore regarded a reduction of the concrete
compressive strength due to transverse tension by 15% as
sufficient if the yield stress of the reinforcement was not less significant for higher values of transverse strain. This
exceeded. Kollegger and Mehlhorn4 evaluated their tests in phenomenon applies exactly to the disturbances observed in
a similar way. They determined the relationship of the the authors’ own tests9 (local failure caused by the PTFE foil
minimum to the maximum relative concrete compressive and segregation of the concrete mixture). In this case, the
strength for each series to be approximately 0.8. Consequently, evaluation, according to Eibl and Neuroth5 and Kollegger
they derived a maximum reduction of the concrete compressive and Mehlhorn,4 suffers from the small variation of strength
strength by 20%. reduction and would significantly underestimate the actual
This kind of evaluation may produce misleading results, reduction of the concrete compressive strength due to
however, if influences other than transverse tension (such as transverse tension.
segregation of the concrete mixture, transverse tensile The tests conducted by Eibl and Neuroth5 in particular show
stresses caused by the support, and nonuniformly distributed no substantial influence of the transverse tensile strain because
load application) lead to a decrease of the strength in the even for plain concrete and RC panels loaded in uniaxial
compression direction at the same time. The effect of such compression, reductions of 25 to 40% existed. Therefore, the
influences may be significant for low transverse strains but significance of these tests should be regarded critically.

ACI Structural Journal/May-June 2011 357


agreement with reductions to between approximately 50 and
60% of the cylinder compressive strength due to average
tensile strains of approximately 0.005. In this range of defor-
mations, the different material models do not produce
substantial differences either. Large transverse tensile strains
were not examined by Schiessl.13 The authors’ own tests,9
however, show that the compressive strength of the panel
hardly decreases any further if the load is applied via even
contact areas. This observation justifies the assumption of a
lower limit βs,lim being independent from the transverse
strain in the authors’ own proposal.
In Fig. 7, the relative concrete compressive strengths βs,exp
achieved in the tests are related to the computed values βs,cal
according to the suggested material model. At this point, βs,0
in Eq. (3) was set to 1.0. The tests conducted by Eibl and
Neuroth5 were omitted due to the disturbances that could not
be clarified. The tests conducted by Bhide and Collins12 and
Belarbi and Hsu2 remained out of consideration because, due
to the kind of load application, only the weakest portion of
Fig. 7—Relation between experimentally obtained reductions the entire cross section determines the ultimate load. Also,
of compressive strength βs,exp and theoretical values βs,cal the test results of Kollegger and Mehlhorn4 considered in
according to Eq. (3). Fig. 7 have to be regarded critically because of the low abso-
lute concrete compressive strength and its large scatter.
Regarding the test results of Schlaich and Schaefer3 Despite this restriction, the deviation between test and model
(Fig. 6(c)), Kollegger and Mehlhorn4 (Fig. 6(d)), and Bhide data amounts—with one exception—to a maximum of
and Collins12 (Fig. 6(a)), it also has to be considered that approximately 20%. When exclusively regarding the authors’
very low specified compressive strengths—currently no own test series9 and the investigations of Schiessl,13 the
longer common for structural concrete—were used within maximum difference results in 11%. Altogether, a quite good
the test series. For these low concrete compressive strengths, the agreement exists between the results of the significant test
absolute reductions of the compressive strength due to series and the proposal according to Eq. (3).
transverse tension lie in the same order of magnitude as the
scatter of the compressive strength itself. In these cases, it CONCLUSIONS
becomes very difficult to differentiate between the influence of In spite of numerous and sometimes very extensive test
the transverse tension and the variation of the concrete quality. series that have been performed during the past decades to
Bhide and Collins12 and Belarbi and Hsu2 also examined study the biaxial compressive strength of cracked RC with
panels under large plastic tensile strains, unlike Schlaich and transverse tensile loading, only a few conclusive results are
Schaefer,3 Kollegger and Mehlhorn,4 and Eibl and Neuroth.5 available. This is due to disturbances resulting especially
From this point of view, the lower relative concrete compressive from the test setup and execution. Past studies, as discussed
strengths σc2/fc,cyl achieved in the tests are not contradictory in this paper, illustrate this situation. Additionally, in former
to the German results of the 1980s. This holds particularly studies, useful data have been primarily obtained for very
true for Belarbi and Hsu,2 who obtained similar reductions as low concrete strengths, which are currently unusual.
the German researchers for small transverse tensile strains A further aspect that has to be critically regarded
(Fig. 6(b)). concerns the evaluation of the test results. In studies in
Regarding the load application, however, the test setups Germany in the 1980s, the reduction of the concrete
used by Bhide and Collins12 and Belarbi and Hsu2 share a compressive strength was determined using the ratio of the
characteristic. The tensile and compression loads were minimum to the maximum relative concrete compressive
applied by several hydraulic jacks at each panel side. The jacks strength of a test series, disregarding the actual low relative
were driven force-controlled, depending on the oil pressure of compressive strength σc2/fc,cyl of the individual panels. By
the hydraulic system. Thus, they acted like independent but critical evaluation, more unfavorable values for the biaxial
identical loads at the panel’s surface. Contrary to rigidly compressive strength can also be derived from these test series.
coupled jacks or deformation-controlled loading of several Further investigations on the biaxial compression-tension
independent jacks, no load redistribution over the panel’s strength showed a more pronounced reduction in comparison to
length—as is possible in real structures—could take place in the uniaxial concrete compressive strength, especially for
the tests. Within a test specimen with nonuniform cracking, large transverse tensile strains. The very low values that were
the load-bearing capacities of the individual compression obtained can be attributed to the kind of load application as a
struts can greatly differ. This holds especially true for large main reason. Bhide and Collins12 and Belarbi and Hsu2 used
tensile strains/crack openings associated with practically no several hydraulic jacks that worked independently from each
aggregate interlock. Therefore, the large reductions to other but were controlled by a common oil pressure. Thus,
between 20 and 30% of the uniaxial concrete compressive the strength determined in this way, especially for large
strength, as observed by Bhide and Collins12 and Belarbi and transverse tension strains, represents only a lower limit as
Hsu,2 represent the strength of the weakest compression strut given by the load-carrying capacity of the weakest portion of
rather than the actual strength of the entire panel. the panel. Hence, the obtained values do not adequately
The comparison of the authors’ own results9 (Fig. 6(g)) represent the strength of an entire panel. Thus, the compressive
and the tests conducted by Schiessl13 (Fig. 6(f)) show a good strength may be underestimated by these investigations.

