Received: 22 March 2017 Revised and accepted: 23 July 2017

DOI: 10.1002/suco.201700059

TECHNICAL PAPER

Size effect in shear and punching shear failures of concrete

members without transverse reinforcement: Differences between
statically determinate members and redundant structures
Miguel Fernández Ruiz | Aurelio Muttoni

École Polytechnique Fédérale de Lausanne,

Lausanne, Switzerland Large efforts have been devoted in the past to understanding size effect in shear
Correspondence failures of members without transverse reinforcement. Experimental works have
Miguel Fernández Ruiz, École Polytechnique demonstrated that increasing the size reduces the nominal shear strength provided
Fédérale de Lausanne, School of Architecture,
that the failure mode is brittle. For large specimen sizes where a linear behavior
Civil and Environmental Engineering (ENAC),
Station 18, CH-1015 Lausanne, Switzerland can be assumed between the opening of the shear cracks and the acting shear
Email: [email protected] force, the size effect agrees with linear-elastic fracture mechanics (LEFM). In this
Funding information case, the reduction of nominal shear strength is characterized with respect to the
Swiss Federal Road Administration, Grant/Award specimen size by an asymptotic slope (in double-log scale) of −1/2. However,
number: AGB 2011-015
such a linear relationship between crack widths and applied shear force is not
present in most reinforced concrete structures, where the response is characterized
by a nonlinear behavior at a cracked stage. In these cases, the influence of size
effect shall be milder, and the slope (in double-log scale) characterizing it shall be
lower than −1/2. In this paper, the influence of size effect on structures character-
ized by internal redundancy (continuous one-way members and slabs in case of
punching shear with a nonlinear response) is investigated in the frame of the criti-
cal shear crack theory. The theory predicts the same influence as LEFM with
respect to size effect when a linear behavior can be assumed, but it yields consist-
ently milder influence of size effect for structurally redundant elements with a
nonlinear response.

KEYWORDS

concrete structures, critical shear crack theory, punching shear strength, shear
strength, size effect

1 | INTRODUCTION For cracked beams failing in shear, a large number of

tests has been performed in the past with varying sizes (typ-
The influence of size effect in shear and punching shear ically from 10 cm up to 4.0 m4). The majority of these tests
failures has been a controversial topic of research in the last correspond to simply supported beams subjected to concen-
decades. Although its influence has been acknowledged trated loads (under three- or four-point bending conditions,
experimentally and theoretically, its significance and impli- refer to Figure 1a). The results of these tests have shown
cations for design do not yet find a consensus. For instance, that the influence of size effect is variable, with almost neg-
with respect to shear design in codes, some do not account ligible influence when a strength or yield criterion governs
for this phenomenon (as ACI 318-141), others propose (typically associated to low member sizes), but quite signifi-
empirical factors (as Eurocode 22) and some calculate it on cant in case of a sudden and brittle crack propagation (typi-
the basis of mechanical models (as MC20103). cally associated with large member sizes).4 For brittle

Structural Concrete. 2017;1–11. wileyonlinelibrary.com/journal/suco © 2017 fib. International Federation for Structural Concrete 1
2

FERNANDEZ RUIZ AND MUTTONI

(a) (b)
2V log ṽ LEFM
limit analysis

b
V V

log d˜

(c) (d)

(f)

(e)

FIGURE 1 Shear and punching shear failures: (a) classical academic (statically determinate) beam shear test; (b) corresponding size-effect law;
(c) redundant structure in the transversal direction (cut-and-cover tunnel); (d) redundant structure in the longitudinal direction (bridge deck slab); (e) slab
sector; and (f ) continuous flat slab

failures with significant size-effect influence, the behavior effect law (according to Bažant et al6). The validity of this
of test specimens is reasonably well approximated by slope for statically determinate members can also be demon-
assuming a linear relationship between the acting (shear) strated by other means as for instance the critical shear
force and the crack widths (see Reference 5 experimental crack theory (CSCT,5 a consistent study on this topic can be
evidence will be discussed in this paper). By using linear- referred elsewhere7).
elastic fracture mechanics (LEFM), the influence of size As it will be shown in this paper, the observed size
effect on the strength of asymptotically large specimens can effect for statically determinate beams is not necessarily
be calculated,6 corresponding to a slope (in double-log representative of actual members failing in shear without
scale) of −1/2, refer to Figure 1b (plotted for normalized transverse reinforcement. These are mostly slabs, with the
values for the size and shear strength parameters, values for capacity to act as two-way members and in many cases
such normalization will be introduced in the next section). with a redundant static system, refer to Figure 1c,d. For
The transition between the two regimes (controlled by a these members, redistribution of internal forces occur after
yield criterion or controlled by LEFM) can be characterized cracking and reinforcement yielding, and the linearity
by a smooth curve that is normally referred to as the size- between acting shear forces and crack widths can be


