International Journal of Impact Engineering 71 (2014) 17e33

Contents lists available at ScienceDirect

International Journal of Impact Engineering

journal homepage: www.elsevier.com/locate/ijimpeng

Review

Assessing punching shear failure in reinforced concrete flat slabs

subjected to localised impact loading
K. Micallef a, J. Sagaseta a, *, M. Fernández Ruiz b, A. Muttoni b
a
Department of Civil and Environmental Engineering, University of Surrey, Guildford GU2 7XH, UK
b
IS-BETON, École Polytechnique Fédérale de Lausanne (ÉPFL), Lausanne CH-1015, Switzerland

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

Article history: Reinforced concrete flat slab structures are used widely in construction projects due to their economic
Received 2 September 2013 and functional advantages. Punching shear failure in such structures can have catastrophic effects in the
Received in revised form case of, for example, multi-storey framed structures and the designer aims to ensure that ductile flexural
27 February 2014
deformation occurs before the brittle shear failure.
Accepted 6 April 2014
Available online 19 April 2014
Shear mechanisms generally govern the behaviour of reinforced concrete structures subjected to
localised impact loads. Existing experimental results investigating punching shear in flat slabs subjected
to impact loading shows that when increasing the loading rate, the punching shear strength also in-
Keywords:
Punching shear
creases whereas the deformation capacity reduces. This behaviour is due to a combination of inertial
Flat slabs effects and material strain-rate effects which leads to a stiffer behaviour of the slab for higher loading
Impact loading rates. This can also lead to a change of mode of failure from flexural to pure punching shear with
Strain-rate effects increasing loading rates. Current empirical formulae for punching shear are unable to predict this
Critical shear crack theory behaviour since the slab deformations are not considered for calculating the punching shear strength.
This paper presents an analytical model based on the Critical Shear Crack Theory which can be applied
to flat slabs subjected to impact loading. This model is particularly useful for cases such as progressive
collapse analysis and flat slab-column connections subjected to an impulsive axial load in the column.
The novelty of the approach is that it considers (a) the dynamic punching shear capacity and (b) the
dynamic shear demand, both in terms of the slab deformation (slab rotation). The model considers in-
ertial effects and material strain-rate effects although it is shown that the former has a more significant
effect. Moreover, the model allows a further physical understanding of the phenomena and it can be
applied to different cases (slabs with and without transverse reinforcement) showing a good correlation
with experimental data.
Ó 2014 Published by Elsevier Ltd.

1. Introduction Fig. 1 gives typical ranges of strain-rates for different loading

sources.
1.1. High-rate loading Dynamic loads are often also idealised as being quasi-static or
replaced by equivalent static loadings but in the case of extreme
The design of most reinforced concrete (RC) structures is typi- events, such as blast or impact loading, such simplifications could
cally governed by ultimate limit state performance of the various be inadequate and further consideration is necessary [2]. Reliable
structural elements when subjected to static loading, e.g. dead (or structural modelling is essential to accurately predict the response
permanent) loads and live (or imposed or variable) loads. Whilst and damage in structures subjected to loads at high strain-rates.
the former are typically static in nature (e.g. structure’s self-weight, The effects of strain-rate effects can be considered on two levels,
finishes etc.), the magnitude of live loading tends to be variable viz. the effects on the material properties of the structure’s con-
with time (e.g. pedestrian or vehicular traffic loading). However, in stituents and the effects on the response of the structure itself.
most cases, such loadings can be idealised as quasi-static, since the
rate at which this loading is applied, typically described by the
strain-rate, 3_ , is of a very small magnitude. 1.2. Effect of strain-rate on material properties

* Corresponding author. Tel.: þ44 (0) 1483 686649; fax: þ44 (0) 1483 682135. The effect of strain-rate on the mechanical properties of most
E-mail address: j.sagaseta@surrey.ac.uk (J. Sagaseta). engineering materials is well-known. This includes the

http://dx.doi.org/10.1016/j.ijimpeng.2014.04.003
0734-743X/Ó 2014 Published by Elsevier Ltd.
18 K. Micallef et al. / International Journal of Impact Engineering 71 (2014) 17e33

Nomenclature Latin upper case

Ax projected contact areas in the X direction for an unit
crack area, [e]
Latin lower case Ay projected contact areas in the Y direction for an unit
a penetration constant, [T2] crack area, [e]
b penetration constant, [L2] A impactor section pressure, [M L1 T2]
b0 punching control perimeter, [L] Ec concrete elastic modulus, [M L1 T2]
c1 tension softening constant, [e] Es steel reinforcement elastic modulus, [M L1 T2]
c2 tension softening constant, [e] Gc concrete shear modulus, [M L1 T2]
cs0 slab initial damping co-efficient, [M T1] GF fracture energy, [M T2]
cs slab damping co-efficient, [M T1] GF0 reference fracture energy, [M T2]
dg concrete maximum aggregate size, [L] K material constant, [M1 L2 T2]
dg0 reference concrete aggregate size, [L] Leff slab effective span, [L]
dv shear-resisting slab effective depth, [L] MRd unit slab flexural capacity, [M L T2]
d slab effective depth, [L] Msd unit slab bending moment, [M L T2]
fc concrete compressive strength, [M L1 T2] N crack normal force component, [M L T2]
fc0 reference concrete compressive strength, [M L1 T2] P(t) contact force, [M L T2]
fct concrete tensile strength, [M L1 T2] _P loading rate, [M L T3]
fy steel reinforcement yield stress, [M L1 T2] P’ section limiting load, [M L T2]
g acceleration due to gravity, [L T2] Ri impactor radius, [L]
hi drop height of impactor, [L] S crack shear force component, [M L T2]
hs slab thickness, [L] V penetration velocity, [L T1]
kc contact stiffness, [M T2] V’ velocity factor, [e]
ks0 slab initial stiffness, [M T2] V0 impactor initial velocity, [L T1]
ks slab stiffness, [M T2] V* velocity constant, [L T1]
mi mass of impactor, [M] Vd shear force, [M L T2]
ms0 slab initial mass, [M] VR punching shear strength, [M L T2]
ms slab mass, [M] X penetration depth, [L]
rc column radius, [L] X€ penetration acceleration, [L T2]
rs position of zero bending moment with respect to
support axis, [L] Greek lower case
rs0 initial position of zero bending moment during contact b shear modulus retention factor, [e]
time, [L] g concrete dilatation angle, [e]
t time, [T] d crack separation, [L]
tc contact time, [T] 3_ strain-rate, [T1]
tE time to peak response, [T] m co-efficient of friction, [e]
ui(t) impactor displacement, [L] n Poisson’s ratio, [e]
u_ i ðtÞ impactor velocity, [L T1] r flexural reinforcement ratio, [e]
€
ui ðtÞ impactor acceleration, [L T2] rc concrete density, [M L3]
us(t) slab displacement, [L] rv shear reinforcement ratio, [-]
u_ s ðtÞ slab velocity, [L T1] sca concrete aggregate interlock normal stress, [M L1 T2]
€ s ðtÞ
u slab acceleration, [L T2] sct concrete tensile stress, [M L1 T2]
vc concrete shear wave velocity, [L T1] sp cement paste plasticisation stress, [M L1 T2]
w crack width, [L] sca concrete aggregate interlock shear stress, [M L1 T2]
wc maximum crack width, [L] 4(x) slab deformed configuration shape function, [e]
f reinforcement bar diameter, [L]
j slab rotation, [e]
jR slab rotation at failure, [e]

constituents of RC structures, namely the concrete and the steel These relationships are valid for strain rates up to 300/s covering
reinforcement. low to moderate impacts. The increases in compressive (fc) and
tensile (ft) strengths are given by (1) and (2) respectively as:

1.2.1. Concrete properties 8

 0:014
It has been shown by many researchers (e.g. Refs. [3e11]) that >
> 3_
>
> ; 3_  30=s
the tensile and compressive strengths of concrete both increase >
< 30  106
fc;dynamic
with loading rate. A very comprehensive review of experimental ¼ (1)
fc;static >
>  1
data in this respect has been carried out by Cotsovos and Pavlovic >
> 3_ 3
>
: 0:012 ; 30=s  3_  300=s
[12]. 30  106
The 1990 and 2010 Model Codes [1,13,14] provide relationships
which give the increase in strength and modulus with strain-rate.
 K. Micallef et al. / International Journal of Impact Engineering 71 (2014) 17e33 19

Fig. 1. Strain-rate ranges for various load sources (after [1]).

