Sie sind auf Seite 1von 12

Engineering Structures 115 (2016) 220–231

Contents lists available at ScienceDirect

Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct

Mechanical model for the shear capacity of R/C beams


without stirrups: A proposal based on limit analysis
Jakob Fisker ⇑, Lars German Hagsten
Aarhus School of Engineering, Aarhus University, Denmark

a r t i c l e i n f o a b s t r a c t

Article history: In this paper a rational model for the shear capacity of beams without shear reinforcement is presented,
Received 11 August 2015 relating to the basic case of a straight beam subjected to combined shear and bending. The model is
Revised 19 February 2016 established using the upper bound principle derived from the Theory of Plasticity, and rests upon the
Accepted 22 February 2016
observation that shear failures in slender beams typically evolve from already existing cracks, developed
Available online 7 March 2016
due to the influence of bending, and shaped under the combined influence of shear and bending. In the
model, this observation is formalised into a characteristic crack pattern and a characteristic pattern of
Keywords:
failure lines along existing cracks and within uncracked concrete. The influence of crack widths is taken
Reinforced concrete
Analytical modelling
into account using a relation between the crack failure condition and the crack width. The estimates of
Concrete mechanics the model are shown to be in good agreement with a wide range of tests as well as with existing models.
Plasticity Ó 2016 Elsevier Ltd. All rights reserved.
Shear
Beams without shear reinforcement

1. Introduction shear forces along the beam prior to failure [2,3]. Today it is widely
accepted that the fundamental difference between ‘‘deep” beams,
Shear failures in reinforced concrete members without shear and ‘‘slender” members is attributed to the favourable influence
reinforcement typically develop suddenly, and without prior warn- of strut and tie action present in deep beams. As the slenderness
ing. Obviously, such failure modes are unwanted and undermine a is increased shear is transferred by the activation of alternative
sound structural design, where, preferably, the incipient collapse mechanisms such as aggregate interlock, residual tensile stresses
should be preceded by warning signs as those anticipating a ductile in cracks and dowel action, all interacting in combination with
failure in bending, such as significant visible cracking and large transfer of shear within the compressive zone [4]. Another appar-
deflections. Unfortunately, the development of shear failures and ent characteristic is, as illustrated in Fig. 1b, the nature of the man-
the corresponding capacity, has proven to be a highly complex ner during which the member fails in shear. As opposed to the
issue, as demonstrated by the great amount of research activity often stable crack/failure line development observed for short
on this topic. Still, even what appears to be the most basic case – beams, failure in slender beams is typically triggered by the sudden
‘‘constant shear” – introduces several questions for which no propagation of an existing flexure-shear crack into the compressive
rational answers can yet be given. The experimental study of shear zone. Although such overall characteristics have been identified,
in concrete members has, at least on a structural level, been con- more recent research suggests, that even in the case of apparently
fined mainly to the case of simply supported beams subjected to identical members significant variations on the relative influence
three- or four point bending. During the 1960s and 1970s pioneer- of the primary shear-carrying mechanisms may be present [5].
ing research revealed the existence of certain characteristics on the These variations are likely to be attributed to the inhomogeneity
behaviour of such beams. From a great number of tests Kani postu- of the material in general, and perhaps more specifically, the, in
lated that ‘‘the laws governing the strength of a beam for ða=dÞ less some aspects, rather random nature of the crack development
than 2.5 and for ða=dÞ greater than 2.5 are totally different and regarding shape and location. This may also lead to differences in
unrelated”, as exemplified in Fig. 1a [1]. Tests by, among others, the crack-kinematic, and, effectively, the activation of various
Taylor and Fenwick & Pauley lead to the identification of so- shear carrying mechanisms. In what follows, a rational model for
called shear carrying mechanisms, responsible for the transfer of the shear capacity of beams without shear reinforcement is pre-
sented. The model relates to the basic case of beams with constant
shear, and only slender members are considered. The model rests
⇑ Corresponding author. upon the upper bound theorem provided by the Theory of
E-mail address: jafi@ase.au.dk (J. Fisker).

http://dx.doi.org/10.1016/j.engstruct.2016.02.035
0141-0296/Ó 2016 Elsevier Ltd. All rights reserved.
J. Fisker, L.G. Hagsten / Engineering Structures 115 (2016) 220–231 221

7 test - shear failure. Short beams: (a/d) < 2.5


test - flexural failure.
6

3 Slender beams: (a/d) > 2.5


2

0
0 1 2 3 4 5 6 7 8 9

(a) (b)
Fig. 1. (a) Influence of shear slenderness as observed in tests [12] and (b) characteristic pattern of failure lines in short and slender beams, respectively.

Plasticity, and upon the aforementioned observations that shear layer of the longitudinal reinforcement. The existing crack through
failures in slender beams typically evolves from already existing which the failure develops is located at a certain distance xcr from
cracks. The proposed model intends to contribute to the ever con- the point of the loading. As seen, the assumed pattern of failure
tinuing development of sound and rational models for this impor- lines represents an idealisation and simplification of the pattern
tant topic, and a comprehensive description can be found in [6]. of failure lines often observed in tests involving slender beams [6].