FERNANDEZ RUIZ AND MUTTONI 3

considered as no longer applicable (neither the size effect 2.1 | The capacity of cracked concrete to transfer
influence predicted by LEFM). Theoretically, the slope shear forces
shall be milder, with an influence depending on the devia- According to the CSCT, the capacity of cracked concrete to
tion with respect to a linear behavior. Despite this fact, the transfer shear forces is characterized by a failure criterion
influence of size effect on shear design formulas has been accounting for the various potential shear-transfer actions7
adapted in several cases from that of statically determinate
(namely aggregate interlock, residual tensile strength of con-
members. crete, dowelling action, and inclination of compression
For punching shear failures, the influence of the non- chord). The failure criterion results from integration of the
linear response on the size effect is even more significant
stresses developed at the critical shear crack accounting for
as two-way slabs are internally redundant members with its shape and kinematics at failure. It can be demonstrated
significant internal force redistributions (in radial and tan- in References 5,7 that the shear failure criterion is in fact
gential directions and between hogging and sagging dependent on the opening (w) and roughness of the critical
regions,8 Figure 1f ) and where the effects of tension stiff- shear crack leading to failure. In addition, it can be demon-
ening are very marked.9 Even for slab sectors (Figure 1e), strated that the capacity of cracked concrete to transfer shear
the response (shear force-crack width) is notably nonlin- forces decreases for increasing crack openings or decreasing
ear due to these aspects (tension-stiffening and radial- crack roughness.
tangential moment redistribution due to cracking and For practical purposes, the failure criterion can be
yielding9). defined by a single function where it is assumed that the
Previous investigations based on analysis of experimen-
critical shear crack width is proportional to a reference
tal results have already shown that the asymptotic slope for strain times the effective depth of the member (w / ε  d 5)
size effect in punching failures governed by brittle failures leading to the following failure criterion:
is milder than −1/2 (expressed in double-log scale), indicat-
ing that the behavior deviates from that of LEFM.10 This VR 1 1
pffiffiffiffi =  ðSI units ½MPa, mmÞ; ð1Þ
phenomenon can be demonstrated theoretically by using the bd fc 3 1 + 120 dεd
dg
CSCT9,11 (ground of current MC2010 punching provi-
where b refers to the width of the member, d to its effective
sions12,13), which yields an influence of size effect on the
depth, fc to the compressive strength of concrete measured
punching strength governed by a slope closer to −1/3.14
in cylinder, and ddg to the equivalent crack roughness
This value is also in agreement to other theoretical
(ddg = dg + 16 mm ≤ 40 mm, where dg refers to the maxi-
approaches for punching shear.15
mum aggregate size). With respect to the reference strain
This paper reviews these aspects by identifying and jus-
(ε), it is considered at a distance equal to 0.6d of the com-
tifying based on experimental evidence the cases where the
pression face and at d/2 of the applied load (location of the
behavior of brittle failures in shear can be approximated by
control section, extended considerations can be consulted
LEFM and the cases where the actual behavior deviates
elsewhere5). It can be noted that in the previous expression,
from it. Based on the theoretical frame of the CSCT, suita-
the crack width accounts both for the size (d) and for the
ble values for the influence of size effect are derived in
strain of the member (ε), thus coupling both phenomena.
terms of the slope describing brittle failures (expressed in
double-log scale). On that basis, recommendations are pro-
vided for design.
2.2 | Statically determinate members—Size effect in
shear for a linear structural response
2 | SIZE E FFE CT IN SHEAR F AILURE S O F The shear strength and the influence of size in statically
O NE -W A Y S L A B S W I T H O UT TR AN S V E RS E determinate members (for instance simply supported beams
R E I NF O R CE M E N T and one-way slabs) has been previously investigated in
detail by means of the CSCT.7 The strength can be calcu-
The influence of size effect on the shear strength of one- lated by intersecting the failure criterion by the law describ-
way slabs and beams without transverse reinforcement is ing the relationship between the acting shear force at the
investigated in this section both for statically determinate control section and the associated crack openings.
(as for instance simply supported beams, as usually tested For statically determinate members, it can be accepted
in laboratory) and redundant members (as continuous one- that the reference strain (ε) and thus the opening of the cracks
way members, which are more representative of practical (w) are linearly dependent on the acting bending moment.5
cases). The aim of this section is to highlight their differ- This assumption is grounded on the reduction and eventual
ences and expected consequences with reference to size loss of reinforcement-to-concrete bond stresses due to the
effect. To that purpose, the theoretical frame of the CSCT is development of horizontal or inclined cracks at the level of
used to analyze the shear strength and to perform the the reinforcement7 limiting the tension-stiffening effects. The
comparisons. development of these cracks (eventually leading to a
4