8 0:018 subjected to impact loading or air blast explosions in a shock tube.

>  3_
>
> Abbas et al. [34] also observed an increase in the load-carrying
>
> ; 3_  10=s
fct;dynamic < 1  106 capacity and stiffness of beams subjected to impact loading when
¼ (2) compared to the members’ static capacities.
fct;static >
>  1
>
> 3_ 3 Cotsovos et al. [30,35] concluded that in RC beams subjected to
>
: 0:0062 ; 10=s  3_  300=s
1  106 impact loading, the effect of the inertial loads on the response leads
to an increase in the member stiffness and its load-carrying ca-
An increase in the fracture energy, GF, with increasing loading pacity. This increase in stiffness is taken into account by considering
rate was also observed by other researchers [5,15e18], although as a reduced span which decreases with increasing loading rate, based
stated in Refs. [1,13] further work is needed in this area. on the member’s dynamic characteristics (shear wave speed and
limiting moment-carrying capacities) and the loading rate.
1.2.2. Steel reinforcement properties Thus, the increase in stiffness observed in such cases of dynamic
The effect of loading rate on metallic materials, such as steel, is loading are attributed to the delayed crack formation due to the
another well-known phenomenon and literature (e.g. Refs. [19,20]) enhanced tensile strength resulting from high strain-rate effects,
confirms the increase of steel reinforcement yield stress with the effects of inertial forces or a combination of the two
loading rate. Various material models, such as the Johnson and phenomena.
Cook [21] or the Cowper and Symonds [22] models are used to take
into account the dynamic effects in the constitutive behaviour of 1.3.2. Slabs
steels. Experimental and numerical work on RC slabs carried out by
Miyamoto et al. [25,26] demonstrated that shear mechanisms
1.3. Effect of high loading rates on structural response dominate the behaviour of RC structures subjected to impulsive
loads and an increase in loading rate is linked with an increase in
The loading rates which structures undergo as a result of severe the failure load, as shown in Fig. 3. Miyamoto et al. also proposed a
blast and impact threats can be significantly larger than those due failure envelope for slabs subjected to impact loading, shown in
to static loading as suggested by Fig. 1. Fig. 2. It was also reported a transition of failure mode from
Indeed, the Model Code 2010 [1,13] recommends that dynamic flexural-punching to punching shear with increasing the loading
effects such as mass and resonance effects are considered. It also rate as shown in Fig. 2.
suggests paying attention to damage mechanisms such as spalling Saito et al. [27] also observed an increase in the failure load of
and scabbing and, in particular, attention to formation of shear various RC structural systems subjected to high speed loading
plugs due to punching shear failure. when compared with the elements’ corresponding static load ca-
Since this paper is principally aimed at investigating the pacities. A change in the deformation and failure mode were also
response of RC flat slabs subjected to impact (rather than blast) observed.
loading, a brief review of experimental and numerical work related Delhomme et al. [28] observed from drop tests on RC slabs that
to this subjected is included in this section. during the very initial stage of impact when the impactor is in
Various researchers (e.g. Refs. [23e39]) have carried out work contact with the slab, there is a reduced radius of the slab which
related to drop impact loadings on RC beams and slabs and the shows bending behaviour. This leads to an increase in stiffness of
principal ones are discussed hereunder. the slab, which is similar to the phenomenon discussed by Cotsovos
et al. [30,35] and Hughes and Beeby [23].
1.3.1. Beams
Hughes and Beeby [23] observed that shear failure may occur in 1.4. Modelling dynamic structural response of RC structures
RC beams due to activation of higher modes under dynamic
loading. Without having to rely on costly large-scale testing of pro-
Saatci and Vecchio [33] carried out impact tests on RC beams totypes, the structural engineer often needs to be able to predict the
and concluded that shear mechanisms are typically critical in such load-carrying capacity of a RC structure subjected to dynamic
scenarios, even in the case of beams which are flexure-critical loading. The alternative methods are use of numerical techniques
under static load conditions. A similar observation was made by (e.g. non-linear finite element analysis) or analytical methods (e.g.
Ozbolt & Sharma [37] and Magnusson et al. [40,41], who have mass-spring models). The former are often difficult to employ in
shown that RC beams which failed in a ductile (flexural) manner practical applications due to their computational cost and software
under static loading changed to a brittle (shear) failure when package limitations in handling, for example, concrete cracking.
20 K. Micallef et al. / International Journal of Impact Engineering 71 (2014) 17e33

Fig. 2. Load-deflection response and failure envelope proposed by Miyamoto et al. (after [25,26]); slab dimensions 1.3  1.3  0.13 m with loading plate 150  150 mm.

Thus, the use of analytical and simple mass-spring models is energy transmitted to the slab are obtained. These are then used as
favoured for practical use and various models have been proposed, initial conditions for the post-contact model, where the slab un-
discussed hereunder. dergoes free vibration and its response is obtained. However, in this
Abdel-Rohman and Sawan [42,43] proposed various analytical case, the model parameters are obtained from experimental data
methods to estimate the dynamic response of RC slabs subjected to and/or numerical models of the impact test and no information on
impact loading, with good correlation with test data being capacity and/or failure is obtained from the model.
observed. The authors make use of the Petry formula to estimate Barbier and Roby [46] used a similar concept where the mass
the impact force and contact time. The Petry formula, originally and stiffness are functions of time but the model parameters are
developed in 1910, is given by: obtained from curve fitting to numerical results.
In addition, these models provide only information on the
X ¼ KARV ’ (3) response (i.e. the shear demand) and there is no information about
the shear strength capacity or whether failure occurs or not.
where X is the penetration of the impactor into the slab, K is a
material constant given as 2.97  102 m3/kN for reinforced con- 1.5. Aim of current study
crete, A is the impactor section pressure, R is a penetration ratio and
V’ is a velocity factor given by: Thus, the aim of the current study is to develop a model which
can be used to predict the dynamic response of RC slabs subjected
  2 
V to impact loading which does not rely on obtaining model pa-
V ’ ¼ log 1 þ 0* (4) rameters from experimental results or numerical simulations and
V
which follows an integral approach in which both the loading on
where V0 is the initial velocity of the impactor and V* is a constant the structure (shear demand) as well as its capacity (shear strength
given by 141.33 m/s. Further discussion of this formula, its pa- supply) are predicted.
rameters and its use is given in Section 3.1. Consideration of strain-rate effects will be made in both aspects
Historically, the CEB Bulletin 187 [44] was an early attempt in of the problem and this will lead to assessing the relative signifi-
modelling the response of RC slabs subjected to impulsive cance of the strain-rate effects and the initial increased stiffness of
loading and potential punching failure by a mass-dashpot-spring the structure associated with inertial effects.
model, as shown in Fig. 3. This model assumes two masses, one The model can then be used to predict the slab’s behaviour
local mass corresponding to the impactor and the punching shear and likelihood of failure due to punching shear and also provide
cone and another mass corresponding to the slab. The main
limitation of this model is that it assumes that the formation of
the punching cone is governed by the tensile stress in the con-
crete, dowel action and shear resisted by the stirrups whereas in
reality the punching strength is also influenced by the slab
deformation [13]. This implies that the stiffness R2 (Fig. 2) de-
pends on the deflection w1 and not only on (w2  w1) as assumed
in Ref. [44]. In addition, for cases where failure does not occur
even if the tensile capacity is exceeded, the CEB Bulletin attri-
butes this fact to the scatter in concrete’s tensile strength, rather
than to the stiffness and/or strength increase due to dynamic and
strain-rate effects.
A significantly improved model to assess the flexural response
of RC slabs during and after impact was proposed by Delhomme
[28,45], where the response was assumed to be split into two
phases, viz. the contact phase and the post-contact phase.
In the former, the slab mass and stiffness are based on a reduced
radius (thus increased stiffness), from which the contact load and Fig. 3. CEB 187 model for punching shear failure (after [44]).
 K. Micallef et al. / International Journal of Impact Engineering 71 (2014) 17e33 21