2.1.1. Evaluation of the shear capacity


2. Shear failure criterion The shear capacity V u will be determined using the upper
bound principle derived from the Theory of Plasticity. Assuming
2.1. Basic shear failure mechanism a rigid plastic behaviour at failure, the total energy dissipated dur-
ing failure consists of five different contributions, and, accordingly,
Consider the beam segment illustrated in Fig. 2a, provided with the following balance between the dissipated energy and the work
a single layer of reinforcement. Owing to the action of combined of external forces, can be established using the work-equation.
flexure and shear a certain pattern of inclined cracks develops V u du ¼ Ai;slid:;pl þ Ai;slid:;cr þ Ai;sep: þ Ai;rot: þ Ai;dow ð1Þ
along the shear span, a. The cracks are assumed to propagate to
the level of the neutral axis, and are idealised as being straight. In this expression Ai;slid:;pl represents the dissipation due to sliding
At a distance d from the applied load the inclination is assumed within the plain concrete in the compressive zone, while Ai;slid:;cr rep-
constant and equal to b. As the applied load is increased, new resents the dissipation due to sliding along an already existing
cracks form and the width of already existing cracks increases. crack. The remaining three contributions follows from the part of
Eventually, the applied load reaches the shear capacity V u , and the failure developing along the longitudinal reinforcement; the
the member fails in shear during the development of a mechanism. assumed vertical translation du introduces a separation failure
The assumed mechanism is shown in Fig. 2b, and consist of three within the concrete at the level of the reinforcement (Ai;sep ), while
rigid parts separated by failure lines. Notice, that the presence of the corresponding rotation introduces two hinges leading to bend-
any load-plate has been ignored. The relative displacements ing of the concrete cover (Ai;rot ) and dowel action of the bars (Ai;dow ).
between the parts are localised into these failure lines, and relates However, the work-equation represents a balance of the rate of dis-
solely to the displacement increment du, introduced instanta- sipated energy at failure, and therefore implies that the contribu-
neously at failure. The pattern of failure lines is characteristic in tions are concurrent in time and can be added up. Based upon
the sense that it involves a failure line along an existing crack of experimental observations this assumption appears rather ques-
a certain width, and a failure line propagating from the tip of the tionable [6]. Specifically, it seems unreasonable to expect the plastic
existing crack to a point just next to the applied load. This latter hinge development in the bars to be compatible in magnitude with
failure line is assumed to develop instantly within the previously the relative displacements due to sliding along cracks and within
uncracked concrete in the compressive zone, referred to as plain the compressive zone [5,7]. Therefore, only the first two contribu-
concrete. Finally, a horizontal failure line develops along the single tions will be considered, so that;

(a) (b)
Fig. 2. Beam segment subjected to shear and bending (a) prior to failure and (b) mechanism at failure.
222 J. Fisker, L.G. Hagsten / Engineering Structures 115 (2016) 220–231

V u ¼ ðAi;slid:;pl þ Ai;slid:;cr Þ=du ð2Þ and the dissipation per unit length of the crack can thus be deter-
mined as;
In order to address the shear capacity a set of material failure con-
ditions must be introduced. In what follows, plain concrete and pffiffiffiffiffiffi 1 þ sinð/cr Þ
W L;slid:;cr ¼ ccr kcr duð1  sinða1 ÞÞb; kcr ¼ ð5Þ
existing cracks will both be regarded as homogeneous materials, 1  sinð/cr Þ
each attached with a unique failure condition. In both cases a
Mohr–Coulomb failure condition with a circular transition curve In (5) kcr is, as shown, defined by the angle of friction /cr . From com-
will be assumed in combination with zero tensile strength. Regard- parison with push-off tests a value of /cr ¼ 22 has been assumed,
ing the latter assumption one can argue that it is the contribution to and thus kcr ¼ 2:2. One should observe that this estimate is derived
the dissipation due to concrete in tension (product of stress and from observations of relative displacements along the crack, and
strain) that is ignored rather than the ability to carry tensile stresses not from the ratio between shear and normal stresses [6]. It thus
per se. relies entirely upon the validity of the normality condition. In
Fig. 3b the relation between the sliding capacity, represented by
2.1.1.1. Sliding within plain concrete. For plain concrete a constant scr , and crack opening is compared with tests by Jacobsen et al.
angle of friction of /pl (=37°) is assumed, representative for con- [11] for a fixed value of a. The comparison is shown for two differ-
crete of low to medium strength with normal strength aggregates ent settings of the maximum crack-width w0 . The tests reveal a
[8]. The plastic strain increments within the failure line is derived rather significant influence of the initial crack-opening for which
using the associated flow rule, and the dissipation per unit length tangential (shear) displacements are imposed. This dependency is
of a failure line can thus be evaluated as; not captured by the simple approach introduced in the model. In
real beams a certain initial crack-width will always exist before
1 crack slip occurs. This follows from the experimental observation
W L;slid:;pl ¼ f duð1  sinða2 ÞÞb ð3Þ
2 c that flexure-shear cracks typically initiates as being more or less
In (3) a2 represents the direction of the relative displacements along straight.
the failure, as shown in Fig. 2b, while f c is the uniaxial compressive
strength [9]. 2.1.1.3. Combining contributions. Referring to Fig. 2b again, and
assuming the width of cracks to be constant along the entire crack
2.1.1.2. Sliding along existing cracks. The properties of the existing length, the shear capacity is simply determined as;
cracks are assumed dependent upon the actual width experienced
V u ¼ W L;slid:;cr lcr þ W L;slid:;pl lpl
along the crack. This dependency is introduced by a fictitious
pffiffiffiffiffiffi 1
cohesion-term, partly taken from [10], and given as; ¼ ccr kcr ð1  sinða1 ÞÞblcr þ f c ð1  sinða2 ÞÞblpl ð6Þ
pffiffiffiffi   2
0:18 f c w0  wcr
ccr ¼ ð4Þ lcr and lpl represents the lengths of failure lines, and are defined
0:3 þ 16þdg
24wcr w0
entirely by the failure location, xcr and the geometry of the assumed
In (4) wcr is the crack width while dg refers to the maximum crack pattern. For further reference, the shear capacity shall be
aggregate-size. The term ððw0  wcr Þ=w0 Þ has been introduced by expressed simply as the sum of the two contributions;
the authors in order to enforce a maximum crack width w0 for V u ¼ V cr þ V pl .
which shear stresses can no longer be transferred across the crack.
The magnitude of this upper limit is likely to be dependent upon 2.2. Characteristics of shear failure criterion
variables such as the actual crack kinematics, maximum aggregate
size, concrete composition etc., and will be discussed later. Fig. 3a In the following, the characteristics of the shear failure criterion
illustrates the assumed failure condition for crack widths given in (6) will be illustrated. Consider for this reason the beam
wcr;1 < wcr;2 < wcr;3 . As in the case of plain concrete, and as indicated segment shown in Fig. 4a. It is assumed that flexure-shear cracks
in Fig. 3a, it will be assumed that the normality condition is fulfilled, can develop at any position within the region bounded by two