FERNANDEZ RUIZ AND MUTTONI

delamination crack) is investigated in detail by Cavagnis On the basis of these considerations, a linear elastic
et al16 (named as type “D” cracks). This fact is shown in cracked behavior can be reasonably assumed. In this case,
Figure 2 showing the results of a specimen (SC70) tested and for the depth of the reference fiber, the strains can be
instrumented consistently with the results presented in Refer- approximated by the following expression (neglecting
ence 16. In that figure, the horizontal opening of the crack at tension-stiffening effects in the sectional analysis, Figure 3):
point “A” is investigated. This point is located at the same
M 0:6d −c
depth as that considered by the CSCT as a reference fiber ε= : ð2Þ
ρbdEs ðd − c=3Þ d − c
(where flexural cracks have already merged). Figure 2b–d
shows the evolution of the crack pattern as well as of the hor- For a simply supported beam subjected to a concen-
izontal opening of the crack. At early cracking stages trated load, the bending moment at the control
(Figure 2b), the behavior is not linear with a rapid increase of section (located at d/2 from the load, see Figure 3b) results
the crack opening. Thereafter (Figure 2c), the relationship M = VR(a − d/2) so that:
between the crack widths and the acting shear force is rather
VR ða −d=2Þ 0:6d −c
linear (with proportional increases for both). This behavior ε= ; ð3Þ
ρbdEs ðd − c=3Þ d − c
continues up to near failure (Figure 2d) when, in the last load
increments (Figure 2e), the linearity is again lost.

a = 3850 mm V
(a)
2Ø28 mm
A
0.6d
M d 600 mm

2Ø28 mm

V 250 mm
(b)
120
100

V [kN]
80
M 60
40
20
0
0 0.1 0.2 0.3 0.4 0.5
V uA [mm]
(c)
120
100
V [kN]

80
M 60
40
20
0
V 0 0.1 0.2 0.3 0.4 0.5
uA [mm]
(d)
120
100
V [kN]

80
M 60
40
20
0
V 0 0.1 0.2 0.3 0.4 0.5
uA [mm]
(e)
120
100
V [kN]

80
M 60
40
20
0
0 0.1 0.2 0.3 0.4 0.5
uA [mm]

FIGURE 2 Analysis of the response of a simply supported beam: (a) geometry, support and loading conditions of specimen SC70; and horizontal crack
opening at point A (uA) for: (b) an early cracking stage; (c) stabilized cracking and development of type D cracks; (d) near failure; and (e) failure


FERNANDEZ RUIZ AND MUTTONI 5

(a) (b)
V control section
b·d √ fc

reinforcement yielding

VR d/2
b·d √ f c
a

failure criterion
linear response 0.6d
ε·d
d dg reference fibre
(c)
VR
log b·d √ fc
0
LEFM limit analysis
-0.4
FIGURE 3 Size effect in shear failures:
(a) failure criterion, flexural behavior and -0.8 2
failure load according to the critical shear 1
crack theory (CSCT); (b) control section and -1.2
reference strain; and (c) size-effect law CSCT size-effect law
according to the CSCT and comparison to the -1.6
database of Reference 5 -1.5 -1 -0.5 0 0.5 1 1.5 2 log dn