information about the levels of deformation at failure. The

application is illustrated by various case studies RC slabs with
and without transverse reinforcement and low to moderate
impacts.
It should be noted at this point that the source of impact load-
ings being considered in this paper is due to impactors which are
relatively rigid and thus non-deformable (e.g. impulsive axial load
in column such as in a progress collapse scenario, falling debris on a
slab, falling objects on a slab and rock falls); such impact loadings
are termed as ‘hard impacts’ in the CEB Bulletin 187 [44] and in this
case it is assumed that the kinetic energy imparted by the impactor
is entirely absorbed by the deformation of the struck body, i.e., the
RC slab.

1.6. Presentation of current study

Following this introduction, the fundamentals of the critical Fig. 4. Critical shear crack through compression strut (after [51]).
shear crack theory will be discussed and how this is extended to
take into account the effects of strain-rate on punching shear ca- VR 0:75
pacity. Section 3 will present a mass-spring-damper model to pffiffiffiffi ¼ jd
(5)
b0 dv fc 1 þ d15þd
predict the flexural deformation response of a RC slab subjected to g0 g

an impact load and the use of this model to investigate various slab
systems will be presented in Section 4. Finally, the main findings where VR is the punching shear capacity, d is the effective depth of
and merits of this paper will be summarised in Section 5. the slab, b0 is the control perimeter at ðd=2Þ from the edge of the
support, dv is the shear-resisting effective depth, dg is the maximum
aggregate size in the concrete and dg0 is a reference aggregate size
2. Dynamic punching shear strength taken as 16 mm.
Knowing the load-rotation response of a particular slab, then
2.1. The critical shear crack theory the punching shear capacity, VR, and the deformation (rotation) at
failure, jR, can be immediately established by the intersection of
Punching is a brittle form of failure observed in RC flat slab the two curves, as shown in Fig. 5.
structures, typically at slab-column connections but also observed In the case of slabs with transverse (shear) reinforcement, the
in many drop impact tests. contribution from the steel reinforcement (which varies with
The punching capacity of slab systems has been investigated rotation due to the varying crack width and thus varying stress in
since the 1950s but most strength models presented in design the shear reinforcement) is added to the concrete contribution,
codes are empirically derived e.g. the American ACI 318-08 [47], the using expressions given in Refs. [1,13,53] to give a new failure cri-
British BS 8110-1 [48] and the European EC 1992-1 [49]. terion, as shown schematically in Fig. 6.
A physically-based mechanical model was proposed in 1988 by The aim is to extend this formulation to take into account the
Muttoni [50] and subsequently formed the basis for punching shear dynamic enhancement of the concrete properties, as discussed in
provisions in various Swiss codes and also in the latest version of Section 1.2.1, by considering the two main constituents of punching
the Model Code [1,13]. capacity in turn (i.e. tensile stress in the concrete and aggregate
The model is based on the critical shear crack theory (CSCT) and interlock). The dynamic effects on the load-rotation response will
assumes that the shear strength is governed by the width and the then be considered in Section 3 to produce a plot similar to Fig. 5 (or
roughness of a shear crack which develops through an inclined Fig. 6) for the dynamic case.
compression strut which carries the shear force, as shown in Fig. 4.
Assuming that the crack width, w, is proportional to the slab
2.2. Influence of strain-rate on fracture toughness
rotation, j, the shear strength is calculated from a set of assumed
kinematics characterised by the rotation of the slab and by inte-
Concrete subjected to tension initially follows a linear elastic
grating the contribution of the concrete tensile stresses and the
stressestrain response, which response is limited by a stress value
aggregate interlock along the failure surface. Most of the shear
stress is transferred at the bottom end of the crack where the crack
width is small while any contribution from dowel action of the
reinforcement is ignored due to the expected spalling of the con-
crete cover.
It can be shown that the punching shear capacity decreases with
increasing rotation since this implies wider cracks, thus reducing
both tensile and aggregate contributions. Further details can be
found in Muttoni and Fernández Ruiz [51,52] for slabs with no
transverse reinforcement and Fernández Ruiz and Muttoni [53] for
slabs with transverse reinforcement, as well as for non-
symmetrical cases investigated by Sagaseta et al. [54].
For design purposes, rather than evaluating the contributions of
concrete tension and aggregate interlock discretely for individual
cases, a failure criterion was proposed [51] to cover most practical
geometric and material configurations, given in SI units (N, mm) by: Fig. 5. Evaluation of punching shear capacity (unreinforced).
22 K. Micallef et al. / International Journal of Impact Engineering 71 (2014) 17e33

qffiffiffiffiffi the variation of GF with fc and noting that

Knowing
fct ¼ 0:3 3 fc2 , it is possible to express the variation of GF and wc
with strain-rate, by using the rate-dependent concrete strengths
given by (1) and (2) with fc in MPa. These are given by (10) and (11)
respectively.
(
0:221GF0 fc0:7 3_ 0:0098 ; 3_  30=s
GF ð_3 Þ ¼ (10)
0:103GF0 fc0:7 3_ 0:2333 ; 30=s  3_  300=s

