Engineering Structures 90 (2015) 130–142

Contents lists available at ScienceDirect

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

Effect of load history on punching shear resistance of flat slabs

Robert Koppitz a, Albin Kenel b, Thomas Keller a,⇑
a
Composite Construction Laboratory (CCLab), École Polytechnique Fédérale de Lausanne (EPFL), Station 16, 1015 Lausanne, Switzerland
b
Institute for Civil and Environmental Engineering (IBU), HSR Rapperswil University of Applied Sciences, Oberseestrasse 10, 8640 Rapperswil, Switzerland

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

Article history: The unloading of reinforced concrete slabs results in residual slab rotations and reloading to the same
Received 9 October 2014 load results in irreversible rotation increases. Unloading and reloading (UR) cycles applied to
Revised 13 January 2015 non-strengthened and strengthened flat slabs may thus affect the punching resistance, which is
Accepted 16 February 2015
rotation-dependent. A quintilinear moment–curvature relationship, which takes concrete softening and
tension stiffening into account, combined with UR cycles, modeled as bilinear envelopes, is developed
to predict residual slab rotations and irreversible rotation increases. A parametric study shows that the
Keywords:
effect of UR cycles on the punching resistance of concrete is normally small, however, it may be sig-
Reinforced concrete
Flat slabs
nificant if the slab is strengthened after unloading, particularly for thin and low-reinforced slabs, which
Punching shear exhibited plastic slab rotations before unloading. Prestressing of the strengthening system may reduce
Load history the residual slab rotations and thus limit or compensate the loss of punching resistance.
Strengthening Ó 2015 Elsevier Ltd. All rights reserved.
Tension Chord Model
Fictitious Crack Model

1. Introduction wcr. This behavior is considered by the Fictitious Crack Model

In contrast to the bending resistance, the punching shear resis-
tance of flat slabs depends on the slab rotation [1]. For slabs with
interior columns, the punching resistance can be estimated with
analytical formulas, which are based on the assumption of a rota-
tion-symmetric slab cutout. Load–rotation curves can be derived
for different, e.g. bilinear or quadrilinear, moment–curvature rela-
tionships [2]. Based on the Tension Chord Model (TCM) [3], the
quadrilinear relationship takes a contribution of the concrete ten-
sile stresses between the cracks into account. This leads to a
decrease of the average steel stresses in the cracked zone and thus
to the so-called ‘‘tension stiffening effect’’, which may influence the
punching resistance of slabs with low reinforcement ratios of the
longitudinal reinforcement [2].
In addition to the consideration of concrete tensile stresses
between cracks, the behavior of concrete in the crack itself was
also investigated. Studies on cement pastes [4,5] showed that their
crystalline structure can hinder or even arrest the growth of micro-
cracks by means of interlocking fibers growing out from the
cement grains. Thus the concrete tensile stresses do not abruptly
drop to zero after tensile strength is reached, but still provide a
fractional contribution up to a critical (fictitious) crack width,
⇑ Corresponding author. Tel.: +41 (21) 693 32 26; fax: +41 (21) 693 62 40.
E-mail addresses: robert.koppitz@epfl.ch (R. Koppitz), albin.kenel@hsr.ch
(A. Kenel), thomas.keller@epfl.ch (T. Keller).
(FCM) [6,7] for instance, where one sharp crack of zero initial
length is assumed, or by the blunt Crack Band Model (CBM) [8,9],
which smears the crack over a fracture process zone of a certain
width.
A Modified Sector Model (MSM) has been proposed by the
authors [10], which takes into account the fact that the punching
resistance also depends on the level of transverse shear loading.
This model considers a quadrilinear moment–curvature relation-
ship including the tension stiffening effect. Comparisons with
experimental results indicated that slab rotations generally agree
very well, but are slightly overestimated immediately after the
cracking of the slab [10]. This overestimation was attributed to
the tensile contribution in the fracture process zone, which had
not been taken into account.
The unloading and reloading of slabs may further influence the
deformation behavior and thus the punching resistance. Prior
unloading of slabs may be required (e.g. by bracings) if existing
slabs have to be strengthened against punching shear in order to
activate the post-installed strengthening systems. For the uniaxial
tension chord it was shown [11] that unloading and reloading (UR)
cycles influence the deformation behavior by affecting the bond
properties between the reinforcing steel and surrounding concrete.
A residual slab rotation, wres, was noticed after unloading of the
slab [12]. After reloading up to the same load as previously applied,
an irreversible increase of rotation, DwUR, was observed [13,14].
Some experiments even failed in punching before reaching the

http://dx.doi.org/10.1016/j.engstruct.2015.02.016
0141-0296/Ó 2015 Elsevier Ltd. All rights reserved.
 R. Koppitz et al. / Engineering Structures 90 (2015) 130–142 131

Nomenclature

Ac concrete cross-sectional area wcr critical crack width where concrete tensile strength is
As, as cross-sectional area of reinforcement (total, per unit exhausted
width) x horizontal coordinate (axial direction), depth of com-
At,ef effective tension area (tension chord) pression zone
B side length of slab y horizontal coordinate (transverse direction)
Ec concrete Young’s modulus z vertical coordinate, lever arm of internal forces
Es steel Young’s modulus DL elongation of beam element
EII uncracked flexural slab stiffness per unit width DNcr axial force increase resulting from specific fracture
= Ec h3/12 energy
II
EI cracked flexural slab stiffness per unit width DVR0 predicted difference of punching shear resistance or
G dead weight portion of punching resistance borne by concrete
GF specific fracture energy DVR1 predicted difference of punching shear resistance
L initial length of beam element (strengthened slab)
M bending moment Dmcr moment increase resulting from specific fracture energy
N axial force Dmunl moment offset of unloading path
Ncr cracking load Dmrel moment offset of reloading path
Nsup, Ninf upper and lower bounds of axial force (UR cycle) D e0 characteristic tension stiffening effect (TCM)
V shear force Decr strain offset
Vflex flexural capacity of non-strengthened slab Dvcr curvature offset
Vmod shear force (MSM), = f(w) Deunl additional tension stiffening for unloading (tension
VRc failure criterion (CSCT), = f(w) chord)
VR,crush crushing resistance of concrete, = f(w) Dvunl additional tension stiffening for unloading (slab)
VR0 predicted punching shear resistance (non-strengthened Dvm tension stiffening curvature offset (Marti, Burns)
slab) or portion of punching shear resistance borne by DvTS tension stiffening curvature offset (CSCT)
concrete DwR0 predicted slab rotation increase at failure (non-
VR1 predicted punching shear resistance (strengthened slab) strengthened slab)
Vskt shear force (CSCT), = f(w) DwR1 predicted slab rotation increase at failure (strengthened
Vsup, Vinf upper and lower bounds of shear force (UR cycle) slab)
b side length of column DwUR slab rotation increase after complete UR cycle
d effective depth DwUR,exp experimental slab rotation increase after complete UR
dg maximum aggregate size cycle
dm average effective depth of both orthogonal directions DwUR,mod predicted slab rotation increase after complete UR cycle
dv shear-resisting effective depth acyl angle of loading cylinder to main axis
fc (cylinder) concrete compressive strength bE reduction factor of steel Young’s modulus
fct concrete tensile strength d bond slip = relative displacement between steel and
fsy yield strength of reinforcing steel concrete
h slab thickness dy slip at which rebar has reached its yield strength
kg factor to take crack roughness into account du slip at which rebar has reached its ultimate strength
ksys factor to take performance of shear reinforcement ec concrete strain
system into account ec,el concrete strain in elastic part of beam
ks bond stress factor es steel strain
mcr cracking moment per unit width esm average steel strains over crack element
mr, mt radial and tangential moments per unit width esy steel strain at yielding stress
mR bending resistance (average of both rebar directions) jV strength reduction factor for reinforcement inside shear
msup, minf upper and lower bounds of moment (UR cycle) crack
n modular ratio = Es/Ec k coefficient for crack spacing
r radius (from slab center) m Poisson’s ratio
r0 radius of critical shear crack (CSCT) = rc + dm q geometrical reinforcement ratio
rc radius of (equivalent) circular column qm average reinforcement ratio of both orthogonal direc-
rcr radius of cracked zone tions
~r cr radius of cracked zone at critical crack width wcr qs,ef reinforcement ratio relating to effective tension area
rcrs radius of zone in which cracking is stabilized (at mcr) rc concrete stress
~r crs radius of zone in which cracking is stabilized rs steel stress
(at mcr + Dmcr) rsr steel stress at crack edge
rq radius of load introduction at perimeter rsr0 steel crack stress
rs (equivalent) radius of circular isolated slab element sb bond shear stress between steel and concrete
= slab radius sb0 initial bond stress for elastic reinforcement = 2 fct
ry radius of yielded zone sb1 initial bond stress after onset of yielding = fct
srm average crack spacing v curvature
u0 control perimeter for punching shear resistance vcr curvature at cracking
(distance dv/2 to supported area) v~ cr curvature at critical crack width wcr
w crack width vcrs curvature at stabilized crack phase (at mcr)
132 R. Koppitz et al. / Engineering Structures 90 (2015) 130–142