where ρ refers to the reinforcement ratio, Es to the modulus in the transversal direction (Figure 1c) as well as in the
of elasticity of the reinforcement, and c refers to the depth longitudinal direction when concentrated loads are acting
of the compression zone5: (Figure 1d). Due to structural redundancy, redistributions
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ! of internal shear forces are possible, which may deviate
Es 2Ec the relationship between the acting shear force and the
c = ρd 1+ −1 ; ð4Þ
Ec ρEs crack openings at the shear-critical region from a
linear one.
and with Ec referring to the modulus of elasticity of
The response of a redundant structure is for instance
concrete.
shown in Figure 4. It corresponds to a continuous one-way
By introducing the value of ε of Equation (2) into
slab subjected to distributed loading, where the response of
Equation (1), the shear resistance (VR) can be determined,
such element is investigated at d/2 (control section) from
which results in the following expression7:
the support region (Figure 4a). Some phases can be differ-
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi entiated, corresponding first to an uncracked behavior, fol-
VR − 1 + 1 + 4dn
pffiffiffiffi = ; ð5Þ lowed by cracking near the support, cracking at mid-span
bd fc 6dn
and eventually yielding of the reinforcement of both
where dn is a parameter accounting for size and strain regions. The evolution of the bending moments at mid-span
effects.7 It can be noted that this expression is consistent and support (Figure 4b) shows that the response is not lin-
with the size-effect law for shear failures by Bažant.6 The ear, deviating due to cracking and yielding occurring at dif-
validity of the expression can in fact be assessed by means ferent load levels at mid-span and support. This influences
of test results as most tests of beams in shear correspond to the response of the element in terms of the shear force-to-
this support and loading conditions, see Figure 3c. For fail- crack opening relationship, leading to a nonlinear behavior.
ures governed by limit analysis (dn ! 0), no size effect is The shear strength can be determined by calculating the ref-
observed whereas for large sizes (dn ! ∞), the slope (in erence strain by means of Equation (2) and by accounting
double-log scale) of the strength corresponds asymptotically for the actual values of the bending moment and shear force
to −1/2 in accordance with the LEFM (Figure 3c). (Figure 4b). In so doing, a nonlinear response is obtained
with a stiffer behavior than that corresponding to a linear
prediction (two calculated cases are shown in Figure 4c). If
2.3 | Size effect in shear for redundant members such analysis is performed for geometrically identical mem-
Despite the fact that the majority of shear tests are per- bers but scaled, one can derive the size-effect law. Two
formed in statically determinate members, actual struc- cases are for instance presented in Figure 4d, one corre-
tures where shear strength governs are in many cases sponding to a moderate flexural reinforcement ratio (ρhog =
redundant. Structural redundancy may be considered both 0.5%) and another to a high flexural reinforcement ratio
6

FERNANDEZ RUIZ AND MUTTONI

(a) (b)
elastic analysis (support)

m
shear control section actual behaviour (support)
q
ρhog m −R

elastic analysis (mid-span)

m +R
actual behaviour (mid-span)
d

ρsag
d/2 m cr
q
(c) L qr1 qr2 qy qR
v
d √ fc
(d) 1

ρhog = 1.0%
elastic prediction ≈3

[ √MPa]
1
actual behaviour
0.1

d·√f c
failure criterion

vR
(thin slabs) ρhog = 0.5%

0.01
failure criterion 0.1 1 10
(thick slabs)
d [m]
ε

FIGURE 4 Influence of size effect on a redundant member: (a) case investigated; (b) evolution of bending moments on the support and mid-span sections;
(c) load-reference strain response; and (d) calculated size effect as a function of the flexural reinforcement ratio plotted in double-log scale (red curve:
ρhog = 1%, ρsag = 0.5%; blue curve: ρhog = 0.5%, ρsag = 0.3%)

(ρhog = 1.0%). From Figure 4d, one can observe that the 3 | S IZ E A ND STR AI N EFF EC TS I N
slope of the size-effect law tends to zero for low sizes (gov- P U NC HI N G F AI L U RE S O F S L AB –C O L U M N
erned by the strength criterion) and that for large size mem- CONNECTIONS
bers it tends to a value of approximately −1/3. This latter
slope is milder than that predicted by LEFM (−1/2) due to 3.1 | Nonlinear response of two-way slabs and shear
the nonlinear response exhibited by the members. The fact strength
that the slope is milder than that of LEFM indicates that the
assumption on the linear behavior for a redundant structure The CSCT can also be used as a consistent frame for investi-
deviates from its actual behavior and leads to safe predic- gating the punching shear strength of two-way slabs.9 With
tions in terms of the shear strength. reference to the opening of the critical shear crack, it is
A similar consideration can be performed with respect assumed to be correlated to the slab rotation (ψ, see Figure 5b)
to structures that are redundant in the transversal direction times the effective depth of the member: w / ψ  d. This term
(due to a two-way slab behavior, refer for instance to the accounts again for both the size and strain level of the mem-
ber. Similarly to the shear strength of one-way slabs, the fail-
deck slab subjected to a concentrated action shown in
Figure 1d). Detailed measurements performed by Natário ure criterion can be described according to Muttoni9 by
et al17 have shown that strong redistributions in the bending a hyperbolic relationship:
and shear fields occur in the region where shear failure
VR 3=4
occurs, with load being transferred to less cracked regions. pffiffiffiffi = ðSI units ½MPa, mmÞ; ð6Þ
Also, crack openings are shown to have a highly nonlinear b0 d fc 1 + 15 ψd
ddg