8
> 1
>
>
>
< 2:97G f 30 3_ 0:0082 ; 3_  30=s
F0 c
wc ð_3 Þ ¼ (11)
>
> 1
>
>
:
2:85GF0 fc30 3_ 0:1 ; 30=s  3_  300=s
Fig. 6. Evaluation of punching shear capacity (reinforced).
The dynamic increase factor for the fracture toughness GF/GF0
given by expression (10) is of similar magnitude to that obtained by
of fct, after which a crack forms. Following the formation of the Oh [18]; for concrete strengths around 30 MPa the difference be-
crack, softening behaviour occurs, until a maximum crack separa- tween the predictions of GF/GF0 given by both approaches is 20%.
tion, wc, is reached when there is complete separation of the two Using these relationships and equating (8) to (9), it is also
crack faces. This is further explained by Hillerborg et al. [55]. possible to obtain strain-rate-dependent values (_3 in 1/s) for the
Various models have been proposed to describe the softening constants c1 and c2 as:
behaviour. One such model is that put forward by Hordijk [56], who
gives an expression for the stressedisplacement variation after the c1 ð_3 Þ ¼ 6:8  0:0061_3 (12a)
initial crack formation as:
c2 ð_3 Þ ¼ 2:7  0:0071_3 (12b)
  3
w c w
 w2 c w
sct =fct ¼ 1 þ c1 e  1 þ c31 ec2 (6) Fig. 8 shows the effect of strain-rate on the tensile strength and
wc wc fracture toughness as described by the rate-dependent form of (6)
for various strain-rate values ranging from 10 to 300/s, representing
where c1 and c2 are constants given by 3 and 6.93 respectively.
low to moderate impact loadings.
Hordijk also gives an expression for the crack separation
displacement as:
2.3. Influence of strain-rate on aggregate interlock
GF
wc ¼ 5:14 (7) After concrete cracks, it nonetheless still has the capacity to
fct
transmit shear (sca) and normal (sca) stresses across the crack sur-
The energy dissipated in opening the crack, GF, or fracture en- faces by means of aggregate interlock action, described graphically
ergy, is given by the area underneath the stressedisplacement in Fig. 9.
curve, or the integral of (6) over wc as: Following the initial crack opening, w0, the displacements of the
crack surfaces in terms of the successive crack translation, d, are
" (  3 ) (
given by:
1 c1 c2 1 1 3 6
GF ¼ fct wc 1þ6 e þ c31 þ þ
c2 c2 c2 c2 c22 c32 w ¼ w0 þ d sin g (13a)
! )#
6 1 3
þ 4 þ 1 þ c1 v ¼ d cos g (13b)
c2 2
(8) where g is the dilatation angle taken as 27 [57].
These are shown in Fig. 10.
For fc not exceeding 80 MPa, Model Code 1990 [14] gives an
expression for GF as:

 0:7
fc
GF ¼ GF0 (9)
fc0

where GF0 is a reference fracture energy (which is a function of dg)

and fc0 is a reference compressive strength, given as 10 MPa.
It should be noted that Model Code 2010 [1,13] gives a simplified
expression in which GF varies only with fc and constant with dg, as
shown in Fig. 7. In this paper, the expression found in Model Code
1990 will be used, since this, in the opinion of the authors, this is
more general.
While the expression given by (6) was proposed by Hordijk
based on the static response of concrete, Weerheijm and Van
Doormaal [9] postulate that the same equation can be used to
describe the softening behaviour even in the dynamic case. Fig. 7. Variation of GF with fc.
 K. Micallef et al. / International Journal of Impact Engineering 71 (2014) 17e33 23

Fig. 10. Initial and final crack translations (after [57]).

Fig. 8. Variation of sct with 3_ .

carried out by Lui [59]. For unreinforced specimens loaded at
10,000 MPa/s (or 3_ approximately equal to 0.3/s), Lui measured a
The aggregate interlock strength component can be evaluated limiting strength of 8.75 MPa. Using Walraven’s model and
by using various physically-based or empirically-derived models. applying the strain-rate enhancement to sp, the maximum stress
In this paper, this is done by using the physical model proposed predicted is 7.36 MPa, which is a very good approximation to the
by Walraven [58] and extended by Guidotti [57], where the two measured value.
stress components are evaluated along the unit crack area as: The small difference between the predicted and observed
 strengths could be attributed to the fact that the value of m adopted
sca ¼ sp Ax  mAy  0 (14a) is the same as for the static case. Although it is debatable as to
whether m also increases with 3_ , in the lack of a statistically-
 significant result database, it is not possible to confirm this asser-
sca ¼ sp mAx þ Ay (14b) tion at this stage. Thus, a value of m ¼ 0.4 will be used throughout
this paper.
where m is the co-efficient of friction between the aggregate
granules and the cement paste (taken as 0.4) and sp is the cement
paste plasticisation stress in MPa, given by: 2.4. Dynamic punching shear failure criterion

sp ¼ 6:39fc0:56 (15) Having established strain-rate-dependent expressions for the

By idealising the aggregate particles as spheres, as shown in two main components of punching shear capacity, it is now
Fig. 11, the projected contact areas, Ax and Ay , are derived (Ref. [58]) possible to obtain the capacity for any slab configuration.
from the probability distributions of an aggregate particle of a It has been shown [51] for the static case that 99 experimental
specific size (limited by dg) being present along the crack. results fall within a very narrow region covering many practical
The crack contact surface areas are obtained by numerical slab and column geometries and slab reinforcement ratios, as
means for various crack kinematics describing different crack shown in Fig. 12.
widths. Further details can be found in Refs. [57] and [58]. Experimental results from various tests at high-rate loading also
Assuming that sp varies with strain-rate according to (1), then the suggest that the results fall within a similar relatively narrow fail-
dynamic aggregate interlock strength can be evaluated for various ure region which however lies above the average failure criterion
cases. given by (5). This is shown in Fig. 13.
Thus, it is necessary to establish such a region for the dynamic
case and propose a dynamic failure criterion similar to (5).
2.3.1. Comparison with test data For this purpose and knowing the relative sensitivity of the
The extension of Walraven’s aggregate interlock model taking various parameters from Refs. [57], three slab configurations (all
into account strain-rate effects is verified by comparison with test having fc ¼ 30 MPa and a cover of 25 mm) were chosen, outlined in
results from experiments carried out on dynamic push-off tests Table 1.

Fig. 9. Aggregate interlock action: (a) initial crack opening due to tension; (b) longitudinal sliding reinstating contact between crack faces; (c) crack kinematics and generation of
normal and shear stresses.
24 K. Micallef et al. / International Journal of Impact Engineering 71 (2014) 17e33

Fig. 11. Idealisation of aggregate particles into spheres (after [58]).

Thus, the total shear force that could be carried by the crack was
found:

NþS
VR ¼ pffiffiffi (17)
2
An illustration of the force variation with crack translation is
shown in Fig. 15.
Thus, for each value of rotation, the maximum shear force that
the slab can carry is found, enabling a plot similar to Fig. 12 to be
derived. The resulting failure region (for the static) case matches
closely the region shown in Fig. 12.
The procedure was repeated for different 3_ values (10/s, 100/s
and 300/s) and the results are shown in Figs. 16 to 18.
These results clearly suggest that the punching shear strength
increases with strain-rate, with a particular increase observed at
low rotation values and most significant at the larger strain-rate
Fig. 12. Punching shear failure region for static loads as a function of critical shear values. From these results, a modified expression for the failure
crack width (after [51]). criterion is proposed in each case, viz.:

VR 0:8
Assuming a crack inclination angle of 45 , a discrete crack model pffiffiffiffi ¼ for_3 ¼ 10=s (18)
jd
was derived for each case. In each case, for different slab rotation b0 dv fc 1 þ d15þd
g0 g
values and crack separations, the normal and shear forces along the
crack surface (shown in Fig. 14) were evaluated using:
Z VR 1
pffiffiffiffi ¼ jd
for_3 ¼ 100=s (19)
N ¼ ðsct þ sca ÞdA (16a) b0 dv fc 1 þ d15þd
g0 g

Acr

Z VR 1:3
pffiffiffiffi ¼ for_3 ¼ 300=s (20)
S ¼ sca dA (16b) b0 dv fc jd
1 þ d15þd
g0 g
Acr
Although further experimental data is required to verify the
above proposed failure criteria, the results suggest that while at the
lower strain-rates the increase from the static case is only by 7%, for
strain-rates of 100 and 300/s, the respective increases are 33% and
73%, which are very significant. For strain-rates higher than 300/s,
which can be the case in ballistic problems, the strain-rate
dependent relationships for the materials used in the model and
the contribution of aggregate interlock would need to be reviewed
to consider additional mechanisms such as aggregate crushing.