7
(0.025 mm) model w 0= 2.0 mm
6
model w 0= 1.0 mm
5
τcr [MPa]

4 (0.04 mm)

2
(0.1 mm)
1

0
0 0.2 0.4 0.6 0.8
w [mm]
(a) (b)
Fig. 3. (a) Crack failure condition with plastic strain diagram superimposed (dn and ds are normal- and shear displacements within a failure zone of height D and unit
thickness), (b) comparison with tests for fixed value a = 40° and different initial crack-openings (bracket numbers). Specimens made from normal strength concrete and
dg = 8 mm.
J. Fisker, L.G. Hagsten / Engineering Structures 115 (2016) 220–231 223

1.0
Vu,rel
Vcr,rel 0.8
Vpl,rel

Vu,rel,Vcr,rel,Vpl,rel
0.6
x cr,min/xcr,max

Position of
0.4
critical crack

0.2

0
1.0 0.8 0.6 0.4 0.2 0
x cr /xcr,max

(a) (b)
Fig. 4. (a) Assumed bounding crack position, xcr;min and xcr;max , (b) plots of V u ; V pl and V cr relatively to max V u .

extreme positions; xcr;min and xcr;max . Each crack within this interval Unfortunately, neither the position of the critical crack, nor the
represents a potential critical crack, through which a failure could load and thus width at which it becomes critical is known a priori.
develop. In line with the presence of a constant shear force, the For this reason an estimate on the crack position and crack widths
crack widths at a given load level V act are assumed to vary linearly along the shear span must be addressed.
along the span.
Fig. 4b displays the variation of the shear capacity related to 2.3. Estimation of cracking development
each possible crack position within the interval, and the variation
of the two contributions that constitutes the total capacity. The In the following, a simplified estimate on the gradual develop-
contribution from V pl attains its maximum just next to the applied ment of cracks will be addressed. More specifically the estimate
load, and then decreases along the span due to the increase of the involves; the spacing between cracks, their height and inclination
angle a2 , defining the direction of the relative displacements. Con- and, finally, their width. Consider for this reason the slender beam
versely, the contribution and variation from V cr is more or less segment illustrated in Fig. 6a. The element is assumed slender in
inversely related to the variation of crack widths along the shear the sense that any influence of arch/direct strut-action with
span, and thus reaches its minimum at xcr;min . As indicated, the respect to the flow of internal forces is ignored.
sum of the contributions reaches a minimum at a position located As mentioned, cracks are assumed to propagate to the level of
somewhere near the center of the shear span. A failure at this posi- the neutral axis as straight lines with an inclination b. With the
tion renders the least resistance, and thus the apparent shear exception of cracks located within a distance of d from the point
capacity of the beam. If, however, V u > V act the actual shear capac- of load application, a value of b ¼ 30 will be assumed in the fol-
ity will be even less due to further increase in crack widths, and the lowing (and consequently, a1 ¼ 30 ). The possible development
location of the critical crack may change. of failures through cracks located closer to the point of load appli-
Using a similar approach it is possible to address the expected cation will be considered later. What regards the height of the
influence of various design parameters such as e.g. longitudinal compressive zone, y0 , this will simply be assumed equal to the
reinforcement ratio or slenderness. Consider for instance the dia- height of the compressive zone in a cracked, elastic RC-section sub-
grams shown in Fig. 5a, where the influence of longitudinal rein- jected to flexure only. What regards the spacing between cracks
forcement ratio is illustrated. An increase of the reinforcement the simple approach proposed in [13] will be applied, where
ratio would lead to a greater compressive zone height, along with scr ¼ 0:7ðd  y0 Þ. Having quantified the pattern of flexure-shear
smaller crack widths for a fixed load level, and, consequently, an cracks, the final piece of information needed for the failure-
increase in capacity when compared to a lighter amount of rein- criterion, the crack widths wcr , can be addressed. Consider for
forcement. Due to the smaller crack widths along the span, the instance the nth crack located at a distance ðn  1Þscr from
variation of V cr will be suppressed to that of V pl . Besides the the point of load application. If the elongation of concrete in ten-
increase in capacity, an increase of the reinforcement ratio there- sion is ignored the width of this particular crack can be estimated
fore also influences the position of the critical crack, which as;
becomes governed mainly by the variation of the contribution
T nav g scr
from V pl . Similar relations holds for the influence of the shear slen- wncr ¼ ð7Þ
E s As
derness ða=dÞ, as illustrated in Fig. 5b. An increase of slenderness
leads to the variation of the contribution from V cr gradually The width expressed in (7) is determined as the sum of reinforce-
becoming more dominating, while the actual resistance towards ment strains along a distance 0.5scr on either side of the particular
sliding decreases due to increased crack widths for equal loading. crack. The influence of tension-stiffening along this distance is
The location of the critical crack thus moves towards the position taken into account using a simplified approach shown Fig. 6b,
of maximum bending moment, while the capacity is decreased. illustrating a small concrete segment located in-between two
Similar overall tendencies regarding this change in the position adjacent cracks. The average tensile force, T nav g , used in (7) can
of the critical crack can be observed, for instance, in tests by Leon- be determined as;
hardt and Walther with varying slenderness [12].
224 J. Fisker, L.G. Hagsten / Engineering Structures 115 (2016) 220–231

2.0 2.0

Vu,rel increased ρ Vu,rel min (a/d)


Vu,rel reduced ρ 1.5 Vu,rel max (a/d) 1.5

Vu,rel

Vu,rel
1.0 1.0

0.5 0.5

0 0
1.0 0.8 0.6 0.4 0.2 0 1.0 0.8 0.6 0.4 0.2 0
x cr /xcr,max x cr /xcr,max

(a) (b)
Fig. 5. Normalised V u plotted against normalised xcr (a) influence of q. The variation of V u is plotted relatively to min. V u for max. q (b) influence of ða=dÞ. The variation of V u is
plotted relatively to min. V u for min. ða=dÞ.