response due to the progression of the critical shear crack

up to failure. These observations show that the structural where b0 refers to the length of the control perimeter
redundancy due to a two-way slab behavior is also a poten- (located at d/2 of the edge of the supported area). As for
tial source of loss of linear response and shall thus mitigate shear failures, the shear strength decreases for increasing
size-effect influence (corresponding to milder slopes of the crack openings (or decreasing crack roughness ddg).
size-effect law for large sizes). More details on the behavior The shear strength is calculated (in a similar manner as
of two-way slabs will be presented in the following for shear failures) by intersecting the punching failure crite-
section with reference to the punching shear strength of rion with the load–rotation response of the slab. With respect
slab–column connections. to the latter, the behavior is markedly nonlinear due to the


FERNANDEZ RUIZ AND MUTTONI 7

(a)
−m
reinforcement yielding
(flexural strength)
− 1r
actual response
tension-stiffening
−m
EI 1 no concrete contribution in tension
1
− 1r

(b) V full yielding of reinforcement

(flexural strength)
(4)
ψ
reinforcemen t
partly yielded (3)
FIGURE 5 Realistic behavior of a slab: V
(a) moment (m)–curvature (1/r) diagram of a
reinforcemen t actual resp onse
slab element with tension-stiffening effects; and (2)
elastic
(b) load–rotation relationship of a slab–column tension-stiffening
connection with different regimes (calculated
uncracked (1) no concrete contribution in tension
assuming a quadri-linear moment–curvature
diagram9) ψ

distribution of bending moments in the slab (in the radial and reinforcement at mid-span and the moment redistributions
tangential directions9) as well as to the moment–curvature that occur.8,18
response of a reinforced concrete section (simplified as a The significance of the tension-stiffening effects can be
quadri-linear response in Figure 5a). observed in Figure 6 with reference to two tests performed
A detailed analysis and derivation of the load–rotation by Guandalini et al19 on geometrically identical specimens
response of slab–column connections has been thoroughly scaled approximately 1:2. Although both specimens failed
presented by Muttoni9 (extended considerations accounting in punching, the smaller exhibited a behavior with a signifi-
for the continuity of flat slabs have also been developed by cantly larger deformation capacity, failing in regime
Einpaul et al8,18). According to these works, the load– (3) (almost beginning of regime (4)). The thicker specimen,
rotation relationship (Figure 5b) comprises a number of however, showed a rather brittle behavior and failed in the
regimes: transition between regimes (2) and (3). Both specimens

• First, the slab behaves in an uncracked regime (linear- 0.8

elastic regime (1) in Figure 5b)
• Second, cracking develops in the member with all rein-
forcements in the elastic regime ((2) in Figure 5b).
• Third, local yielding of the reinforcement occurs in the 0.6
Test PG-3
ρ = 0.33%
vicinity of the column ((3) in Figure 5b) but still a part d = 456 mm

of the reinforcement remains elastic.

[√MPa]

• Fourth, all reinforcement is yielded, and the slab eventu-

Test PG-10
ally reaches the flexural strength ((4) in Figure 5b) 0.4 ρ = 0.33%
d = 210 mm
b0·d√f c
V

The complete load–rotation curve describing the

response of the member can be observed to be highly non- failure criterion PG10
0.2
linear, with a progressive loss of stiffness as load increases. d = 210 mm

The response is in fact to a large extent controlled by

tension-stiffening effects. In the radial direction, tension
failure criterion PG3, d = 456 mm
stiffening can near failure be reduced due to a delamination 0.0
crack as for one-way slabs. Nevertheless, in the tangential 0.000 0.020 0.040 0.060
ψ [-]
direction, tension stiffening is very relevant as it affects the
whole slab surface and develops until failure. In actual (con- FIGURE 6 Experimental response of thick and thin slabs for
tinuous) flat slabs, the response of the member (character- geometrically scaled specimens (tests PG-3 and PG-10 from Reference 19)
ized by its load–rotation curve) is also influenced by the and comparison to predicted flexural behavior and failure criterion
8