Table 1
Slab scenarios considered for discrete crack models.

Case 1 Case 2 Case 3

r (%) 0.5 1 1.5

f (mm) 12 16 25
h (mm) 200 200 250
rc (mm) 150 150 200
dg (mm) 16 16 32
Fig. 13. Punching shear failure results from various high-rate tests.
 K. Micallef et al. / International Journal of Impact Engineering 71 (2014) 17e33 25

Fig. 14. Punching shear strength components.

Fig. 17. Failure regions for Cases 1e3 (static and 100/s).

3.1. Modelling the impact event

Fig. 15. Strength contributions from tension and aggregate interlock (Case 1, slab
rotation ¼ 0.001). In this work, it is assumed that the loading source is an impactor
of known mass, mi, and radius, Ri, which is dropped from a known
height, hi, thus having an initial velocity equal to 2ghi from simple
energy conservation considerations.
3. Prediction of dynamic flexural response
In order to determine the impact force, P(t), imparted to the slab,
it is necessary to determine the contact time, tc, and the accelera-
The discussion in Section 2 was concerned with the shear ca-
tion of the impactor’s penetration into the slab, from which the
pacity of flat slabs in the case of dynamic loading.
contact force is then easily obtained.
In this section, the response of a slab when subjected to an
From Newton’s second law of motion:
impact loading will be investigated. Thus, the slab’s displacement
(and hence rotation) are obtained and the load-rotation response is
€ ¼ mV dV
used in conjunction with the failure criterion discussed in Section 2. PðtÞ ¼ mi X i (21)
dX

where X€ is the acceleration of the impactor’s penetration and V is

the velocity of penetration.
Using the Petry formula described by (3) and using (4), then it
can easily be shown through some algebraic manipulation that (21)
may be written as:

€ ¼ aXebX 2
X (22)

where the constants a and b are given by:

2
2:3pRi V *
a ¼ (23a)
mi KR

4:6pRi
b ¼ (23b)
mi KR
It is obvious that (22) is a non-linear non-homogenous ordinary
differential equation and it can be solved by numerical means (in
this work, the fourth-order RungeeKutta method was employed),
pffiffiffiffiffiffiffiffiffi
ffi
subject to the initial conditions X ¼ 0 and X_ ¼ V0 ¼ 2ghi at t ¼ 0.
The contact time is represented by the time required for X_ to
become zero.
The constant R in (3) and (23) is the penetration ratio, repre-
senting the penetration into a slab of finite thickness to the pene-
Fig. 16. Failure regions for Cases 1e3 (static and 10/s). tration into a slab of infinite thickness.
26 K. Micallef et al. / International Journal of Impact Engineering 71 (2014) 17e33

Fig. 19. Two-phase model: (a) Contact phase; (b) Post-contact phase.

These two phases are described in further detail in Sections 3.2.1

and 3.2.2 respectively.

3.2.1. Contact phase

The various model parameters are outlined below.
Fig. 18. Failure regions for Cases 1e3 (static and 300/s).

e Slab stiffness
Thus, R ¼ 1 represents an infinitely thick slab. Abdel-Rohman
and Sawan [42,43] suggest using R ¼ 1.1 or 1.2 for typical RC slab As discussed in Section 1.3, experimental and numerical results
thicknesses to obtain more realistic results. suggest that during the initial response of a slab subjected to an
This was verified by means of the following parametric study impact event, an increased stiffness is observed. In this paper, the
based on results reported by Delhomme et al. [28,45]. Delhomme increase in stiffness is modelled by considering a reduced slab ge-
et al. dropped a 450 kg mass from a height of 30 m on a 0.28 m thick ometry or radius, rs0, on the basis of which the flexural stiffness is
slab and measured a contact time of 3.5 ms and a maximum contact computed.
force of 5 MN. Modelling this loading scenario, the contact time and The evaluation of rs0 is based upon the Leff concept proposed by
force were evaluated for various values of R, summarised in Table 2. Cotsovos et al. [30,35], whereby it is assumed that immediately
It can be observed that a good correlation with experimental after contact in the case of high-rate loading, a reduced portion of
results is obtained for the case of R ¼ 1.1, suggesting that the Petry the full span reacts to the applied load. While Cotsovos developed
formula can be used to provide a simple yet accurate method of the concept for RC beams, similar observations were made by
estimating the contact time and contact force history for typical RC Delhomme et al. [28,45] for the case of RC slabs.
slab applications. This reduced span (termed Leff by Cotsovos) is limited by the
load-carrying capacity based on the lower part of the RC section
(MRd) and upper part of the RC section ðMRd ’ Þ, suggesting that the
3.2. Modelling the slab response
slab behaves essentially as a fixed-ended member irrespective of its
actual boundary conditions, as shown graphically in Fig. 20.
Knowing the contact time, it is now possible to develop a
The moment capacity for a RC section is obtained from simple
simplified mass-spring-dashpot model to simulate the impact on
equilibrium and is given by:
an RC flat slab. For this purpose, a two-phase model is proposed,
based on the model developed by Delhomme et al. [28,45] and  
rfy
shown graphically in Fig. 19. MRd ¼ rd2 fy 1  (24)
2fc
In the initial phase, the impactor is in contact with the slab
between t ¼ 0 and t ¼ tc. The slab parameters are based on an initial where r is the reinforcement ratio at the level being considered
reduced slab radius, leading to an increased geometric stiffness, ks0, (top or bottom) and fy is the steel reinforcement yield stress.
and a reduced slab mass, ms0. The dashpot, cs0, represents the slab’s
dissipating action while the contact stiffness, kc, is set such that
ui(tc) ¼ us(tc). At the end of this phase, the slab displacement and
energy are obtained and these are used as initial conditions for the
next phase, during which the slab mass, ms, stiffness, ks and
dashpot, cs, are based on the actual slab geometry, giving the
complete slab response.

Table 2
Contact results for various R values.

R tc Pmax
(ms) (MN)

1.0 3.41 5.14

1.1 3.57 4.91
1.2 3.73 4.7
Fig. 20. Definition of Leff (after Cotsovos [30,35]).
 K. Micallef et al. / International Journal of Impact Engineering 71 (2014) 17e33 27

The limiting moment capacity at the top of the section is either

the moment capacity which can be generated by any reinforcement  
3 rs fy Msd 3=2
present at the top of the section as determined by (24) or the j¼ (31)
2 d Es MRd
section’s cracking moment (Mcr) for the case of a section of thick-
ness hs with no top reinforcement, given by:
where rs indicates the position of zero radial bending moment with
respect to the support axis (typically taken as z0.22L for flat slabs
fct h2s of span L), Es is the elastic modulus of steel reinforcement, Msd is the
Mcr ¼ (25)
6 average bending moment per unit length in the slab’s column
Defining the velocity, vc, at which shear waves travel through (support) strip and MRd is the average flexural strength per unit
concrete as: length in the column strip, given by (24).
The various terms in (31) affect the crack width (and thus the
sffiffiffiffiffiffi
rotation). The term ðrs =dÞ represents the slenderness of the slab
Gc
vc ¼ (26) while the term ðMsd =MRd Þ is the bending moment demand ratio.
rc The strain in the reinforcement at yielding is considered by the
term ðfy =Es Þ.
where Gc is the concrete shear modulus and rc is the concrete For internal columns, from the Model Code [1,13], Msd is related
density, it is possible to find the time, Dt, required to reach the to the load Vd by:
limiting moment capacities as per Fig. 20 as:

Leff Vd
Dt ¼ (27) Msd ¼ (32)
2vc 8
Substituting (32) and the appropriate value of rs0 into (31) leads
’ to be generated at the slab’s
Defining P’ as the load required for MRd to a simple expression for load-displacement and hence its gradient
upper face, then. to give the slab stiffness, ks0.