v~ crs curvature at stabilized crack phase (at mcr + Dmcr) wres,exp experimental residual slab rotation after unloading
vr, vt radial and tangential curvatures wres,mod predicted residual slab rotation after unloading
vy curvature at yielding of reinforcing steel x mechanical reinforcement ratio = q  fsy/fc
w slab rotation outside column region xmin, xmax minimum and maximum mechanical reinforcement
wG elastic rotation under dead weight ratios
wres residual slab rotation after unloading Ø rebar diameter

same load as previously applied [14]. Additional deformations kf ct 1  q

resulting from such load histories may thus reduce the punching
resistance and their accurate modeling or prediction is important.
In this paper, the modeling of the effects of UR cycles on the
deformation behavior of the uniaxial tension chord is therefore
extended to bidirectional slabs, focusing on single UR cycles
required for slab strengthening. On the basis of the MSM, the resi-
dual slab rotations after unloading and irreversible rotation
increases after reloading of slabs are described. Furthermore, the
tensile contribution of the fracture process zone is taken into
account by a softening branch in the concrete stress–strain rela-
tionship in order to eliminate the above-mentioned overestimation
of deformations during cracking. A parametric study is carried out
to quantify the effects of these parameters on the punching shear
resistance of non-strengthened and strengthened flat slabs.
2. Concrete contribution in tension
2.1. Tension Chord Model
The contribution of the concrete between the cracks in tension
can be considered using the TCM [3], which assumes a stepped
rigid–perfectly plastic bond shear stress–slip relationship for
monotonic loading [15]. As long as the rebar stress is below the
yield strength, fsy, an upper bond shear stress of sb0 = 2 fct is
assumed. After yielding, the bond shear stress is reduced to a lower
value of sb1 = fct to take an irreversible bond reduction into account.
A linear elastic behavior is assumed for concrete up to the tensile
strength, fct; no softening behavior is subsequently considered.
In a tension chord with applied axial force, N, bond stresses
between the cracks reduce the average steel stresses and thus
the average steel strains, esm, inside the crack element with con-
stant average crack spacing srm, compared to the case where fct is
neglected (equal to naked steel characteristic). Applying the TCM
results in a horizontal shift of the average steel strains by a con-
stant factor De0, see Fig. 1a and Eq. (1):
De0 ¼ ð1Þ
2Es q
where q = As/Ac with As = p Ø2/4 (centrally arranged rebar with dia-
meter Ø), Ac = gross cross-sectional area of tension chord; and
0.5 6 k 6 1.0. The two boundaries of k represent minimum and
maximum values for crack spacing. The actual value is determined
by the spacing of the transverse reinforcement.
The behavior of the uniaxial tension chord was also investigated
for UR cycles [11]. Based on Kenel [16,17], an irreversible reduction
of bond shear stress sb = ks fct between rebar and surrounding con-
crete from the initial value of ks = 2 to ks = 1 was suggested for
unloading and reloading cycles. The bond stress–slip path is there-
fore altered during the UR cycle, see Fig. 2a, resulting in the load–
average steel strain curve of Fig. 2b (abscissa normalized by De0).
The path is represented by BHI, whereby the deformation at I is
higher than at B due to the irreversibly reduced bond properties.
Small circles between B and H, and H and I, respectively, denote
full slip reversal in the tension chord, where all shear stresses along
the rebar have reversed their direction. Further unloading (or
reloading) results in a slope equal to the steel axial stiffness at a
distance of ks/2  De0 to the curve where fct is neglected.
In beams subjected to bending moments, M, the contribution of
concrete in tension can be assumed as being analogous to the uni-
axial case. The zone around the rebars in Fig. 1b is considered as
tension chord with crack spacing srm and effective reinforcement
ratio qs,ef = As/At,ef. Marti [18] approximated the effective tension
area, At,ef, as one third of the area below the neutral axis according
to [19,20] and for beams he derived a constant curvature offset,
Dvm, due to tension stiffening, see Eq. (2), and Fig. 1c.
De0 kf ct
Dvm ¼  ð2Þ
d  xII 6as Es
where xII = depth of concrete compression zone in cracked elastic
state, as = cross-sectional area of tensile reinforcement per unit
width. Later Burns [21] proposed a solution for Dvm using an

(a) N (c) M
Ec Ac Δεcr EI I/h Δχcr
Δε0 Δχm
1 1/h
Ncr Es As Mcr EI II /d
1 1/d

εsm χm

(b) -
x II
≈ (d − x) / 3

M crack element M
d − x II

h s rm = L
M y
x y
x + χm
z z
εsm

b A t,ef srm = L As

Fig. 1. (a) Tension chord with applied load–average steel strain diagram in uncracked and cracked elastic state, (b) denomination of beam and (c) transformation to applied
moment–average curvature diagram in uncracked and cracked elastic state acc. to [18].
 R. Koppitz et al. / Engineering Structures 90 (2015) 130–142 133

(a) τb (a) (b) σc

w
A B C 2fct = τb0 fct elastic
part
fct = τb1 N N
L + ΔL Ec
Ι D F
1
O,P
ΔL = εel L + w
δy δu δ ε el
σc (c) σc
H fracture
fct fct fct process
zone
(b) N 1 − kτ / 2 1

kτ / 2 kτ / 2 w GF
Ec Ac B Ι
Nsup
Δε 0 ΔL wcr w
P A Es As
Ncr
Δε 0 Fig. 3. Fictitious Crack Model [6,7]: (a) behavior of concrete beam loaded in
tension; (b) stress–strain relationship of elastic beam parts; (c) stress–crack
full slip reversal opening relationship inside fracture process zone (adapted from [15]).

Ninf H
O
functions [31], or e.g. balances of elastic and plastic energy [32];
see summary by Kenel [27]. The area below the stress–crack open-
εsm
ing curve represents the specific fracture energy in tension, GF,
Δε 0 which is dissipated per unit area of the fracture process zone until
Fig. 2. (a) Rigid–perfectly plastic bond–slip relationship for UR cycles [11] and (b)
complete separation of the two specimen parts occurs. Wittmann
elastic UR load path for tension chord. et al. [33], Hordijk [34] and Trunk [35] experimentally confirmed
a primary dependence of GF on the concrete tensile strength and
the aggregate size. Following CEB-FIP Model Code 90 [36], Marti
effective reinforcement ratio compatible with the Bernoulli bending [18,37] proposed an empirical formula for the specific fracture
theory. To take into account a loss of bond stress due to creep under energy, depending on concrete tensile strength, fct (in MPa), and
sustained loads she reduced Dvm. As a limit case she investigated maximum aggregate size, dg (in mm):
an unloading of a tension chord for a complete loss of bond stress pffiffiffiffiffi
(equivalent to setting ks = 0); a reloading case, however, was not f ct 4 dg
GF ¼ ðin N=mmÞ ð4Þ
investigated. 80 mm3=4
In the rotation-symmetric Quadrilinear Sector Model (QSM) for
slabs, see Section 3, Muttoni [2] further revised Eq. (2) by setting
k = 1, q  as/h, and reducing the steel Young’s modulus by a factor 3. Modified Sector Model
bE to take an orthogonal instead of radial and tangential reinforce-
ment layout into account, see also [10]: In the rotation-symmetric sector model based on Kinnunen and
Nylander [1] isolated rigid slab segments are considered, which
f ct rotate around the edge of a central column with a rotation angle,
DvTS ¼ ð3Þ
6qhbE Es w, see Fig. 4. Equilibrium of moments results in a shear force as a
function of the slab rotation, Vskt(w) [2]:

2.2. Fictitious Crack Model V· Δϕ

2π
In displacement-controlled tensile tests on short concrete spe-
cimens a softening behavior instead of a sudden brittle failure
was observed after tensile strength was reached at the weakest
point inside the specimen [22–25]. The softening at this point mr mt
accompanied deformation localization in a single fracture process rq
zone because of progressive micro-cracking between the cement rs
matrix and aggregates [15,26,27]. Δϕ
The Fictitious Crack Model (FCM) was proposed by Hillerborg
V· Δϕ rc r0
[6,7] in which the total elongation of a concrete beam (with initial
2π
length L, see Fig. 3a) after cracking, DL, is composed of the elastic
unloading deformation of the intact beam part (Fig. 3b) and a local r0
deformation due to the softening inside the fracture process zone.
A corresponding stress–crack opening relationship (Fig. 3c)
assumes that the fracture process zone has an initial length of zero,
∼ ψ ·dm
i.e. is a fictitious crack. The tensile strength decreases with increas-
dm

ing crack opening, w, until it drops to zero at a critical value, wcr,

ψ
and the fictitious crack becomes a real crack.
The softening behavior of concrete in the fracture process zone rc
was described by various analytical functions. Common relation-
ships are either constant [28] or linear [7,29], bilinear [30], power Fig. 4. Sector model with rigid outer slab segment and acting forces [2].
134
 Z rs
2p
V skt ðwÞ ¼  mr ðwÞ  r 0 þ mt ðwÞ dr
rq  rc ro
A quadrilinear moment–curvature relationship is suggested and
the radial curvature vr of the radial moment mr at r = r0 is assumed
as vr = w/r0 and the tangential curvature as vt = w/r along the seg-
ment. The maximum slab resistance is limited by the flexural capa-
city, Vflex, which is reached as soon as the bending resistance, mR, of
the entire radial and tangential reinforcement is fully activated.
The Modified Sector Model (MSM) [10] introduces a strength
reduction factor, jV, for the shear crack-crossing longitudinal rein-
forcement, thus taking the influence of shear on the flexural behav-
ior into account. jV depends on the mechanical longitudinal
reinforcement ratio, x = q fsy/fc. Eq. (6) follows from Eq. (5), multi-
plied by a reduction factor with mr calculated at r = r0.
1
V mod ðwÞ ¼  V skt ðwÞ
1 þ jV mmr ðwÞ r0
R rs
In the Critical Shear Crack Theory (CSCT) Muttoni [2] proposed a
concrete failure criterion, VRc(w), as follows:
pffiffiffiffi
0:3u0 dv f c
V Rc ðwÞ ¼
0:4 þ 0:125wdm kg
where u0 = control perimeter at dv/2 distance to support area, dv =
shear-resisting effective depth, dm = average effective depth,
kg = 48/(dg + 16 mm), dg = maximum aggregate size, and fc = cylin-
der compressive strength of concrete. For slabs without shear rein-
forcement the intersection of Eqs. (6) and (7) determines the
(theoretical) punching resistance of the slab, VR0.
The punching resistance of slabs with shear reinforcement can-
not exceed the crushing resistance of the concrete strut near the
column [38], VR,crush, which can be estimated as follows [39]:
V R;crush ðwÞ ¼ ksys V Rc ðwÞ
where VRc(w) from Eq. (7) is multiplied by an empirical system fac-
tor, ksys.
4. Effect of UR cycles on slab behavior
4.1. Extension of Modified Sector Model by concrete softening
The consideration of the softening behavior of concrete after
cracking, which was neglected in the MSM, results in a further
deformation reduction, in addition to the tension stiffening effect.
The Dugdale model [28] provides the simplest approach to take the
effect of interlocking cement fibers into account, as the tensile
strength is maintained constant until it drops to zero at wcr
(Fig. 5a) where all fibers are pulled out. This model has the advan-
tage of omitting a crack-opening dependence of the transferrable
stress in the fracture zone below wcr. Here, the fracture zone
(a) (b)
σc σc
R S R S
fct fct
GF
Q T Q T
wcr w ΔL
Fig. 5. (a) Stress-softening diagram acc. to Dugdale [28] and (b) smeared crack
approach.
R. Koppitz et al. / Engineering Structures 90 (2015) 130–142

developing in each crack element is smeared along its length srm.
ð5Þ
Upper and lower boundaries for srm are given according to the
TCM [3]. In fact srm is often determined by the spacing of the trans-
verse reinforcement. The area QRST below the stress–strain curve
in Fig. 5b thus represents the specific fracture energy GF divided
by srm. Accordingly, the load–average strain and moment–curva-
ture diagrams in Fig. 6a and b are modified by this additional area
QRST, which is now GF At,ef/srm. After the cracking load Ncr (or the
cracking moment per unit width mcr) is reached, the curves ascend
reflecting the cracked axial (flexural) stiffness until the critical
crack opening is reached. Subsequently, the curves are horizontally
shifted back to the original quadrilinear curve by Decr (Dvcr,
respectively), as already shown in Fig. 1.
Applying Eq. (6) using the quintilinear moment–curvature rela-
tionship shown in Fig. 6b (OQRSCDF) results in
ð6Þ 1 2p
V mod ðwÞ ¼ 
1 þ jV mmr ðwÞ
R rs
r0 r  r
q c
  
~r crs
 mr ðwÞ  r 0 þ mR hr y  r 0 i þ EIII w ln
ry
 
  rs
þEIII DvTS ~r crs  ry þ ðmcr þ Dmcr Þh~rcr  ~r crs i þ EII w ln
ð7Þ r cr
  
r cr
þEIII w ln þ EIII ðDvTS þ Dvcr Þhrcr  ~r cr i ð9Þ
~r cr
where mcr = cracking moment, mR = bending resistance (Dmcr and
Dvcr according to Fig. 6b), EII and EIII = uncracked and cracked flex-
ural stiffness (DvTS according to Eq. (3)), r0 6 ri = w/vi 6 rs with sub-
script i = cr, crs or y. Tilde symbols denote curvatures and
corresponding radii at points R, S instead of P, A (e.g.
v~ cr ¼ vcr þ Dmcr =EIII , ~rcr ¼ w=v~ cr ). Operator hxi is x for x P 0 and 0
for x < 0.
Omitting GF (corresponding to the quadrilinear relationship in
ð8Þ
Fig. 6b, path OPTCDF) results in Dmcr and Dvcr equal to zero, and
~rcr and ~rcrs are identical to rcr and rcrs. The last line of Eq. (9) is
dropped and the remaining terms in the squared brackets are
equivalent to Eq. (6) in [2]. Additionally neglecting fct (correspond-
ing to the bilinear relationship in Fig. 6b) leads to DvTS = 0. Only
the first three terms in the squared brackets of Eq. (9) remain,
where rcr and rcrs derive to rs. For jV = 0 Eq. (9) corresponds to Eq.
(7) in [2].
4.2. Unloading and reloading of slabs
Every point in the slab segment is unloaded and reloaded from a
different stress level, because the slab moments are radius-depen-
dent, see Fig. 7a. To avoid having different UR paths for every slab
point, a simplification is made which is explained for the elastic
uniaxial tension chord in Fig. 7b. By assuming that full slip reversal
is always reached, i.e. a sufficiently large load difference Nsup  Ninf
is ensured, the curve progression can be approached by a rigid–ful-
ly cracked envelope. Unloading from B to H is modeled by an
immediate load drop to the parallel line at ks/2  De0 distance to
the naked steel characteristic (BY), and a further curve progression
along that parallel line down to Ninf (YH). Reloading is modeled
likewise by path HZI.
Fig. 7c illustrates the modified loading path OPRB1 where point
GF Ec B1 is at Ncr + DNcr. Accordingly, the unloading path is now shifted
srm downwards by a constant value Es As Deunl (= distance UV) where
1 Deunl is determined by equalizing the two gray areas (QRST and
UVWX), thus assuming that the fracture energy during loading is
the same as the energy required to push the fibers back during
L
unloading. Now unloading takes place along path B1UVH2 (at full
unloading). H1 = X is actually shifted horizontally to H2 and not
vertically to W, giving a small error indicated by the extra triangle
 R. Koppitz et al. / Engineering Structures 90 (2015) 130–142 135

(a) (b)
N mr , mt
C,D C,D
NR mR
I
Ec Ac Es As EI /h EI II /d
Δε cr Δχcr
1 1 1 /h 1/d
S R
Δε 0 R S
ΔχTS
Ncr+Δ Ncr mcr+Δ mcr
Ncr P,Q A,T
mcr P,Q A,T bilinear
quadrilinear
O O
εsm χcr χcrs χy χr , χt
χ~cr χ~crs

Fig. 6. Extension of (a) load–average elongation and (b) moment–curvature diagram with stress-softening behavior of concrete after cracking.