FERNANDEZ RUIZ AND MUTTONI

showed a markedly nonlinear behavior and their response 4 | I M P L I C A T I ON S F OR DE S I GN—

was characterized by the tension-stiffening effects, closely A P P R O A C H O F F IB M C 2 0 1 0
following the theoretical prediction described in Reference
9 (refer to Figure 6). It is also interesting to observe that the fib MC20103 punching shear provisions are based on the
size effect does not only influence the strength and deforma- CSCT13 by incorporating the so-called levels-of-
tion capacity, but even the failure mode. approximation (LoA) approach.20 To that aim, a simplified
As a consequence of the nonlinear behavior of slabs, it load–rotation relationship is used (based on a number of
can be stated that the size effect cannot be described for assumptions justified in Reference13) leading to the follow-
large specimen sizes by the LEFM (as for simply supported ing expression:
beams without transverse reinforcement failing in shear).  
For these cases, the dependence of the strength with respect rs fy ms 3=2
ψ = 1:5 ; ð7Þ
to the size cannot correspond to a slope −1/2 (in double-log d E s mR
scale), but shall be milder (see also the next section).
where rs refers to the distance from the center of the column
to the line of contraflexure of bending moments, fy to the
3.2 | Size effect in punching failures according to yield strength of the reinforcement, Es to its modulus of
the CSCT elasticity, ms to the average bending moment acting in the
The CSCT can be used as previously explained in order to support strip of the slab, and mR to the average flexural
predict the influence of size on the punching shear strength. strength in the support strip of the slab.
For instance, Figure 7a plots the predictions of this theory The parabola of Equation (7) approximates with a single
compared to the experimental programme of Guandalini expression the nonlinear response of an actual slab (see
et al19 (geometrically identical specimens but scaled with Figure 7c-right) enhancing the easiness of use of the
d = 100456 mm). The influence of size effect is observed approach. The coefficients of the failure criterion
to be variable consistently with the size-effect law. For low (Equation (6)) are also slightly adapted to account for the
sizes, the size-effect influence is limited as the behavior is scatter of test results13 and in order to provide a suitable
mostly controlled by the strength criterion. For large speci- level of safety for design.
men sizes, the influence of size becomes more significant. Design can be performed following various LoA to esti-
With respect to the slope, the tension-stiffening effects and mate the punching strength of a slab–column connection.
the nonlinearity of the load–rotation relationship do not This allows performing simple and safe designs for non-
yield to the asymptotic slope predicted by LEFM (−1/2 in critical members, but the accuracy can be refined upon
double-log scale) but to a milder value (approximately −1/3 necessity by a more accurate evaluation of the physical
in double-log scale), refer to Figure 7a. The slope depends parameters implied.
in addition on the amount of flexural reinforcement, as this The most simple LoA proposed in MC2010 (LoA I) is
parameter influences the plastic deformations at failure and aimed at verifying the dimensions of the slab–column con-
different regimes of the load–rotation curve may govern nection. This LoA assumes yielding of the flexural rein-
(refer to Figure 5b). forcement prior to punching (ms = mR) so to ensure a
The significant role of tension stiffening in this response ductile behavior of the connection. Additionally, this con-
can be observed in Figure 7b, where the size effect is calcu- sideration provides a safe estimate of the actual strength
lated assuming no contribution of concrete in tension lead- (as connections may be over-reinforced in bending to
ing thus to a linear response of the slab before yielding of increase their punching strength).
For LoA II and higher, the value of ms is to be calcu-
the reinforcement. In this case, the slope is steeper, corre-
lated as a function of the acting shear force. This allows
sponding (exactly) to −1/2 in agreement to the LEFM pre-
determining the punching strength in case the reinforcement
diction (the mathematical derivation of this value can be
in the support strip is not fully yielded and thus accounting
consulted elsewhere14). This confirms that the CSCT is con- for the potential gain in punching strength when flexural
sistent to LEFM provided that the response of the member reinforcement is not fully yielded (corresponding to lower
is linear. However, as previously shown (refer to Figure 6 crack widths). Figure 7c plots the predicted punching
for instance), the actual behavior of two-way slabs is not strength by using LoA II for the same cases as previously
linear and thus the size-effect influence has to be milder investigated with the CSCT (Figure 7a). The calculated size
than that of LEFM. effect is observed similar to the one derived from the CSCT
The value −1/3 obtained for the influence of size in accounting for a more realistic load–rotation relationship
punching failures is interestingly similar to the one previ- (Figure 7a), with an slope in double-log scale of approxi-
ously calculated for redundant one-way slabs, showing the mately −1/3 for large members (asymptotic value of −1/
significance of the structural response and tension stiffening 2.5). This slope is again milder than that of LEFM and con-
on this phenomenon. sistent with the detailed approach of the CSCT.