8MRd’
e Slab mass
P’ ¼ (28)
Leff
From the computed Leff, it is also possible to work out the
If P_ is the loading rate (obtained from the Petry formula as effective mass of the slab for the initial phase, by assuming the
described in Section 3.1), then P’ is simply given by P_ Dt. Using this slab’s deformed shape to be similar to that of a built-in section and
expression, (27) and (28), Leff is found as. using this as a shape function for the evaluation of the generalised
mass:
’ v
MRd c
Leff ¼ (29) !2 !3
P_ x x
4ðxÞ ¼ 12  16 (33)
Thus, from (29), Leff f1=P_ for any given section, confirming Leff Leff
physical observations that the stiffness is observed to increase (i.e.
Leff decreases) with increasing loading rate. The mass is then worked out using:
From the value of Leff, rs0 can easily be found purely from geo-
metric considerations. Two limiting cases are identified, viz. the Leff

case with no top reinforcement and the case where the top and
Z2p Z 2
bottom reinforcement are equal, shown in Fig. 21. In the latter case, ms0 ¼ 2prc h2s ½4ðxÞ2 dx dq (34)
the limiting moment is Mcr  MRd and thus may be ignored. 0 0
This leads to the following range for rs0:
Substituting (33) into (34), then it can easily be shown that
ms0 z2800h2s Leff assuming rc ¼ 2400 kg/m3, which is typical for
Leff
 rs0  Leff (30) normal weight concrete.
2
Model Code 2010 [1,13] gives expressions for the load-rotation e Slab damping
response of a RC slab by using the levels of approximation
method [60]. In this instance, a Level II approximation is used, The damping parameter, cs0, is obtained using the standard
whereby the load-rotation relationship is given by: expression from structural dynamics as:

Fig. 21. Reduced slab span: (a) Equal top and bottom reinforcement; (b) No top reinforcement.
28 K. Micallef et al. / International Journal of Impact Engineering 71 (2014) 17e33

sffiffiffiffiffiffiffiffiffi
exceed around 2 m), the transition time is unlikely to exceed a
ks0
cs0 ¼ 2zms0 (35) duration of the order of 2 ms.
ms0 Similarly, ms is determined using (34) with the shape function
based on the static deformation of the slab. From (38), the slab
where z is the structure’s damping co-efficient, which depends on response is determined and thus the full load-rotation response can
the structural type. be established.
Further discussion on the value of this parameter is done in The entire procedure is summarised in Appendix A.
Section 4.

e Contact stiffness 3.2.3. Strain-rate

From the displacements obtained using the two phase model
Finally, the contact stiffness, kc, is set such that the impactor’s described, the slab rotations can easily be established by assuming
and slab’s displacement are equal at t ¼ tc and by assuming the that the slab has a conical form beyond the critical crack [51,52] and
Hertz contact law given by: thus the rotation can be evaluated by considering the displacement
at any point x away from the critical shear crack.
Fc ¼ kc ðui  us Þ3=2 (36) Thus, the crack width, w, can be determined from the funda-
The value of contact stiffness is obtained by an iterative process mental assumption of the CSCT:
until it has a value such that the above condition is satisfied.
wfjd (40)
e Governing equations of motion In this paper, it is assumed that the proportionality constant is
given by 0.5, as suggested by Fernández Ruiz and Muttoni [53].
Having established all the mass, spring and damping parame- Thus, the strain at the crack (assuming the reinforcement is
ters, it is now possible to write the governing equations of motion within the elastic domain) can be established by using a strain-
for the first phase: crack width relationship, such as the square root model sug-
( gested by Fernández Ruiz et al. [61]:
€ i þ gÞ þ kc ðui  us Þ3=2 ¼ 0
mi ðu 0 12=3
(37)
€ s þ cs0 u_ s þ ks0 us  kc ðui  us Þ3=2 ¼ 0
ms0 u
B3sb;max jdC
3 ¼ @ pffiffiffiffiffiffiffiffiffi A (41)
The system of equations shown in (37) is solved by numerical 2f Es fy
means (e.g. fourth-order RungeeKutta pffiffiffiffiffiffiffiffiffiffi method), subject to the
initial conditions us ¼ 0 and u_ i ¼ 2ghi at t ¼ 0. 2=3
where sb,max is the maximum bond stress given by fc . The square
At the end of the contact phase, the slab displacement us is
root model for bond is based on the affinity hypothesis of the slip
obtained and this is used as an initial condition for the post-contact
distribution along long anchored bars and neglects the local loss of
phase. In addition, the strain and kinetic energy in the slab are also
bond stiffness and strength due to the formation of local diagonal
evaluated from which the slab velocity is obtained.
cracks [61].
The strain-rate 3_ can be determined from (41) as:
3.2.2. Post-contact phase 0 12=3
In the post-contact phase, the slab is assumed to undergo free 
d B 3sb;max d C 2=3
vibration subject to the initial conditions obtained at the end of the 3_ ¼ ð3 Þ ¼ @ pffiffiffiffiffiffiffiffiffiA j_ (42)
dt 2f Es fy
contact phase. The governing differential equation, as per Fig. 19(b),
is simply given by:
Knowing the strain-rate from (42), it is possible to evaluate the
€ s þ cs u_ s þ ks us ¼ 0
ms u (38) corresponding effect on the failure criterion, as discussed in Section
2.4, and also to adjust fc and fy while establishing the slab load-
The slab load-rotation relationship (hence ks) is obtained using
(31) and (32) but the slab radius rs now increases from rs0 to the
actual slab dimension. This increase is assumed to occur linearly
over time at a velocity of vc given by (26). A word on the value of Gc
is in order here. The elastic value of Gc is given from classical me-
chanics as:

Ec
Gc ¼ (39)
2ð1 þ nÞ

where Ec is the concrete elastic modulus (varying from 30 to 40 GPa

for typical fc values [1]) and n is the Poisson’s ratio (equal to 0.2 [1]).
Thus, from (39), Gc has values typically ranging from around 12e
17 GPa.
As the concrete cracks, the modulus decreases and hence Gc
decreases by a factor b which reduces from 1 to 0 depending on the
crack width. This reduction of Gc affects vc, according to (26), which
could delay or interrupt the transition from rs0 to rs. However, the
effect on vc is not significant, with the wave speed still remaining
over 1000 m/s even for low Gc values, as shown in Fig. 22. This
implies that in typical slab applications (where rs is expected not to Fig. 22. Variation of yc with Gc.
 K. Micallef et al. / International Journal of Impact Engineering 71 (2014) 17e33 29

Fig. 25. Load-rotation response and failure criterion for T5.