(a) χt
ψ
r0 < χy χ~crs
χ~cr χ
crs χcr

mt r0 rs r
mR quintilinear
quadrilinear
mcr+Δ mcr bilinear
mcr

r0 r~crs rcrs rcr s r r

r~cr

(b) N Δε cr Δε 0 (c) N
B2 Ι2
B Ι Nsup
Nsup R S,B1
Ncr+ ΔNcr kτ
2 Δε 0
kτ
2 Δε 0
P A
Ncr Ncr P,Q A,T
Y
Y1,U Y2
Z
Z2
V
Es As
Ninf H 1 Δεunl
O
O N inf
εsm H1,X
H2 ε sm
W

Δχcr ΔχTS Δψ UR
(d) mr , mt (e) V
B2
B2 Ι2 Vsup
msup Ι2
R S
mcr+Δmcr kτ
ΔχTS
2 Y2
mcr P,Q

Y2
Δ m unl

Z2
Z2
Δ m rel

Δχunl Vcr
minf Vinf
O H2 χ~crs χr, χt ψ resO H1 H2 ψ

Fig. 7. (a) Tangential moment distribution along segment (loading in elastic state), (b) simplification of UR load path, (c) elastic UR load path of tension chord, (d) moment–
curvature relationship of slab, and (e) shear force–rotation curve of slab.

XWH2 below the abscissa that is considered in the calculation of the naked steel characteristic down to Ninf (point H2) where a resi-
Deunl. dual elongation of the rebar remains (distance OH2). The reloading
Unloading starting at B2 > B1 does not provide any supplemen- path back to Nsup, H2Z2I2, runs along a parallel line at ks/2  De0 dis-
tal vertical shift. Thus the unloading path B2Y2H2 runs along the tance to the naked steel characteristic, and thus no softening is tak-
parallel line at a total horizontal distance of Deunl + ks/2  De0 to en into account in reloading. For reloading the same value of ks was
136 R. Koppitz et al. / Engineering Structures 90 (2015) 130–142

assumed as for unloading [11]. Therefore back at Nsup a horizontal

gap between I2 and B2 of (1  ks/2) De0 remains after completion of
(a) rc B/ 2
the UR cycle. π
The moment–curvature relationship (Fig. 7d) is modified in the 8
π
same way. Unloading from msup to minf (path B2Y2H2) is 4
V
8
accomplished by an immediate vertical drop of Dmunl, Eq. (10),
and a subsequent linear decrease with slope EIII. Reloading (path
H2Z2I2) exhibits a vertical rise of Dmrel, Eq. (11), and the same rq
linear increase back to msup (point I2), where there also
remains a horizontal gap in relation to the initial load path,
DvUR = (1  ks/2) DvTS.
(b) b/ 2

b/ 4

 B/ 2
IIks
Dmunl ¼ EI 1þ DvTS þ Dvunl ð10Þ
2 αcyl
Dmrel ¼ EIII ½ks DvTS þ Dvunl  ð11Þ V
e/2 8
where
Dmcr Dvcr rq
Dvunl ¼   ð12Þ

f
EIII v~ crs  k2s DvTS

Based on the moment–curvature relationships for the UR cycle

B2Y2H2Z2I2 load–rotation curves can be calculated, see Fig. 7e. It is
assumed that the entire prefactor in Eq. (9), which takes transverse (c) b/ 2

b/ 4
B/ 2
shear into account, remains constant at its msup-value. Thus the UR
load–rotation curves are also linear. The increase of rotation result-
ing after the UR cycle of a slab in the cracked elastic state,
π
DwUR ¼ wI2  wB2 , can be determined for a known load difference π 12
12
Vsup  Vinf, resulting in a constant value due to the rigid–fully V
20
cracked approximation:
rq
 
1  k2s DvTS r s
b/ 4

DwUR ¼  ð13Þ b/ 2 B/ 2
1 þ ln rr0s

π
4.3. Validation with experiments π 12
12
V
4.3.1. Experimental database 24
rq
In the literature only a small number of experiments were V
found with single unloading and reloading cycles before the load 48 Sr2 only
was increased up to failure. At ETH Zürich four slabs (P2–3, P5–
6) without post-tensioning were tested in the 1970s [13,14]. All
(d) rc rs
slabs had an octagonal shape, a central column with diameter
300 mm and were loaded by eight cylinders, see Fig. 8a. Slabs P2,
P3 and P5 had an orthogonal reinforcement layout, P6 included
only tangential ring reinforcement, but radially arranged stirrups
around the column. Here, a secant rotation was calculated from V
i i = # loading points
the average deflection at the slab edge, assuming zero deflection rq
at the column edge. Two of the slabs used in Lips’ [40,41] investi-
gation on the punching resistance of square slabs with large
amounts of shear reinforcement, PF3 and PF5, were also unloaded Fig. 8. Slab dimensions and yield-line mechanisms: (a) fan mechanism for round
column (slabs P2–3, P5–6); four-point star mechanisms for (b) eight (slabs PF3,
and reloaded. The layout of the two slabs with orthogonal rein-
PF5), (c) 20 (Sd1–2, Sh2–4, Sr1–2) and 24 load points (slab Sr3); (d) truncated cone
forcement is given in Fig. 8b. The slab rotations of the weak axis mechanism of equivalent circular slab.
were measured directly using inclinometers that were installed
near the applied load.
An experimental campaign on several strengthening systems system) on the punching resistance of the non-strengthened slabs.
for existing RC flat slabs was carried out in [12,42,43]. Three slabs, The slab properties and yield-line mechanisms (YLMs) of all speci-
Sh2–4, were strengthened against punching shear by an externally mens are summarized in Fig. 8a–c and Table 1. For the MSM the
prestressed steel head placed underneath the slab. Another three slab and column geometries were transformed to an equivalent
slabs, Sr1–3, contained non-laminated and prestressed carbon circular slab with the same flexural capacity, Fig. 8d. For actual cir-
fiber-reinforced polymer (CFRP) straps installed crosswise around cular slabs and columns bE  0.6 was used [2], for square ones
the column in predrilled and precut openings, and anchored on bE  0.7 [10]. Slab rotations were calculated from deflections mea-
the bottom side of the slab by a steel compression frame. The slabs sured near the column and the slab edge. The maximum rotation of
(for properties see Fig. 8c) were preloaded and unloaded before the the weak axis was selected, except for Sr1–3 where only deflec-
strengthening systems were installed and applied. Two further tions along the strong axis were recorded during preloading.
slabs, Sd1–2, were investigated without subsequent strengthening, The zero measurements in the experiments prior to loading
but to quantify the effect of boreholes (required for the strap define the origin Ô of the load–rotation curves, see Fig. 9. Thus
 R. Koppitz et al. / Engineering Structures 90 (2015) 130–142 137

Table 1
Slab properties for model validation.