(a) (b)
Fig. 6. (a) Development of cracks along shear span, (b) diagram of reinforcement forces in-between flexure-shear cracks. T-forces determined from sectional analysis.
! 2
!
V act ða  n þ 1Þscr ldeb In Fig. 7b the test-results are plotted along with the results of the
T nav g ¼ þ V act tanðbÞ þ kb ldeb  0:25scr  ; model including the effectiveness factors, hence, V u;eff ¼ mV u . The
d  13 y0 scr
designation ‘‘eff” will, however, be omitted. The model estimates
kb ¼ pDnsb ð8Þ obtained using mlow and mupp define the vertical extension of the
The expression inside the first parenthesis represent the total force grey-shaded region containing the apparent scatter in the tests
in the n-reinforcement bars at the considered cracked section while results. The introduced effectiveness factor is seen to originate from
the expression given in the second parenthesis accounts for the tests on normal strength concrete only. As will be shown, however,
influence of tension stiffening due to bond with ldeb representing a it appears that the same effectiveness factors can be used success-
local region with zero bond stresses, and taken as 0.65c [6]. Using fully also outside of this range. It is of interest to notice that the
expressions (6) and (7) the shear capacity can thus be addressed effectiveness factors are almost identical to the one traditionally
as a function of the crack position and the applied load. applied in the case of beams with stirrups, ð0:7  f c =200Þ [15]. In
the present case, however, the effectiveness factors are seen to carry
a slightly less dependency upon the concrete strength.
3. Effectiveness factor

The model overestimates the shear capacity when compared to 4. Comparison with selected tests
test, and in order to counter this an empirical correction in terms of
an effectiveness factor is introduced. The effectiveness factor will In the following the model will be compared with various test
be determined from a comparison with tests by Moody et al. results. Focus has been on what appears to be the governing design
[14], which, to the knowledge of the authors, represents the only parameters; the concrete strength, represented by f c , the amount
experimental investigation where the influence of the compressive of reinforcement, q and size effects owing to the effective height
strength of normal strength concrete has been investigated across d and slenderness a=d. All tests refer to simply supported beams
a broader range through multiple tests. The test results naturally or slabs subjected to three of four point bending.
include a certain degree of scatter, and for this reason upper and
lower values of the effectiveness factor have been estimated. The 4.1. Concrete strength
comparison revealed that good agreement could be obtained using
the following factors; As already seen from the comparison with the tests of Moody
fc fc fc et al. the model shows that an increase of the concrete strength
mmid ¼ 0:71  ; mlow ¼ 0:65  ; mupp ¼ 0:77  ð9Þ
380 380 380 has a positive influence on the shear capacity. According to the
J. Fisker, L.G. Hagsten / Engineering Structures 115 (2016) 220–231 225

2.00

1.50

τu [MPa]
1.00

0.50 model with ν mid


Tests by Moody et al

0
0 10 20 30 40 50
fc [MPa]
(a) (b)
Fig. 7. (a) Test-setup and properties of B-series, (b) comparison including effectiveness factors.

expression in (6) the contribution from sliding within the compres- observations are consistent with findings in other experimental
sive zone is proportional to f c . For moderate amounts of reinforce- studies [17,18], and are likely to be attributed to the increased brit-
ment, however, the compressive zone is rather narrow, and when tleness of the concrete. The apparent drop in capacity may have
combined with a large angle a2 the contribution from sliding along been amplified by the use of small-size aggregates and the – unin-
existing cracks therefore tends to be dominating. Consequently, as tended – use of a different aggregate-material for the group B4,
observed from the expression in (4), the net increase in shear where mainly flint was used as opposed to mixed sea-material
capacity decreases as the compressive strength is further for the other groups. The only way to account for such changes
increased. This tendency is amplified by the corresponding in the overall brittleness of the concrete, and especially the conse-
increase in crack widths due to increased capacity, and, unfortu- quent changes of the surface texture of cracks, is by modifying the
nately, also from the minor dependency of the effectiveness factor effective maximum aggregate size and also the assumed value of
upon the concrete strength. w0 . In the case of the latter parameter, it appears reasonable to
Fig. 8a shows a comparison between the model estimates and expect the upper limit represented by w0 to be critically influenced
tests performed at Aarhus University [16]. The only intended vari- by changes in the roughness. The results of such manipulations are
ation was the concrete strength, and the beams were subdivided shown in Fig. 8b where the dashed line illustrates the influence of a
into four groups of different strength (B1 to B4) with three dupli- reduced value of w0 .
cates per group. The capacity is seen to be reasonable estimated Fig. 9a illustrates a comparison between the model and tests by
by the model when considering the groups B1 to B3. For the last Podgorniak-Stanik [19]. Notice, that for these tests only half the
group, B4, however, the capacity is significantly overestimated amount of longitudinal reinforcement was provided when com-
due to a somewhat surprising but though consistent decrease in pared with the tests of Moody et al., from which the effectiveness
capacity observed in the tests. A subsequent investigation of the factors were determined. In the diagram, the abbreviations BN and
fractured surfaces (the critical crack as well as some ‘‘non- BH refers to beams made from normal- and high-strength concrete,
critical” cracks) revealed a noticeable difference in the crack sur- respectively. Again, the comparison suggests that the effective
face roughness with significantly more fractured aggregates in maximum aggregate-size, and perhaps also the size of w0 , must
the case of the last group with the highest concrete strength. These be reduced in order not to overestimate the capacity in the case

1.50 1.50
a/d=4.52, ρ=1.56%, d=404 mm, d g=8 mm a/d=4.52, ρ=1.56%, d=404 mm, d g=8 mm

1.00 1.00
τu [MPa]

τu [MPa]

B3
B2 B4
B1

0.50 0.50
for f c ≥50 MPa:
model using ν mid
model using ν mid , w0=2.0mm, dg=0mm
mean of group
single test model using ν mid , w0=0.5mm, dg=0mm