Fig. 23. Effect of increasing material properties due to strain-rate on slab stiffness.
whereas the yield strength of the flexural reinforcement was
fs ¼ 500 MPa. The slabs were subjected to impact loadings from a
450 kg mass from a height of 30 m and 15 m (T5 and T1).
rotation curve. However, the effects of strain-rate on the slab
stiffness are found to be small (although not insignificant), as
e Test T5
illustrated by Fig. 23 for a typical flat slab structure (hs ¼ 280 mm,
fc ¼ 30 MPa, r ¼ 0.7%). For the case of 3_ ¼ 0.1/s. the increase in
The displacementetime history was recorded and this was
stiffness is by 4% while for 3_ ¼ 300/s, the increase is almost 15%.
compared with the numerical result, as shown in Fig. 24.
At this point, a comment on alternative methods of estimating
Good correlation in terms of the maximum displacements and
the strain-rate is in order. A simplified equation is presented in the
also temporally was observed, with the maximum predicted
American UFC 3-340-02 [62] to estimate the strain-rate in the
displacement being 28 mm and the maximum recorded displace-
concrete as:
ment being 23 mm. The differences in the post-contact stage are
0:002 expected as the analytical model neglects any bouncing effect be-
3_ ¼ (43)
tE tween the impactor and the slab after impact (i.e. relative move-
ment between the impactor and the slab is assumed to be zero).
where tE is the time duration from the initial impact to the peak As described in Section 3.1, the contact time and force were also
response. The results obtained using (42) and (43) will be discussed accurately correlated.
and compared in Section 4.3. From the displacements, the rotations were extracted and the
load-rotation curve was assembled. The failure criterion from (5)
was adjusted to take into account the strength contribution of the
4. Case studies
shear reinforcement, using the method proposed in Ref. [53]. Thus,
the shear demand and strength are plotted together as shown in
In this section, the use of the integral model proposed in this
Fig. 25. In this instance, the strength and stiffness are based on the
paper, as described in Sections 2 and 3, will be illustrated by
static values without any adjustment for strain-rate effects.
assessing various impact loading scenarios on RC slabs.
It can be observed that during contact, the stiffness is consid-
erably higher than the post-contact phase.
4.1. Slabs with transverse reinforcement In addition, both during loading and the free vibration phases,
the maximum shear demand does not exceed the failure criterion,
In this section, the tests carried out by Delhomme et al. [28,45] albeit that the slab is on the cusp of failure. Delhomme et al. record
will be investigated. Delhomme tested 12  4.8  0.28 m slabs
reinforced top and bottom equally with r ¼ 0.71% and having shear
reinforcement rv ¼ 0.36%. The concrete used had a compressive and
tensile strengths of fc ¼ 40.5 MPa and ft ¼ 5.8 MPa respectively

Fig. 24. Numerical and experimental displacementetime histories for T5. Fig. 26. Effect of strain-rate on load-rotation response and failure criterion for T5.
30 K. Micallef et al. / International Journal of Impact Engineering 71 (2014) 17e33

Fig. 29. Load-rotation response and failure criterion for T1.

A similar comparison between experimental results and nu-

merical predictions was done in this case, with the correlation
being as shown in Fig. 28.
The load-rotation and failure criterion are shown in Fig. 29.
Fig. 27. Effect of Y parameter on displacementetime history for T5.
Since a strain-rate of around 0.08/s was predicted, the static values
were used for the load-rotation curves, on the basis of the discus-
severe concrete cracking, indicating that concrete capacity is sion earlier. Clearly, the maximum shear force is less than the ca-
exceeded. However, complete punching was prevented by the pacity in this case.
shear reinforcement. Comparing Figs. 29 and 25, it can be observed that in the case of
the higher drop height (T5), the stiffness is significantly higher,
matching experimental observations. It can also be seen that the
4.1.1. Strain-rate effect
maximum rotation predicted is also higher in the case of the higher
As discussed in Section 2.4, strain-rate has a favourable effect on
drop height (or larger contact force).
the failure criterion (thus the shear capacity) and also on the slab
stiffness. In this case, using the procedure described in Section
3.2.3, the maximum strain-rate was estimated, using (42), to be 0.1/
4.2. Slabs with no transverse reinforcement
s for this case. These effects are illustrated in Fig. 26.
It can be observed that, in this case, there is little modification
In this section, the tests carried out by Bhatti et al. [36] will be
due to strain-rate effects to the load-rotation curve and the failure
investigated.
criterion and using the static curve for the latter will provide a
Bhatti tested 1.65  1.65  0.15 m slabs reinforced at the bottom
conservative and accurate estimate of the slab’s capacity.
with r ¼ 0.6% and having no reinforcement at the top and also no
transverse reinforcement. Only control slabs without FRP strips are
4.1.2. Damping ratio considered in this work. The slabs were subjected to impact load-
The only parameter which was not explicitly derived in the two ings from a 300 kg mass at 4 m/s and 5 m/s (N-IS-4 and N-IS-5). In
phase model is the damping parameter (cs0 and cs), which depends both cases, punching shear failure was observed, as shown in
on the choice of z. Fig. 27 shows the effect of the z on the response, Fig. 30.
suggesting that the correlation between the experimental and
numerical results is best achieved with z ¼ 10%. Indeed, Delhomme e Test N-IS-4
et al. measured z ¼ 11%.
Whilst the selection of z remains a judicious choice to be done The numerical prediction for this case is shown in Fig. 31. In this
by the engineer, based on experience on similar structures, it is case, the maximum strain-rate was estimated using (42) to be
suggested that the value of z is taken between 5 and 10%. 0.025/s and thus the static values were retained for both the stiff-
ness and failure criterion.
e Test T1 The prediction is that the slab fails due to punching shear and
this was confirmed by the test, as shown in Fig. 30(a).

e Test N-IS-5

Fig. 30. Slab cross-sections showing punching failure: (a) N-IS-4; (b) N-IS-5 (after
Fig. 28. Numerical and experimental displacementetime histories for T1. [36]).
 K. Micallef et al. / International Journal of Impact Engineering 71 (2014) 17e33 31

Table 3
Comparison of strain-rate values.

Test 3_ (/s)

Proposed method UFC method

eq’n (42) eq’n (43)