FERNANDEZ RUIZ AND MUTTONI 9

FIGURE 7 Predicted size effect by the critical shear crack theory (CSCT) and comparison to the test series presented in Reference 19 plotted in double-log
scale (tests PG-2b and PG-9 for ρ = 0.24%; tests PG-3 and PG-10 for ρ = 0.33%, tests PG-7 and PG-19 for ρ = 0.75% and PG-1 and PG-6 for ρ = 1.50%):
(a) accounting for tension-stiffening effects; (b) without consideration of tension-stiffening effects; and (c) according to fib MC2010

It is also interesting to observe that in Figure 7c, a 5 | C O NC L U S IO N S

dashed line is plotted indicating the condition at which all
flexural reinforcement yields at punching failure (regime This paper investigates the influence of size effect for fail-
(4) governing). This dashed line represents actually the con- ures in shear and punching shear. The investigation high-
dition of LoA I, as slabs designed complying with LoA I lights the differences between statically determinate and
ensure reaching their flexural capacity before failing in redundant members, accounting for the loss of linearity
punching. Size effect is much stronger in this case (slope between the acting shear force and crack widths in the lat-
−1/1) as, for increasing sizes, the nominal punching shear ter. Its main conclusions are summarized below:
strength decreases (refer for instance to Figure 6) and thus
to ensure full yielding of the flexural reinforcement the rein- 1. The significance of size effect depends not only upon
forcement ratio of the slab shall be reduced accordingly. the size of the member (low significance for low sizes
Consequently, slabs designed using LoA I cannot be inves- and high significance for large sizes), but also on the
tigated for a size effect, but for the influence of size on the capacity of a structure to redistribute internal forces due
transition from failures governed by the shear capacity to to cracking and reinforcement yielding.
failures governed by bending capacity (as not only the size 2. Classical tests on statically determinate beams provide a
but also the flexural reinforcement ratio varies, and they size-effect influence consistent to LEFM for asymptoti-
present a combined size and strain effect). cally large sizes (slope of −1/2 in double-log scale).
10

FERNANDEZ RUIZ AND MUTTONI

However, size effect shall deviate from LEFM in actual rs distance from center of column to the line of contra-
(mostly redundant) structures as the shear action and flexure of bending moments
shear resistance (proportional to crack widths and thus uA horizontal opening of crack at point A
to bending moments) are no longer proportional. v unitary shear force
3. It has been observed experimentally that size effect in w crack width
punching of slab–column connections (which are inter- ε reference strain
nally redundant members) is milder than that predicted ρ reinforcement ratio
by LEFM. This is in agreement to the previous consid- ψ rotation
eration and can be theoretically explained by the CSCT.
4. According to the CSCT, the predicted size-effect slope
for redundant members is closer to −1/3 (instead of −1/ REFERENCES