0 0
0 10 20 30 40 50 60 70 0 10 20 30 40 50 60 70
fc [MPa] fc [MPa]
(a) (b)
Fig. 8. (a) Comparison without modifications of dg and w0 , (b) comparison including modifications.
226 J. Fisker, L.G. Hagsten / Engineering Structures 115 (2016) 220–231

of higher strength concrete. If doing so, however, good agreement great variations on the failure position can occur, without any sig-
is obtained using the effectiveness-factors. It is interesting to nificant influence on the shear capacity – at least for members that
notice that the capacity of the beam BH50 was virtually unaffected are not lightly reinforced (q J 1:5%). This tendency can be illus-
by the greatly increased compressive strength. trated using the grey shaded regions in the diagrams. The regions
In Fig. 9b, c and d the estimates of the model have been com- indicate failure locations for which the increase in capacity is less
pared with the results of additional test series, and good agreement than or equal to a 5% increase of the capacity at the predicted crit-
is generally obtained. The test shown in Fig. 9b were performed at ical crack position. For unknown reasons the group B1 consistently
Aarhus University [20], among other things, in order to supplement failed in previously developed cracks located closer to the load and
the case of low strength concrete. The estimated capacity is based the actual position is mostly outside the grey-shaded region. The
on the average concrete strength of the four beams, S1 to S4. average capacity of this group was, however, 10% higher than the
predicted (using mmid ), which in turn appears to be in agreement
with the trend of the model.
4.1.1. Final remark on the comparison with tests by Eriksen and
Similar tendencies regarding this apparent insensitivity of the
Pedersen capacity towards the actual failure position has been observed by
In the tests, reported in [16], certain variations on the actual
Mphonde et al., also from test on slender members with a rather
position of the failure were observed. For the groups B2 to B4 these
significant amount of reinforcement. In contrast, a substantial vari-
variations can be summarised by the bounding locations xcr;min and
ation of the capacity was observed for the shorter members in these
xcr;max shown on the pictures in Fig. 10b. There was no apparent tests, and was attributed to the fluctuations of failure position [21].
correlation to the concrete strength. In Fig. 10a the corresponding
predicted position of the critical crack is illustrated. In addition, the
diagram includes the capacity related to a failure developing 4.2. Reinforcement ratio
through neighbouring cracks along the shear span. Initially, these
discrepancies between the predicted and actual positions of failure A great number of tests on beams without shear reinforcement
may be interpreted as a model-limitation. On the other hand, bear- have been performed on members containing very large amount of
ing in mind natural fluctuations of material properties and crack reinforcement [22]. In the following comparison, however, atten-
development, the model seems to support the observation that tion is primarily given to the few tests series including also mem-

2.00 2.00
a/d=4.3, ρ=1.73%, d=364 mm, d g=8 mm
f c =35 MPa f c =95 MPa f c =95 MPa f c =95 MPa avg. f c =22.7 MPa
std=2.8 MPa
1.50 1.50
τu [MPa]

τu [MPa]

1.00 1.00

w 0=2.0mm w 0=2.0mm w 0=2.0mm w 0=1.0mm


d g=10 mm d g= 10 mm d g=0 mm d g=0 mm
0.50 0.50
test tests
model with ν mid model with ν mid

0 0
BN25 BN50 BH25 BH50 BH25 BH50 BH25 BH50 S1 S2 S3 S4
Beams Beams
(a) (b)
2.00 2.00
a/d=4.0, ρ=1.19%, d=270 mm, d g=13 mm a/d=4.0, ρ=2.44%, d=270 mm, d g=13 mm

1.50 d g=0 mm 1.50


w 0=1.0 mm
τu [MPa]

τu [MPa]

d g=0 mm
1.00 1.00 w 0=1.0 mm

0.50 0.50
model using ν mid model using ν mid
test test

0 0
0 10 20 30 40 50 60 70 80 90 0 10 20 30 40 50 60 70 80 90
fc [MPa] fc [MPa]

(c) (d)
Fig. 9. Comparison with tests. (a) Podgorniak-Stanik [19], (b) Kolt and Holdflod [20], (c) and (d) Elzanaty et al. [18].
J. Fisker, L.G. Hagsten / Engineering Structures 115 (2016) 220–231 227

(a) (b)
Fig. 10. Position of critical crack. (a) Model predictions including region due to 5% increase of min V u , (b) xcr;min and xcr;max as observed in tests of groups B2, B3 and B4.

2.0 2.0
a/d=3.78, fc =24 MPa, d=402 mm, d g=25.4 mm a/d=3.4, f c =40 MPa, d=360 mm, d g=19 mm

1.5 1.5
τu [MPa]
τu [MPa]

1.0 1.0

0.5 0.5
model with ν mid model with ν mid
test test

0 0
0 0.4 0.8 1.2 1.6 2.0 2.4 2.8 0 0.4 0.8 1.2 1.6 2.0 2.4 2.8
ρ [%] ρ [%]
(a) (b)
2.0 2.0
a/d=4.0, f c =20 MPa, d=270 mm, d g=13 mm a/d=4.0, f c =40 MPa, d=270 mm, d g=13 mm

1.5 1.5
τu [MPa]

τu [MPa]

1.0 1.0

0.5 0.5
model with ν mid model with ν mid
test test

0 0
0 0.4 0.8 1.2 1.6 2.0 2.4 2.8 0 0.4 0.8 1.2 1.6 2.0 2.4 2.8
ρ [%] ρ [%]
(c) (d)
Fig. 11. Comparison with tests. (a) Mathey and Watstein [23], (b) Tureyen and Frosch [24], (c) and (d) Elzanaty et al. [18].