Reference # unla rela h (mm) dm (mm) qm (–) fc (MPa) fct (MPa) Ec (GPa) fsy (MPa) Es (GPa) Vflex/mR (–) xm (–) jV (–)
b
Marti et al. [13] P2 Yes Yes 180 143 1.48% 37.3 3.4 37.2 558 206 7.71 0.22 0.46
P3 Yes Yes 180 152 1.40% 35.8 3.3b 36.6 558 206 7.71 0.22 0.46
Pralong et al. [14] P5 Yes Yes 180 154 1.31% 27.1 2.7b 37.3 515 204 7.71 0.25 0.53
P6 Yes Yes 180 154 1.31% 30.0 2.5b 38.1 515 204 7.71 0.22 0.48
Lips [40,41] PF3 Yes Yes 250 209 1.50% 37.1 3.3c 33.4c 583 200 8.14 0.24 0.50
PF5 Yes Yes 400 354 1.50% 33.4 3.1c 32.2c 580 200 7.74 0.26 0.57
Unpublished Sd1 Yes Yes 256 199 1.58% 51.0 3.9b 38.5 517 205 7.39 0.16 0.33
Sd2 Yes Yes 257 202 1.56% 52.2 3.9b 39.5 517 205 7.39 0.15 0.31
Lips et al. [43] Sh2 Yes No 260 199 1.58% 59.1 4.0b 37.4 514 205 7.39 0.14 0.28
Sh3 Yes No 260 203 1.55% 47.6 3.9b 40.0 569 205 7.39 0.19 0.38
Sh4 Yes No 255 207 1.52% 45.3 3.4b 40.3 569 205 7.39 0.19 0.40
Koppitz et al. [12] Sr1 Yes No 257 200 1.57% 48.1 3.6b 41.0 517 205 7.39 0.17 0.35
Sr2 Yes No 187 138 1.46% 43.1 3.4b 35.8 521 205 7.39 0.18 0.36
Sr3 Yes No 325 264 1.44% 44.2 3.4b 35.1 525 205 7.24 0.17 0.36
a
unl = unloading path, rel = reloading path of non-strengthened slabs.
b
Values from double punch tests on cylinders (h = Ø = 150 mm) [44,45].
pffiffiffiffiffi 2=3
c
Values not measured, calculated with [46]: Ec ¼ 10; 000 MPa2=3  3 f c ; f ct ¼ 0:3 MPa1=3  f c ðboth in MPaÞ.

V Q
VR,exp
zero measurement
G
O ψ − ψG
O ψ R,exp
Fig. 9. Shift of origin for experimental curves.
the dead weight G of the slab and the equipment and the resulting
slab rotation wG had to be added and the coordinate system was
shifted from Ô to O. wG was estimated using elasticity theory.
Assuming an uncracked behavior of an equivalent circular slab,
which is ring-supported at rc, and ring-loaded along the slab edge
with rs  rq, maximum slab rotation at the slab edge is [47]:
3G 1  m r 2q  r 2c
wG ¼ ð14Þ
p Ec h3 rq
where m = Poisson’s ratio  0.2.
In the following modeled and measured load–rotation curves of
non-strengthened slabs are compared in normalized diagrams
(Figs. 10 and 11) according to [2]. On the ordinate the equivalent
pffiffiffiffi
shear stress, V/(u0 dv) is divided by f c , assumed as being propor-
tional to the concrete shear strength. On the abscissa, the product
w dm kg is plotted.
ψ
4.3.2. Monotonic loading considering concrete softening
In Fig. 10, the observed monotonic load paths (solid lines) of five
slabs are compared to calculated load paths, Eqs. (6) and (9), with-
out (dotted) and with (dashed) consideration of concrete softening
by using the estimation of GF in Eq. (4). The shapes of the dashed
curves show better agreement with the measurements than the
dotted curves. For higher applied loads both calculations
gradually result in the same rotations, since the stiffening effect
due to the bridging fibers is exhausted as soon as all crack widths
exceed wcr.

0.4 Legend

0.3
PF3 PF5 Load-rotation:
Experiment
MSM: Vmod (ψ ), Eq. (9),
0.2 with GF = 0
MSM: Vmod (ψ ), Eq. (9),
with GF acc. to Eq. (4)
0.1
Special points:
0 UR cycle: Vsup
0 0.5 1 1.5 2 2.5 0 0.5 1 1.5 2 2.5
0.4
Vsup
0.3
Sr1 Sr2 Vsup Sr3 Vsup

0.1
Rotation ψA : Rotation ψA : Rotation ψA :
0.2 d = 211 mm d = 147 mm d = 276 mm
ρ = 1.49% ρ = 1.37% ρ = 1.38%

0
0 0.5 1 1.5 0 0.5 1 1.5 0 0.5 1 1.5
ψ · d m · k g [-] ψ · d m · k g [-] ψ · d m · k g [-]

Fig. 10. Monotonic loading, comparison of quadri- and quintilinear MSM to experimental results.
138 R. Koppitz et al. / Engineering Structures 90 (2015) 130–142

0.8
Legend
0.6 PF3 PF5
Vflex Load-rotation: Special points:
1+κV r0 / rs Experiment Failure (V R,exp)
0.4
UR cycle: V sup
MSM
0.2 Vmod (ψ ), Eq. (9), Transition point (ry = r0)
GF acc. to Eq. (4) Flexural capacity (ry = rs)
0
0 5 10 15 0 2 4 6
0.4
0.3 Sr1 Sr2 Sr3 Sh2

0.2

0.1

0
0 0.4 0.8 1.2 0 0.4 0.8 1.2 0 0.4 0.8 1.2 0 0.4 0.8 1.2
0.4
0.3 Sh3 Sh4 Sd1 Sd2

0.2

0.1

0
0 0.4 0.8 1.2 0 0.4 0.8 1.2 0 0.4 0.8 1.2 0 0.4 0.8 1.2
0.4
0.3 P2 P3 P5 P6

0.2

0.1

0
0 1 2 3 0 1 2 3 0 1 2 3 0 1 2 3
ψ · d m · k g [-] ψ · d m · k g [-] ψ · d m · k g [-] ψ · d m · k g [-]

Fig. 11. Comparison of predicted and observed unloading (and reloading) cycles.

Table 2
Residual slab rotation wres and rotation increase after reloading (DwUR,mod according to Eq. (13)).

# wres,exp (mrad) wres,mod (mrad) wres;mod

(–) Average (–) C.o.V. (–) DwUR,exp (mrad) DwUR,mod (mrad) DwUR;mod
(–) Average (–) C.o.V. (–) DwUR;mod
(–)
wres;exp DwUR;exp wres;exp

P2 2.61 1.77 0.68 0.57 0.48 0.84 0.22

P3 1.83 1.49 0.82 0.43 0.50 1.14 0.23
P5 0.47 0.88 1.84 0.17 0.43 2.45 0.36
0.92 0.54 1.21 0.58
P6 2.21 1.32 0.60 0.50 0.40 0.79 0.23
PF3 22.0 24.4 1.11 –a 0.40 –a –
PF5 1.62 0.78 0.48 0.30 0.25 0.84 0.19
Sd1 1.72 1.19 0.69 0.33 0.35 1.08 – 0.19
Sd2 1.74 1.22 0.70 0.38 0.36 0.96 0.22
Sh2 1.99 1.35 0.68 – 0.36 – –
Sh3 1.65 1.24 0.75 – 0.36 – –
0.80 0.13 1.02
Sh4 1.12 1.02 0.92 – 0.33 – –
Sr1 1.18 1.04 0.88 – 0.35 – –
Sr2 2.31 2.00 0.86 – 0.46 – –
Sr3 0.82 0.73 0.89 – 0.30 – –
Total (14 slabs) 0.85 0.38 Total (7 slabs) 1.16 0.51 0.23
a
Failure prior to UR cycle completion.

4.3.3. Residual rotation after unloading and rotation increase after
reloading
Fourteen slabs were completely unloaded to V = G, eight of
them were reloaded, see Table 1. Slab PF3 failed before the UR
cycle could be completed. The experimentally obtained paths are
displayed in Fig. 11, together with modeled unloading and reload-
ing paths according to Fig. 7e (dashed curves). In the model the
maximum load before unloading, Vsup, resulted from adjusting
the computational slab rotation to the measured one before
unloading the slab. The slopes of the measured UR paths are steep-
er than the modeled ones because of the rigid–fully cracked
approach of the moment–curvature relationship. The residual slab
rotations after unloading are underestimated by the model by 15%
on average, see Table 2. The Coefficient of Variation (C.o.V.) of 38%,
however, is quite high; it is much lower for the complete datasets
of the authors’ experiments (13%). The predicted rotation increases
after completion of the full UR cycle, according to Eq. (13), overes-
timate the experimental values by 16% on average, the C.o.V. is
51%, see Table 2. By excluding the UR cycle of P5, the mean value
decreases to 0.94 with a C.o.V. of 15%. In the observed experiments
the rotation increases are only 23% (on average) of the residual
rotations, see also Table 2.
 R. Koppitz et al. / Engineering Structures 90 (2015) 130–142 139

Table 3
Parametric study: increase of slab rotation and decrease of punching resistance resulting from UR cycle (calculations 5 and 13 correspond to Fig. 14).