2). This indicates that design on the basis of size-effect 1. American Concrete Institute Committee 318. Building Code Requirements
for Structural Concrete ACI 318-14. Farmington Hills, MI: American Con-
factors calibrated for shear failures in statically determi- crete Institute; 2014:519.
nate members may be too safe for actual structures. 2. CEN European Committee for Standardization. Eurocode 2. Design of Con-
5. fib MC2010 design provisions for punching already crete Structures—General Rules and Rules for Buildings, EN 1992-1-1.
Brussels, Belgium; 2004:225.
account for a nonlinear behavior in the flexural response
3. Fédération internationale du béton. fib Model Code for Concrete Structures
of a slab and the resulting slope for the size effect (in - 2010. Germany: Ernst & Sohn; 2013:434.
double-log scale) is similar to −1/3, in consistency with 4. Collins MP, Bentz EC, Quach PT, Proestos GT. The challenge of pre-
the CSCT and the previous theoretical considerations. dicting the shear strength of very thick slabs. Concr Int.
2015;37(11):29-37.
5. Muttoni A, Fernández Ruiz M. Shear strength of members without trans-
verse reinforcement as function of critical shear crack width. ACI Struct J.
ACKNOWLEDGMENTS 2008;105(2):163-172.
This work was supported by the Swiss Federal Road 6. Bažant ZP, Kim J-K. Size effect in shear failure of longitudinally reinforced
beams. ACI J Proc. 1984;81(5):456-468.
Administration (research grant AGB 2011–015 “Shear 7. Fernández Ruiz M, Muttoni A, Sagaseta J. Shear strength of concrete
strength of slabs and cut-and-cover tunnels”) members without transverse reinforcement: a mechanical approach to
consistently account for size and strain effects. Eng Struct.
2015;99:360-372.
8. Einpaul J, Fernández Ruiz M, Muttoni A. Influence of moment redistribu-
N OT AT I O N tion and compressive membrane action on punching strength of flat slabs.
EC MODULUS OF ELASTICITY OF CONCRETE Eng Struct. 2015;86:43-57.
Es modulus of elasticity of reinforcement 9. Muttoni A. Punching shear strength of reinforced concrete slabs without
transverse reinforcement. ACI Struct J. 2008;105(4):440-450.
EI1 uncracked flexural stiffness 10. Dönmez A, Bažant ZP. Size Effect on Punching Shear Strength of Rein-
M bending moment forced Concrete Slabs Without and With Shear Reinforcement, Report
V shear force No. 15-12/936s, Department of Civil and Environmental Engineering,
Northwestern University, IL; 2016, 23.
VR shear strength 11. Muttoni A, Fernández Ruiz M. The critical shear crack theory for punching
a shear span design: from a mechanical model to closed-form design expressions, Joint
b width of beam ACI-fib International Symposium “Punching Shear of Structural Concrete
Slabs”, Philadelphia, PA, October 23–27, 2016.
b0 control perimeter length
12. Muttoni A, Fernández Ruiz M. MC2010: the critical shear crack theory as a
c thickness of compression zone mechanical model for punching shear design and its application to code
d effective depth (distance from the centroid of the provisions. In: fib Federation for Structural Concrete, Bulletin 57;
flexural reinforcement to the outermost compressed 2010:31-60.
13. Muttoni A, Fernández Ruiz M, Bentz E, Foster SJ, Sigrist V. Background
fiber) to the Model Code 2010 shear provisions—part II punching shear. Struct
~
d normalized effective depth Concr. 2013;14(3):195-203.
14. Fernández Ruiz M, Muttoni A. Size effect on punching shear strength: dif-
ddg equivalent crack roughness
ferences and analogies with shear in one-way slabs. In: fib Federation for
dg maximum aggregate size Structural Concrete, Bulletin 81; 2017:59-72.
dn dimension parameter 15. Broms CE. Tangential strain theory for punching failure of flat slabs. ACI
fc concrete compressive strength measured in cylinder Struct J. 2016;113(1):95-104.
16. Cavagnis F, Fernández Ruiz M, Muttoni A. Shear failures in reinforced
fy yield strength of the reinforcement steel concrete members without transverse reinforcement: a critical analysis on
m unitary bending moment the basis of test results. Eng Struct. 2015;103:157-173.
mcr unitary bending moment at cracking 17. Natário F, Fernández Ruiz M, Muttoni A. Shear strength of RC slabs under
concentrated loads near linear supports. Eng Struct. 2014;76:10-23.
ms average bending moment in the support strip of 18. Einpaul J, Ospina CE, Fernández Ruiz M, Muttoni A. Punching shear
a slab capacity of continuous slabs. ACI Struct J. 2016;113(4):861-872.
mR average flexural strength in the support strip of a slab 19. Guandalini S, Burdet O, Muttoni A. Punching tests of slabs with low rein-
~v normalized shear force forcement ratios. ACI Struct J. 2009;106(1):87-95.
20. Muttoni A, Fernández Ruiz M. The levels-of-approximation approach in
q distributed load MC 2010: applications to punching shear provisions. Struct Concr.
r radius of curvature 2012;13(1):32-41.


FERNANDEZ RUIZ AND MUTTONI 11

AUTHOR'S BIOGRAPHIES
How to cite this article: Fernández Ruiz M,
Miguel Fernández Ruiz PhD Muttoni A. Size effect in shear and punching shear fail-
Senior Lecturer ures of concrete members without transverse reinforce-
École Polytechnique Fédérale de ment: Differences between statically determinate
Lausanne members and redundant structures. Structural Concrete.
Station 18, CH-1015 Lausanne 2017;1–11. https://doi.org/10.1002/suco.201700059
Switzerland
[email protected]

Aurelio Muttoni PhD,

Professor
École Polytechnique Fédérale de
Lausanne
Station 18, CH-1015 Lausanne
Switzerland [email protected]