bers with low or moderate amounts of longitudinal reinforcement. leading to an increase with respect to crack sliding resistance,
A comparison with four different test series is shown in Fig. 11. In and (ii) increase of the compressive zone height and thereby also
all cases, the model seems to reasonably capture the general ten- the resistance towards sliding within the compressive zone. Since
dency observed in the tests. crack sliding represents the dominant contribution, the overall ten-
According to the model the increase in shear capacity due to an dency mainly follows the inverse relation between crack widths
increased amount of longitudinal reinforcement is attributed and the amount of reinforcement, as seen for instance from the
mainly to two different relations; (i) reduction of crack widths expression in (7).
228 J. Fisker, L.G. Hagsten / Engineering Structures 115 (2016) 220–231

2.0 were used due to the large yield strength, the bond-properties of
a/d=4.1, f c =28 MPa, d=265 mm, d g=12.7 mm
which are known to be inferior to traditional deformed bars. As
shown by Leonhardt and Walter, however, this may have a direct
1.5 influence on the shear capacity due to reduced flexural cracking
and activation of ‘‘strut and tie action” [12].
τu [MPa]

1.0 4.3. Size effect

The model correctly predicts a size effect with respect to the


effective height, and the relation rests partly upon the characteris-
0.5
model with ν mid tics of the assumed system of cracks. Since the ratio (y0 /d) is con-
test stant for all other parameters but the effective height being fixed,
the ratio between primary crack-spacing and effective height (scr /
0 d) will therefore also remain fixed. When the effective height is
0 0.4 0.8 1.2 1.6 2.0 2.4
increased the distance between primary cracks will thus be
ρ [%] increased proportionally, and the number of primary cracks will
Fig. 12. Comparison with tests by Rajagopalan and Ferguson [25].
be almost constant. Consequently, greater crack widths will be
experienced for a lower load (relative to the effective height
Fig. 12 illustrates a comparison with tests by Rajagopalan and (V act /d) when compared to a member of less height. The resistance
Ferguson [25]. Unfortunately, the results are rather scattered for from crack sliding therefore reaches a critical magnitude for a
the case of low reinforcement degrees, which may perhaps be lower load.
related to the properties of the reinforcement used in the tests. Fig. 13a shows a comparison between the model and tests by
For these small ratios, prestressing strands (without prestressing) Podgorniak-Stanik [19]. The continuous curves displaying the

1.5
a/d=3.0, f c =35 MPa, d g=10 mm

1.0 0.91%
τu [MPa]

0.89%

0.81%

0.5 0.76%
test
model with ν mid
model with ν mid & w0=1mm

0
0 200 400 600 700 800
d [mm]

(a) (b)
1.5 1.2
a/d=3.0, f c =23 MPa, d g=16 mm
a/d=3.0, f c =24 MPa, ρ=1.35%
1.1 2.4 mm
9 mm
1.0
τu/τu, d=233 mm

1.0 19 mm
ρ=0.83%
τu [MPa]

0.9 38 mm
ρ=0.74%
ρ=0.79% 0.8
0.5
0.7
model with ν mid Model, scaling d g
0.6 Model, constant d g=9 mm
tests

0 0.5
0 200 400 600 700 800 0 200 400 600 800 1000
d [mm] d [mm]
(c) (d)
Fig. 13. Comparison with tests. (a) and (b) Podgorniak-Stanik [19], (c) Walraven [27], (d) Taylor [26].
J. Fisker, L.G. Hagsten / Engineering Structures 115 (2016) 220–231 229

1.50 mid-height have been included. Bearing in mind that the compared
a/d=3.0, f c =38 MPa, d g=10 mm, d=440mm, ρ=0.91%
quantities relates to the – in some aspects – rather random nature
of cracking, the level of agreement is surprisingly good.
Part of the size effect predicted by the model is attributed to the
missing scaling of aggregates. This is illustrated in Fig. 13d, where
1.00
the beam properties, maximum aggregate size and aggregate-
τu [MPa]

scaling have been chosen identical to the case tested by Taylor


[26]. The predicted size effect when scaling the aggregate size is
similar to that observed in the actual test. Conversely, if dg remains
0.50 constant a greater size effect is predicted. This latter scenario was,
however, not tested.
test A final remark on the geometrical size effects regards the influ-
model with ν mid ence of member width, b. Fig. 14 shows a comparison with tests by
Sherwood et al., addressing the influence of member-width [28].
0
0 500 1000 1500 2000 2500 3000 As expected, the increase of width has no influence on the shear
b [mm] capacity, when expressed as a formal shear stress, V u =ðbdÞ. In other
words, the capacity V u is simply proportional to the width, b.
Fig. 14. Comparison with tests by Sherwood et al. [28].

trend from the model have been obtained by interpolation of the 5. Comparison with the Critical Shear Crack Theory
reinforcement ratio in-between the tested values of d. The dashed
line in the figure represents the model estimates in case the value In Fig. 15 the tendencies predicted by the model is compared to
of w0 is lowered to 1 mm as opposed to 2 mm. In Fig. 13b the sys- the Critical Shear Crack Theory (CSCT), proposed by Muttoni and
tem of cracks assumed in the model is compared with the one Ruiz [29]. The model has been proposed on the basis of an evalua-
observed in the tests. The comparison relates to the value of m giv- tion of a great number of tests (>120) [30]. The predictions given
ing the smallest deviation with respect to the measured capacity. by the CSCT are generally contained inside the shaded region
What regards the ratio (scr /d) from the tests, only cracks reaching obtained using mlow and mupp , although the CSCT is seen to predict

1.5 1.5
d=300 mm, fc =35 MPa, ρ=1.0%, d g=16 mm d=300 mm, fc =35 MPa, d g=16 mm, (a/d)=5

1.0 1.0
τu [MPa]

τu [MPa]

0.5 0.5

model with ν mid model with ν mid


CSCT CSCT

0 0
0 1 2 3 4 5 6 7 8 9 0 0.50 1.0 1.50 2.0 2.50
(a/d) ρ [%]
(a) (b)
1.5
1.5 f c =35 MPa, ρ=1.0%, d g=16 mm, (a/d)=5
d=300 mm, ρ=1.0%, d g=16 mm, (a/d)=5

1.0 1.0
τu [MPa]
τu [MPa]