# Parameters UR cycle, ks = 1 UR cycle, ks = 0 2rc 3rc 2Vflex 2Vflex and 1.5EI

q (%) fc d DwR0 DVR0 DwR0 DVR0 DwR1 DVR1 DwR1 DVR1 DwR0 DVR0 DwR0 DVR0
(MPa) (mm) (%) (%) (%) (%) (%) (%) (%) (%) (%) (%) (%) (%)
1 2.1 30 200 +3 1 +5 1 +4 1 +3 1 +7 2 +11 2
2a 2.1 30 100 +1 1 +3 1 +6 3 +3 2 +17 5 +25 6
3 2.1 30 400 +4 1 +8 2 +7 2 +5 2 +10 2 +16 3
4 2.1 60 200 +3 1 +6 2 +5 2 +3 2 +8 2 +12 3
5a 2.1 60 100 +2 1 +4 2 +7 4 +4 3 +22 8 +32 9
6 2.1 60 400 +5 1 +9 3 +7 3 +6 2 +11 3 +17 4
Average (q = 2.1%) +3 1 +6 2 +6 2 +4 2 +13 4 +19 4
Max. DwR; min. DVR +5 1 +9 3 +7 4 +6 3 +22 8 +32 9
11a 0.5 30 200 +3 2 +7 4 +6 3 0 0 +46 17 +66 20
12a 0.5 30 100 +1 2 +4 3 +1 0 0 0 +78 30 +100 33
13a 0.5 30 400 +6 2 +12 5 +7 3 +5 3 +15 6 +23 7
14a 0.5 60 200 +4 2 +8 5 0 0 0 0 +69 26 +88 28
15a 0.5 60 100 +2 2 +4 3 0 0 0 0 +104 39 +128 43
16a 0.5 60 400 +7 3 +14 7 +8 5 +4 2 +32 13 +46 15
Average (q = 0.5%) +4 2 +8 4 +4 2 +2 1 +57 22 +75 24
Max. DwR; min. DVR +7 3 +14 7 +8 5 +5 3 +104 39 +128 43
Average (total) +3 2 +7 3 +5 2 +3 1 +35 13 +47 14
Max. DwR; min. DVR +7 3 +14 7 +8 5 +6 3 +104 39 +128 43
a
Including plastic rotations before unloading (Vsup above transition point).

4.4. Parametric study

0.8
GF = 150 N/m
Based on the validated models for the residual rotation and the 100
rotation increase after one UR cycle, a parametric study for non- 0.6 86
50
strengthened and strengthened slabs was performed to evaluate 0
the sensitivity of the results to some assumptions that were made 0.4
in the models. In the non-strengthened case, the following three r
ea
parameters were investigated: the influence of GF on the deforma- in
0.2 bil
tion behavior, the influence of Vsup on the residual rotation and the ρ high
influence of the rotation increase due to an UR cycle, DwUR, on the
punching resistance. Two slabs with the same geometry and mate- 0
rial properties (rc = 159 mm, rq = 1500 mm, rs = 1575 mm, 0.4
h = 250 mm, dm = 200 mm, G = 75 kN, fc = 30 MPa, Ec = 35 GPa, GF = 150 N/m
100
fct = 2.9 MPa, dg = 32 mm, fsy = 500 MPa, Es = 205 GPa, bE = 0.7), but 0.3 86
different reinforcement ratios (qhigh = 2.1%, qlow = 0.5%) were 50
selected for the first two parameters. For the third parameter q 0
)
(0.5% and 2.1%), dm (100, 200 and 400 mm, d/h = 0.8) and fc (30
0.2 =0
(G F
and 60 MPa), and thus also fct (2.9 and 4.6 MPa) and GF (86 and ear
a d rilin
137 N/m), were varied, see Table 3. qu
0.1 ar
These parameters were also used for the strengthened case in bili
ne ρ low
which slabs were strengthened after a preloading up to VR0 and a
0
full unloading down to zero to investigate the maximum possible 0 1 2
effect of an UR cycle on the punching resistance of concrete. ψ · d m · k g [-]
Because of modified slab properties after strengthening and/or
different failure criteria the residual rotation wres (= origin of Fig. 12. Effect of GF on load–rotation curve.
new V–w-curve) is the critical parameter instead of DwUR. Four
strengthening cases were investigated: The first one was a dupli-
cation and triplication of the support area (increase of the con- the prestressed and non-prestressed systems were compared,
trol perimeter, u0, and of the flexural capacity, Vflex), the second both assumed to be reached at VR,crush, according to Eq. (8), with
case was a duplication of Vflex by neglecting a stiffness increase ksys = 2.5 [10].
(i.e. in the case of the application of very thin CFRP sheets The load–rotation curves for different specific fracture energies
[48]), the third one assumed a duplication of Vflex and a stiffness GF are compared in Fig. 12 (GF = 86 N/m according to Eq. (4) and
increase of 50% (e.g. by applying an additional upper concrete GF = 0, 50, 100, 150 N/m were selected, whereas GF = 0 is equivalent
layer [49] or an external steelhead placed between column and to a quadrilinear moment–curvature relationship). The effect of GF
slab [43]). In the last case, the effect of a prestressed shear on the deformations is only visible immediately after cracking and
strengthening system was investigated, which is able to compen- gradually disappears as flexural capacity is approached. In the ser-
sate the residual slab rotation [12]. The UR cycle curves of the viceability limit state, it is therefore sufficient to refer to a simple
strengthened slabs were calculated as if they were monotonically estimation of GF as in Eq. (4). GF can be neglected in the
loaded and then horizontally shifted by wres. In the first three moment–curvature relationship for the determination of the
cases, the effect of residual slab rotations on the concrete contri- punching shear capacity.
bution to the punching resistance, according to Eq. (7), was In the experimental validation of the model, Vsup was calibrated
quantified. In the fourth case, the punching resistances, VR1, of to start unloading at the same rotation as in the measurements. In
140 R. Koppitz et al. / Engineering Structures 90 (2015) 130–142

1 qhigh a decrease from 1241 kN to 900 kN (28%) results in almost

Vsup : no reduction of wres (2%).
0.8 1241 kN The effect of a UR cycle on punching resistance is shown in
Fig. 14a and b for two non-strengthened slabs with different rein-
0.6 900 kN forcement ratios (qhigh = 2.1%, qlow = 0.5%), effective depths
(dm = 100 and 400 mm) and concrete compressive strengths
0.4 (fc = 60 and 30 MPa). The intersection of the load–rotation curve
with the failure criterion, Eq. (7), denotes the punching resistance,
0.2 ρ high VR0. If the slabs are unloaded and reloaded, rotation wR0 increases
by DwUR and, accordingly, VR0 decreases. The limit case of a com-
0
0.4
plete loss of bond between concrete and rebar, i.e. setting ks = 0
instead of ks = 1, doubles DwUR. Nevertheless, the slab rotations
Vsup : at failure increase by only 4% (Fig. 14a) and 12% (Fig. 14b), and
0.3
429 kN by 7% on average (total) and maximum 14% if considering all para-
350 kN
meter combinations, see Table 3. The resulting decreases of VR0
0.2 remain small: 2% and 5% for Fig. 14a and b, 3% on average (total)
and 7% maximum.
In the first strengthening case (see Fig. 14c and d) the increase
0.1
ρ low of the column radius, rc, increases the control perimeter, u0, and
simultaneously the flexural capacity of the slab by a reduction of
0 the denominator (rq  rc), see Fig. 8. wres causes practically no resis-
0 1 2 3
tance decrease (smaller than 5%), see Table 3.
ψ · d m · k g [-]
In the second case with doubled Vflex and no flexural stiffness
Fig. 13. Effect of Vsup on wres.
increase, a horizontal shift of the load–rotation curves by wres
caused by the UR cycle increases wR0 and thus decreases VR0 (see
Fig. 14e and f). The amount of increase and decrease, respectively,
some slabs the difference between the calculated Vsup and the is influenced by irreversible plastic slab rotations that have
actual value was more than 15%. The curves in Fig. 13 illustrate occurred before unloading and strengthening in some of the slabs
that, as long as no part of the slab’s longitudinal reinforcement (indicated in Fig. 14 by triangles, and in Table 3). In the case of
yields (below the marked triangle), the variation of residual rota- highly reinforced slabs (qhigh = 2.1%) the reinforcements of the
tion is low. In the case of the slab with qlow, a decrease of Vsup from two thin slabs were only plasticized around the column. Thus the
429 kN to 350 kN (18%) reduces wres by 11%. For the slab with rotation increase is 13% on average (22% for the slab in Fig. 14e)