358 ACI Structural Journal/May-June 2011


The material model suggested by the authors specifies the 6. ACI Committee 318, “Building Code Requirements for Structural
biaxial compression-tension strength of cracked RC with Concrete (ACI 318-05) and Commentary,” American Concrete Institute,
Farmington Hills, MI, 2005, 430 pp.
and without fibers and the stress-strain relationship in 7. BS EN 1992-1-1:2004, “Eurocode 2: Design of Concrete Structures,
compression, depending on the principal tensile strain. Part 1-1: General Rules and Rules for Buildings,” British Standards Institute,
Validated by the authors’ own test results9 and foreign test London, UK, Dec. 2004, 230 pp.
8. DIN 1045-1, “Tragwerke aus Beton, Stahlbeton und Spannbeton: Part 1:
data, and after a thorough critical discussion of the relevance Bemessung und Konstruktion,” Normenausschuss Bauwesen (NABau) im
of individual tests and test series, the model offers a more DIN Deutsches Institut fuer Normung e. V., Beuth Verlag, Berlin,
realistic characterization of material behavior and a goal- Germany, Aug. 2008. (in German)
9. Fehling, E.; Leutbecher, T.; and Roeder, F.-K., “Biaxial Compression-
oriented design of biaxially loaded structures. Tension-Strength of Reinforced Concrete and Reinforced Steel Fibre
Concrete in Structural Panels,” Structural Materials and Engineering
REFERENCES Series, No. 11, Kassel University Press, Kassel, Germany, 2008, 139 pp.
1. Vecchio, F. J., and Collins, M. P., “The Modified Compression Field 10. Kupfer, H. B., “Das Verhalten des Betons unter Mehrachsiger
Kurzzeitbelastung unter Besonderer Beruecksichtigung der Zweiachsigen
Theory for Reinforced Concrete Elements Subjected to Shear,” ACI JOURNAL,
Beanspruchung,” Deutscher Ausschuss fuer Stahlbeton (DAfStb), No. 229,
Proceedings V. 83, No. 2, Mar.-Apr. 1986, pp. 219-231. Beuth Verlag, Berlin, Germany, 1973. (in German)
2. Belarbi, A., and Hsu, T. T. C., “Constitutive Laws of Softened 11. Fehling, E.; Leutbecher, T.; Roeder, F.-K.; and Stuerwald, S.,
Concrete in Biaxial Tension Compression,” ACI Structural Journal, V. 92, “Structural Behavior of UHPC under Biaxial Loading,” Proceedings of the
No. 5, Sept.-Oct. 1995, pp. 562-573. Second International Symposium on Ultra-High-Performance Concrete,
3. Schlaich, J., and Schaefer, K., “Zur Druck-Querzug-Festigkeit des Structural Materials and Engineering Series, No. 10, Kassel University
Stahlbetons,” Beton und Stahlbetonbau, Ernst & Sohn, Berlin, Press, Kassel, Germany, 2008, pp. 569-576.
Germany, V. 78, No. 3, Mar. 1983, pp. 73-78. (in German) 12. Bhide, S. B., and Collins, M. P., “Reinforced Concrete Elements in
Shear and Tension,” Publication No. 87-02, Department of Civil
4. Kollegger, J., and Mehlhorn, G., “Biaxiale Zug-Druckversuche an
Engineering, University of Toronto, Toronto, ON, Canada, 1987, 147 pp.
Stahlbetonscheiben,” Research Report No. 6, Chair of Concrete Structures, 13. Schiessl, A., “Die Druckfestigkeit von Gerissenen Scheiben aus
University of Kassel, Germany, 1988, 103 pp. (in German) Hochleistungsbeton und Selbstverdichtendem Beton unter besonderer
5. Eibl, J., and Neuroth, U., “Untersuchungen zur Druckfestigkeit von Beruecksichtigung des Einflusses der Rissneigung,” Deutscher Ausschuss
Bewehrtem Beton bei Gleichzeitig Wirkendem Querzug,” Research Report fuer Stahlbeton (DAfStb), No. 548, Beuth Verlag, Berlin, Germany, 2005,
T 2024, IRB Verlag, Stuttgart, Germany, 1988. (in German) 187 pp. (in German)

ACI Structural Journal/May-June 2011 359


View publication stats

You might also like