0.5 0.5
model using νmid , dg=0 mm, w 0=2.0 mm
model with ν mid
model using νmid , dg=0 mm, w 0=0.75 mm
CSCT
CSCT
0 0
0 10 20 30 40 50 60 70 80 0 200 400 600 800 1000
fc [MPa] d [mm]

(c) (d)
Fig. 15. Comparison between presented model and CSCT. (a) slenderness, (b) reinforcement ratio, (c) compressive strength, (d) effective height.
230 J. Fisker, L.G. Hagsten / Engineering Structures 115 (2016) 220–231

1.25 Primary domain of existing


shear-tests - no yielding Yielding prior to shear failure

1.00 1.0

0.75
Vu/Vflex

Vu/Vflex
0.50

Model with ν mid


0.25 Test shear fail.
Test flex. fail.

0
0 1 2 3 4 5 6 7 8 9 10 (a/d) min (a/d)
(a/d)
(a) (b)
Fig. 16. (a) Valley of Kani resulting from the tests by Leonhardt and Walter [12], (b) ‘‘outside? Kani‘s Valley. Shear failure may be triggered by plastic hinge formation.

a somewhat greater size effect and a greater increase in shear The model rests upon the observation that shear failures in slender
capacity for larger reinforcement ratios. In both cases this may beams typically evolves from already existing cracks, developed
relate to the influence of dowel action, which is not directly due to the influence of bending, and shaped under the combined
included in the presented model. influence of shear and bending. In the model, this observation is
formalised into a characteristic crack pattern and a characteristic
pattern of failure lines along existing cracks and within plain,
6. Influence of reinforcement yielding
uncracked concrete.
Uncracked concrete as well as existing cracks are identified as
Fig. 16a shows a comparison between the model and tests by
plastic materials, with the fundamental difference being that the
Leonhardt and Walter, addressing the influence of slenderness
capacity of cracks is inversely related to the width. Using the The-
[12]. When plotted relative to the flexural capacity the comparison
ory of Plasticity as basis a shear failure criterion is established and,
reveals the presence of the so-called Kanis Valley, giving a graphical
according to this criterion, the shear capacity is governed by a
presentation of the domain in which the flexural capacity exceeds
sliding-failure, involving sliding along an existing crack and sliding
the shear capacity [1]. In the tests, the most slender beams failed in
along a failure line within uncracked concrete.
flexure due to concrete crushing, without reinforcement yielding.
The comparison with tests revealed the necessity of an effec-
This mode of flexural failure is, however, in contrast to the
tiveness factor. With the effectiveness factor included the model
approach followed in practical design, where members, if possible,
has shown to be in good agreement with test results and allows
are designed to fail in flexure subsequent to the development of
for a rational explanation regarding the influence of important
considerable plastic deformations. Especially in case of statically
design parameters. Bearing in mind that the effectiveness factor
indeterminate structures where rotation capacity is needed in
only includes a minor influence of concrete compressive strength,
terms of plastic hinge development. Examining various databases
the predicted trends with respect to the aforementioned parame-
one will find that the vast majority of ‘‘shear tests” have been con-
ters follows mainly from the model itself.
ducted on members belonging to the region, where the shear
Obviously, the model represents a compromise between reality
capacity governs the magnitude of the load to be applied. Little is
and simplicity, and for this reason a number of assumptions have
therefore known with respect to the behaviour of more moderately
been introduced. Some of them represent reasonable approxima-
reinforced members, in which reinforcement yielding is initiated
tions, while others have been introduced simply due to a lack of
for load levels less than the apparent shear capacity. That is, as
better knowledge. This regard first and foremost the treatment of
illustrated in Fig. 16b, the shear behaviour of members belonging
cracks; the pattern and shape, the assumed kinematics and the
next to the region defined by Kanis valley.
mechanical properties when subjected to shear- and normal dis-
So far, the conditions close to the point of load application
placement. Evidently cracks are not straight with constant inclina-
(xcr K d) have not been discussed in details, and, as long as rein- tion, and the missing correlation between i.e. slenderness and the
forcement yielding is not initiated, the shear resistance in this crack pattern is not entirely consistent with observations. A value
region will not become critical, se e.g. Fig. 4b. The initiation of of b somewhere within the interval 25–35° does, however, appear
yielding and significant plastic steel strains will, however, lead to reasonable, and variations within this interval will not significantly
greater crack widths and reduced compressive zone height, and, influence the results when compared to b = 30°. The assumption of
consequently, a loss of shear resistance inside the hinge. In [6] it straight cracks of a constant width implies that similar conditions
is shown – within the concept of the presented model – how this exist along the entire length of the crack. If the contact between
may trigger a shear failure inside the plastic hinge. The obtained crack-faces is lost, due to an insufficient shear displacement, this
results have shown to be in good agreement with the few tests is not captured by this simplified approach, and can therefore only
investigating this important aspect [31], what regards the load- be included through the effectiveness factor, representing the
and rotational capacity and the position of the failure. influence of aspects not – adequately or at all – addressed in the
model.
7. Discussion and conclusion Another issue that may be unintentionally captured by the
effectiveness factor is the influence of dowel action. That is the
A rational model have been presented for the basic scenario of a unfortunate consequence of fitting. However, in light of the com-
beam element subjected to combined constant shear and bending. parison with tests and also the CSCT it appears that the effective-
J. Fisker, L.G. Hagsten / Engineering Structures 115 (2016) 220–231 231

ness factor should not include any term specifically related to the [9] Nelsen MP, Hoang LC. Limit analysis and concrete plasticity. 3rd ed.. CRC Press;
2010.
longitudinal reinforcement. The bending of dowels could be
[10] Vecchio FJ, Colins MP. The modified compression-field theory for reinforced
included directly by adding the dissipation due to hinge- concrete elements subjected to shear. ACI Struct J 1986;83(2):219–31 [March].
 