0.5 Legend (a) and (b)

0.4
(a) (b) Monotonically loaded

UR cycle, kτ = 1
0.3
ΔψUR ΔψUR
UR cycle, kτ = 0
0.2 VR0 VR0
Failure criterion: concrete,
0.1 VRc (ψ ), Eq. (7)
(c) and (d)
0 Non-strengthened slab
0 1 2 3 4 5 0 1 2 3 4 5
1 Strengthened, 2rc
3rc
0.8
(c) (d) Strengthened, 3rc
ψres 2rc ψres
0.6 Failure criterion: concrete,
3rc VRc (ψ,rc ), Eq. (7)
ψres 1rc ψres
0.4 ψres 2rc ψres (e) and (f)
Non-strengthened slab
0.2 1rc
Strengthened, 2Vflex
0
0 2 4 6 8 0 1 2 3 4 5 Strengthened, 2Vflex , 1.5EI II
1
Failure criterion: concrete,
0.8
(e) ΔVR1 (f) ΔVR1 VRc (ψ ), Eq. (7)
Failure criterion: concrete
0.6 ψres ψres crushing, VR,crush (ψ ), Eq. (8)
Special points:
VR0 ψres VR1 VR0 ψres VR1 Transition point (ry = r0 )
0.4
ψres fc = 60 MPa ψres fc = 30 MPa UR cycle: Vsup (= VR0)
0.2 dm = 100 mm dm = 400 mm
Concrete resistance (VR0)
ρm = 2.1% ρm = 0.5%
Failure acc. to MSM (VR1)
0
0 1 2 3 4 5 0 1 2 3 4 5
ψ · d m · k g [-] ψ · d m · k g [-]

Fig. 14. Effect of UR cycle on: (a) and (b) VR0 of non-strengthened slab, (c) and (d) VR1 of strengthened slab (increase of support area), (e) and (f) VR0 of strengthened slab
(increase of flexural capacity and/or stiffness) and DVR1 due to prestressed system. The diagrams were evaluated for the parameters indicated in the boxes in (e) and (f),
respectively.
 R. Koppitz et al. / Engineering Structures 90 (2015) 130–142
and the decrease of the concrete contribution to the punching
resistance is less than 8% when compared to wres = 0 (4% on aver-
age). This decrease is much higher for low reinforced slabs
(qlow = 0.5%) with much larger plastic deformations before unload-
ing, especially for the thin slabs. An average decrease of concrete
resistance of 22% results (only 6% for the slab in Fig. 14f), with a
maximum loss of 39%. If additionally the flexural stiffness is
increased by 50% (third case), the obtained increases of wR0 are
approximately 30% higher; the corresponding reductions of VR0
are 10% higher on average compared to those without stiffness
increase.
In the fourth case, the applied prestressing force is adjusted in
such a manner as to fully compensate wres in cases 2 and 3 and thus
does not result in a reduced contribution of the concrete to punch-
ing resistance. The prestressed system (with wres = 0) is compared
to the non-prestressed one (with wres > 0). If only Vflex is doubled
(as in case 2, e.g. using prestressed CFRP straps [10]), strap pre-
stressing leads to an average increase of punching resistance,
DVR1, of 5%, see Fig. 14e and f. If the prestressed shear strengthen-
ing system further increases slab stiffness by 50% (as in case 3, e.g.
by a prestressed external steelhead [43]), a DVR1 of 14% results on
average.
5. Conclusions
Using the Modified Sector Model (MSM) load–rotation curves
were derived to discuss the effect of unloading and reloading
(UR) cycles on the deformation behavior and punching shear resis-
tance of non-strengthened and strengthened flat slabs. A compar-
ison of the model results to the authors’ and literature
experiments, and a parametric study, result in the following
conclusions:
1. A quintilinear moment–curvature relationship – which takes
into account a softening branch in the concrete stress–strain
relationship and the tension stiffening effect – together with a
UR cycle, modeled as a bilinear envelope assuming full slip
reversal, are able to predict residual slab rotations and irre-
versible rotation increases after a full UR cycle.
2. UR cycles increase the slab rotation and thus reduce the punch-
ing shear resistance part of the concrete compared to a mono-
tonically loaded slab; the effect however is small. The effect
may increase significantly if the slab is strengthened after
unloading or for slabs with low reinforcement ratios due to irre-
versible plastic slab rotations. Prestressing of the strengthening
system, however, may compensate the residual slab rotation
and thus prevent the loss of punching resistance.
3. In the serviceability limit state, slab rotations are more accu-
rately predicted if the concrete softening branch is considered.
Approaching the ultimate limit state – for estimation of the
punching shear resistance – the effect of concrete softening dis-
appears and may thus be neglected.
Further research is required to take into account time-depen-
dent effects like shrinkage or creep of concrete. These effects
may lead to additional deformations and may also result in a
decreased punching resistance. In the proposed model this may
be considered by the superposition of an additional time-depen-
dent slab rotation, and by a reduction of material properties of
concrete.
Acknowledgments
The authors wish to acknowledge the support and funding pro-
vided by the Swiss Federal Commission for Technology and
141
Innovation CTI (Grant No. 11569.2 PFIW-IW) and F.J.
Aschwanden AG, Lyss, Switzerland.
References
[1] Kinnunen S, Nylander H. Punching of concrete slabs without shear
reinforcement. Trans. Royal Institute of Technology, No. 158, Stockholm,
Sweden; 1960. p. 112.
[2] Muttoni A. Punching shear strength of reinforced concrete slabs without
transverse reinforcement. ACI Struct J 2008;105(4):440–50.
[3] Marti P, Alvarez M, Kaufmann W, Sigrist V. Tension chord model for structural
concrete. Struct Eng Int 1998;8(4):287–98.
[4] Higgins DD, Bailey JE. Fracture measurements on cement paste. J Mater Sci
1976;11(11):1995–2003.
[5] Grudemo Å. Microcracks, fracture mechanism and strength of the cement
paste matrix. Cem Concr Res 1979;9(1):19–33.
[6] Hillerborg A, Modéer M, Petersson P-E. Analysis of crack formation and crack
growth in concrete by means of fracture mechanics and finite elements. Cem
Concr Res 1976;6(6):773–81.
[7] Hillerborg A. Analysis of one single crack. In: Wittmann FH, editor. Fracture
mechanics of concrete. Amsterdam, The Netherlands: Elsevier; 1983. p.
223–49.
[8] Bažant ZP, Oh BH. Crack band theory for fracture of concrete. Mater Struct
1983;16(93):155–77.
[9] Bažant ZP, Ožbolt J, Eligehausen R. Fracture size effect: review of evidence for
concrete structures. ASCE J Struct Eng 1994;120(8):2377–98.
[10] Koppitz R, Kenel A, Keller T. Effect of punching shear on load–deformation
behavior of flat slabs. Eng Struct 2014;80:444–57. http://dx.doi.org/10.1016/
j.engstruct.2014.09.023.
[11] Koppitz R, Kenel A, Keller T. Tension chord model modification for uniaxial
unloading and reloading in elastic and plastic states. ASCE J Struct Eng
2014;140(10):04014077. Erratum, 140(10):08014002. http://dx.doi.org/10.
1061/(ASCE)ST.1943-541X.0000999 and http://dx.doi.org/10.1061/(ASCE)ST.
1943-541X.0001142.
[12] Koppitz R, Kenel A, Keller T. Punching shear strengthening of flat slabs using
prestressed carbon fiber-reinforced polymer straps. Eng Struct
2014;76:283–94. http://dx.doi.org/10.1016/j.engstruct.2014.07.017.
[13] Marti P, Pralong J, Thürlimann B. Schubversuche an Stahlbeton-Platten. Inst. of
Struct. Engrg., ETH Zürich, Switzerland. Report No. 7305-2; 1977. p. 123 [in
German] http://dx.doi.org/10.3929/ethz-a-000138231.
[14] Pralong J, Brändli W, Thürlimann B. Durchstanzversuche an Stahlbeton- und
Spannbetonplatten. Inst. of Struct. Engrg., ETH Zürich, Switzerland. Report No.
7305-3; 1979. p. 89 [in German]. doi:http://dx.doi.org/10.3929/ethz-a-
000185615.
[15] Sigrist V. Zum Verformungsvermögen von Stahlbetonträgern. Doctoral thesis
ETH Zürich, Switzerland, IBK-Report No. 210; 1995. p. 159 [in German]. doi:
http://dx.doi.org/10.3929/ethz-a-001492371.
[16] Kenel A, Marti P. Faseroptische Dehnungsmessungen an einbetonierten
Bewehrungsstäben. ETH Zürich, IBK-Report No. 271; 2002. p. 93 [in
German]. doi: http://dx.doi.org/10.3929/ethz-a-004306146.
[17] Kenel A, Nellen P, Frank A, Marti P. Reinforcing steel strains measured by bragg
grating sensors. ASCE J Mater Civ Eng 2005;17(4):423–31.
[18] Marti P. Stahlbeton Grundzüge I [Structural concrete Fundamentals Pt. I].
Lecture Notes, Inst. of Struct. Engrg., ETH Zürich, Switzerland; 1999. p. 106 [in
German].
[19] Martin H, Schießl P, Schwarzkopf M. Berechnungsverfahren für Rißbreiten aus
Lastbeanspruchung. Forschung Straßenbau und Straßenverkehrstechnik
1980;309:34–66 [in German].
[20] Schießl P. Grundlagen der Neuregelung zur Beschränkung der Rißbreite.
Deutscher Ausschuss für Stahlbeton, vol. 400. Berlin, Germany: Beuth; 1989. p.
157–75 [in German].
[21] Burns C. Serviceability analysis of reinforced concrete based on the tension
chord model. Doctoral thesis ETH Zürich, Switzerland, IBK-Report No. 342;
2012. p. 149.
[22] Weibull W. A statistical theory of the strength of materials. The Royal Swedish
Institute for Engineering Research, No. 151, Stockholm, Sweden; 1939. p. 45.
[23] Hughes BP, Chapman GP. The complete stress–strain curve for concrete in
direct tension. RILEM Bull New Ser 1966;30:95–7.
[24] Evans RH, Marathe MS. Microcracking and stress–strain curves for concrete in
tension. Mater Struct 1968;1(1):61–4.
[25] Heilmann HG, Hilsdorf H, Finsterwalder K. Festigkeit und Verformung von
Beton unter Zugspannungen. Deutscher Ausschuss für Stahlbeton, vol.
203. Berlin, Germany: Ernst & Sohn; 1969. p. 93 [in German].
[26] Kaufmann W. Strength and deformations of structural concrete subjected to
in-plane shear and normal forces. Doctoral thesis ETH Zürich, Switzerland.
IBK-Report No. 234; 1998. p. 147. doi:http://dx.doi.org/10.3929/ethz-a-
001972931.
[27] Kenel A. Biegetragverhalten und Mindestbewehrung von Stahlbetonbauteilen.
Doctoral thesis ETH Zürich, Switzerland. IBK-Report No. 277; 2002. p. 115 [in
German] doi:http://dx.doi.org/10.3929/ethz-a-004470989.
[28] Dugdale DS. Yielding of steel sheets containing slits. J Mech Phys Solids
1960;8(2):100–4.
[29] Zhu Y. Evaluation of bond strength between new and old concrete by means of
fracture mechanics method. Doctoral thesis Royal Institute of Technology.
TRITA-BST Bull. No. 157, Stockholm, Sweden; 1991. p. 112.
142 R. Koppitz et al. / Engineering Structures 90 (2015) 130–142
[30] Petersson P-E. Crack growth and development of fracture zones in plain
concrete and similar materials. Lund Institute of Technology, REPORT TVBM –
1006; 1981. p. 174.
[31] Foote RML, Mai Y, Cotterell B. Process zone size and crack growth
measurement in fiber cements. ACI SP 1987;105–3:55–70.
[32] Barenblatt GI. The mathematical theory of equilibrium cracks in brittle
fracture. Adv Appl Mech 1962;7:55–129.
[33] Wittmann FH, Rokugo K, Brühwiler E, Mihashi H, Simonin P. Fracture energy
and strain softening of concrete as determined by means of compact tension
specimens. Mater Struct 1988;21(1):21–32.
[34] Hordijk DA. Tensile and tensile fatigue behaviour of concrete; experiments,
modelling and analyses. HERON 1992;37(1):1–79.
[35] Trunk BG. Einfluß der Bauteilgröße auf die Bruchenergie von Beton. Doctoral
thesis ETH Zürich, Switzerland; 1999. p. 155 [in German] doi:http://dx.doi.org/
10.3929/ethz-a-002053523.
[36] CEB-FIP Model Code 1990: Design Code. Comité Euro-International du Béton,
Thomas Telford, London, UK; 1993. p. 437.
[37] Marti P, Alvarez M, Kaufmann W, Sigrist V. Tragverhalten von Stahlbeton. IBK
Special Publication SP-008, Inst. of Struct. Engrg., ETH Zürich, Switzerland;
1999. p. 301 [in German].
[38] Beutel RRK. Durchstanzen schubbewehrter Flachdecken im Bereich von
Innenstützen. Doctoral thesis RWTH Aachen University, Germany; 2003. p.
369 [in German].
[39] Fernández Ruiz M, Muttoni A. Application of critical shear crack theory to
punching of reinforced concrete slabs with transverse reinforcement. ACI
Struct J 2009;106(4):485–94.
[40] Lips S. Punching of flat slabs with large amounts of shear reinforcement.
Doctoral thesis École Polytechnique Fédérale de Lausanne, Switzerland, Thèse
No. 5409; 2012. p. 217 doi:http://dx.doi.org/10.5075/epfl-thesis-5409.
[41] Lips S, Fernández Ruiz M, Muttoni A. Experimental investigation on punching
strength and deformation capacity of shear-reinforced slabs. ACI Struct J
2012;109(6):889–900.
[42] Keller T, Kenel A, Koppitz R. Carbon fiber-reinforced polymer punching
reinforcement and strengthening of concrete flat slabs. ACI Struct J
2013;110(6):919–27. http://dx.doi.org/10.14359/51686148.
[43] Lips S, Koppitz R, Kenel A, Keller T. New punching strengthening systems for
flat slabs. In: 4th Int. fib congress, Mumbai, India, February 10–14, 2014. p. 11.
[44] Chen WF. Double-punch test for tensile strength of concrete. Proc ACI J
1970;67(12):993–5.
[45] Marti P. Size effect in double-punch tests on concrete cylinders. ACI Mater J
1989;86(6):597–601.
[46] SIA 262:2013. Betonbau [Concrete Structures]. Swiss Society of Engineers and
Architects, Zürich, Switzerland; 2013. p. 102 [in German].
[47] Beyer K. Die Statik im Stahlbetonbau. 2nd ed. Berlin, Germany: Springer; 1956.
p. p. 804 [in German].
[48] Harajli MH, Soudki KA. Shear strengthening of interior slab-column
connections using carbon fiber-reinforced polymer sheets. ASCE J Compos
Constr 2003;7(2):145–53.
[49] Amsler M, Thoma K, Heinzmann D. Mit Aufbeton verstärkte Durchstanzplatte:
Versuch und Nachrechnungen. Beton- und Stahlbetonbau 2014;109(6):
394–402 [in German].