formation, and leading to V dow ¼ f y nD =3ða  xcr Þ. The addition [11] Jacobsen JS, Poulsen PN, Olesen JF. Characterization of mixed mode crack
opening in concrete. Mater Struct 2012;45(1–2):107–22.
of this contribution does, however, not introduce any significant [12] Leonhardt F, Walter R. Schubversuche an einfeldigen Stahlbetonbalken mit
changes when q K 1% or if, in case of greater amounts of reinforce- und ohne Schubbewehrung. Deutscher Ausschus für Stahlbeton, heft 151;
ment, the member is not too short. 1962.
[13] Reineck K. Ultimate shear force of structural concrete members without
As shown, the effectiveness factor appears to be applicable for transverse reinforcement derived from a mechanical model. ACI Struct J
both normal- and high strength concrete. For concrete of higher 1991;88(5):592–602 [September].
strength (f c J 50 MPa), however, the comparison with tests shows [14] Moody KG, Viest IM, Elstner RC, Hognestad E. Shear strength of reinforced
concrete beams, Part 1 – Tests of simple beams. ACI J Proc 1954;26(4):317–32.
that a maximum aggregate-size of dg = 0 must be assumed, while a [15] DS/EN 1992-1-1. Eurocode 2.
reduction of w0 to 0.5–1 mm appears reasonable. For normal [16] Eriksen JO, Pedersen JG. Behaviour of shear-critical reinforced concrete beams
strength concrete w0 = 2 mm have been assumed. As mentioned, without transverse reinforcement. Masters Thesis, Aarhus University,
Architectural Engineering; 2014.
these manipulations are likely to reflect the consequence of [17] Sherwood E. One-way shear behaviour of large, lightly-reinforced concrete
increased concrete brittleness, and should in principle therefore beams and slabs. PhD thesis, Department of Civil Engineering, University of
also influence the dissipation with respect to sliding in the plain Toronto; 2008.
[18] Elzanaty AH, Nilson AH, Slate FO. Shear capacity of reinforced concrete beams
concrete in the compressive zone, although this is difficult to
using high-strength concrete. ACI J Proc 1986;83(2):290–6 [March–April].
include directly in the model. From the comparison with tests it [19] Podgorniak-Stanik B. The influence of concrete strength, distribution of
appears, however, that this part is reasonably captured simply by longitudinal reinforcement, amount of transverse reinforcement and
the dependency of the effectiveness factor on concrete strength. member size on shear strength of reinforced concrete members. University
of Toronto; 1998.
In its entirety, the model may not be suitable for practical [20] Kolt A, Holdflod K. Shear in reinforced concrete members without transverse
design. In [6] it is illustrated how a simplified procedure can be reinforcement. Experimental report. Aarhus University, Architectural
established without loss of mechanical transparency. Finally, it is Engineering; 2013.
[21] Mphonde AG, Frantz GC. Shear-tests of high- and low-strength concrete beams
discussed how the presented procedure can be further extended without stirrups. ACI Struct J 1984;81(4):350–7 [July–August].
in order to cover other loading scenarios. [22] Reineck KH, Bentz EC, Fitk B, Kuchma DA, Bayrak O. ACI-DAfStb database of
shear tests on slender reinforced concrete beams without stirrups. ACI Struct J
2013;110(5) [September–October].
References [23] Mathey RG, Watstein D. Shear strength of beams without web reinforcement
containing deformed bars of different yield strengths. ACI Struct J 1963;60
[1] Kani G. Basic facts concerning shear failure. ACI J Proc 1966;63(6):675–92. (2):183–207.
[2] Taylor H. Investigation of the forces carried across cracks in reinforced [24] Tureyen AK, Frosch RJ. Shear tests of FRP-reinforced concrete beams without
concrete beams in shear by interlock of aggregate. Cement and Concrete stirrups. ACI Struct J 2002;99(4):427–34.
Association. Technical report, 42.447; 1970. [25] Rajagopalan KS, Ferguson PM. Exploratory shear tests emphasizing percentage
[3] Fenwick RC, Pauley T. Mechanisms of shear resistance of concrete beams. J of lightweight steel. ACI J Proc 1968;65(8):634–8.
Struct Div ASCE* 1968;94(ST10):2325–50. [26] Taylor HPJ. Shear strength of large beams. J Struct Div 1972;98(ST11):2473–90
[4] Muttoni A. Die Anwendbarkeit der PlastizitÃd’tstheorie in der Bemessung von [November].
Stahlbeton. ETH ZÃijrich, Institut frÃijr Baustatik und Konstruktion, Bericht Nr. [27] Walraven J, Lehwalter N. Size effects in short beams loaded in shear. ACI Struct
176; 1990. J 1994;91(5):585–93 [September].
[5] Campana S, Fernández RM, Anastasi A, Muttoni A. Analysis of shear-transfer [28] Sherwood EG, Lubell AS, Bentz EC, Collins MP. One-way shear strength of thick
actions on one-way RC members based on measured cracking pattern and slabs. ACI Struct J 2006;103(6):794–802 [November–December].
failure kinematics. Mag Concr Res 2013;56(6):386–404. [29] Muttoni A, Ruiz MF. Shear strength of members without transverse
[6] Fisker J. Shear capacity of reinforced concrete beams without shear reinforcement as function of critical shear crack width. ACI Struct J
reinforcement. PhD thesis, Aarhus University, Aarhus, Denmark; 2014. 2008;105(2):163–72 [March–April-.
[7] Dei Poli S, di Prisco M, Gambarova PG. Cover and stirrup effects on the shear [30] Muttoni A. Schubfestigkeit und Durchstanzen von Platten ohne
response of dowel bar embedded in concrete. ACI Struct J 1993;90(4):441–50 Querkraftbewehrung. Beton Stahlbetonbau 2003;98(2):74–84.
[July–August]. [31] Rodrigues RV, Muttni A. Shear strength of RC slab strips without stirrups after
[8] Zhang JP. Strength of cracked concrete. Part 2. Report R, No. 17. Technical yielding. Test report. Laboratoire de Construction en Beton, EPFL, Lausanne;
University of Denmark, Department of Structural Engineering and Materials; 2004.
1997.

Das könnte Ihnen auch gefallen