T5 0.1 0.112
T1 0.08 0.130
N-IS-4 0.025 0.075
N-IS-5 0.035 0.072

compared with values obtained using (43), which has been his-
torically been used to estimate strain-rates since the 1990s (e.g. in
TM 5-1300 [63]).
The comparison is shown in Table 3.
It can be observed that, in all cases, the two equations predict
Fig. 31. Load-rotation response and failure criterion for N-IS-4. strain-rates within the same order of magnitude which, given that
both stiffness and strength are only significantly affected by strain-
rate when this changes by several orders of magnitude, then it can
The numerical prediction for this case is shown in Fig. 32. The be concluded that the proposed method produces relatively accu-
maximum strain-rate was predicted to be around 0.035/s, from rate estimates of strain-rate values which complement the
(42), and thus even in this case the static values were retained for remainder of the model presented in this paper.
both stiffness and strength. The model also predicts due to
punching shear and this was also confirmed by the experimental
test, as shown in Fig. 30(b). 5. Concluding remarks
Comparing the crack formation in the two specimens as shown
in Fig. 30, it can be observed that in the case of the lower loading In this paper, a two-phase model which can predict the dynamic
rate (N-IS-4), the flexural cracks leading to the shear cracks are response of RC slabs subjected to impact loading has been pre-
wider than in the case of the higher loading rate (N-IS-5). This is sented. This model provides the shear demand on a structure,
also predicted by the model, which suggests that the normalised taking into account the effects of strain-rate on the material
slab rotation (hence crack width) at failure is around 0.018 in the properties and also the initial increased stiffness of the structure
case of N-IS-5 and approximately 0.025 for N-IS-4. due to inertial effects.
This highlights the ability of the model to provide not only The slab stiffness and response is evaluated during the contact
verification on whether the impacted slab has failed or not but also and post-contact phases and it is found that the increase in stiffness
an indicative value of the slab deformation at failure. This is reported by other researchers is dominated primarily by the
important data for the designer since in most codes of practice (e.g. effective reduction in the slab’s span rather than by enhancements
UFC 3-340-02 [62]) use slab rotations to specify levels of acceptable to material properties due to strain-rate effects, except for very high
damage in the design of RC structures subjected to extreme loading. strain-rates (_3 a100/s). The model is used to assess the behaviour of
a slab subjected to a known impact scenario.
In addition, in this paper the failure criterion proposed in the
4.3. Strain-rate estimation
latest version of the Model Code to predict punching shear failure
in the static case has been extended to take into account the effects
The strain-rates in the preceding sections were computed using
of strain-rate and thus provide a variation of the dynamic shear
the method proposed in this paper, i.e., using (42). In order to
strength supply with slab deformation (rotation). It is found that
validate the suitability of the assumptions made in the derivation of
punching shear strength is only markedly increased for strain-rates
(42) as discussed in Section 3.2.3, the estimated strain-rates are
in excess of 100/s.
The model’s governing equations are solved numerically and
used for various slabs with and without transverse reinforcement
which are subjected to localised impact loading. The model is used
to predict the slab’s behaviour and assess the occurrence of failure
due to punching shear. The model also provides information about
the level of deformation (i.e. slab rotation) at failure. The good
correlation between experimental (from Refs. [28,36,46]) and
predicted results suggests the plausibility of the proposed model.
The advantage of the proposed method is that, unlike previous
methodologies, the current work follows an integral approach in
which both the loading on the structure (shear demand) as well as
its capacity (shear strength supply) are predicted.

Acknowledgements

This work is part of a research project financially supported by

the Engineering and Physical Sciences Research Council (E.P.S.R.C.)
Fig. 32. Load-rotation response and failure criterion for N-IS-5. of the U.K. (grant reference: EP/K008153/1).
32 K. Micallef et al. / International Journal of Impact Engineering 71 (2014) 17e33

Appendix A

References [8] Schuler H, Mayrhofer C, Thoma K. Spall experiments for the measurement of
the tensile strength and fracture energy of concrete at high strain-rates. Int J
[1] Bulletin 65: model code 2010 (volume 1) final draft. CEB/fib; 2010. Impact Eng 2006;32(10):1635e50.
[2] Cormie D, Mays G, Smith P. Blast effects on buildings. 2nd ed. London: ICE [9] Weerheijm J, Van Doormaal JCAM. Tensile failure of concrete at high loading
Publishing; 2012. rates: new test data on strength and fracture energy from instrumented
[3] Yon J-H, Hawkins NM, Kobayashi AS. Strain-rate sensitivity of concrete me- spalling tests. Int J Impact Eng 2007;34(3):609e26.
chanical properties. ACI Mater J 1992;89(2):146e53. [10] Yan D, Lin G, Chen G. Dynamic properties of plain concrete in triaxial stress
[4] Malvar LJ, Ross CA. Review of strain-rate effects for concrete in tension. ACI state. ACI Mater J 2009;106(1):89e94.
Mater J 1998;95(6):735e9. [11] Lu YB, Li QM. About the dynamic uniaxial tensile strength of concrete-like
[5] Lambert DE, Allen Ross C. Strain-rate effects on dynamic fracture and strength. materials. Int J Impact Eng 2011;38(4):171e80.
Int J Impact Eng 2000;24(10):985e98. [12] Cotsovos DM, Pavlovi&cacute MN. Numerical investigation of concrete sub-
[6] Grote DL, Park SW, Zhou M. Dynamic behaviour of concrete at high strain- jected to high rates of uniaxial tensile loading. Int J Impact Eng 2008;35(5):
rates and pressures: I. experimental characterisation. Int J Impact Eng 319e35.
2001;25(9):869e86. [13] Bull 66: model code 2010 (volume 2) final draft. CEB/fib; 2010.
[7] Li QM, Meng H. About the dynamic strength enhancement of concrete-like [14] Model code 1990. CEB/fib; 1993.
materials in a split Hopkinson pressure bar test. Int J Solids Struct [15] Mindess S, Banthia N, Yan C. The fracture toughness of concrete under impact
2003;40(2):343e60. loading. Cem Concr Res 1987;17:231e41.
 K. Micallef et al. / International Journal of Impact Engineering 71 (2014) 17e33 33

[16] Brara A, Klepaczko JR. Fracture energy of concrete at high loading rates in [40] Magnusson J, Hallgren M, Ansell A. Air-blast-loaded, high-strength concrete
tension. Int J Impact Eng 2007;34(3):424e35. beams. Part I: experimental investigation. Mag Concr Res 2010;62(2):127e36.
[17] Zhang XX, et al. Fracture behaviour of high-strength concrete at a wide range [41] Magnusson J, Ansell A, Hansson H. Air-blast-loaded, high-strength concrete
of loading rates. Int J Impact Eng 2009;36(10e11):1204e9. beams. Part II: numerical non-linear analysis. Mag Concr Res 2010;62(4):235e42.
[18] Oh BH. Fracture behaviour of concrete under high-rates of loading. Eng Fract [42] Abdel-Rohman M, Sawan J. Impact effect on RC slabs e analytical approach.
Mech 1990;35:327e32. ASCE J Struct Eng 1985;111(7):1590e601.
[19] Campbell JD. The dynamic yielding of mild steel. Acta Metall 1953;1(6):706e [43] Sawan J, Abdel-Rohman M. Impact effect on RC slabs e experimental
10. approach. ASCE J Struct Eng 1986;112(9):2057e65.
[20] Mainstone RJ. Properties of materials at high rates of straining or loading. [44] CEB. Bulletin 187: concrete structures under impact and impulsive loading.
Mater Struct 1975;8(2):102e16. CEB; 1988.
[21] Johnson GR, Cook W. Fracture characteristics of three metals subjected to [45] Delhomme F. Étude du comportement sous impact d’une structure pareblocs
various strains, strain rates, temperatures and pressures. Eng Fract Mech en béton armé. Universite de Savoie; 2005.
1985;21(1):31e48. [46] Barbier F, Roby M. Simplified unified model for concrete structures under
[22] Cowper GR, Symonds PS. Strain-hardening and strain-rate effects in the impact. In: Fib symposium 2013; 2013. Israel.
impact loading of cantilever beams. Brown University (Dept. of Appl. Math.); [47] ACI. ACI 318-08-building code requirements for structural concrete; 2008.
1957. Michigan (USA).
[23] Hughes G, Beeby AW. Investigation of the effect of impact loading on concrete [48] BSI. BS 8110-1-structural use of concrete. Code of practice for design and
beams. Struct Eng 1982;60B(3):45e52. construction; 1997. London (UK).
[24] Ross TJ, Krawinkler H. Impulsive direct shear failure in RC slabs. ASCE J Struct [49] CEN. EN 1992-1-1-design of concrete structures e part 1-1: general rules and
Eng 1985;111(8):1661e77. rules for buildings; 2004. Brussels (Belgium).
[25] Miyamoto A, King MW, Fujii M. Analysis of failure modes for RC slabs under [50] Muttoni A. SIA draft code proposal (SIA 162/AG5). Zürich, Switzerland: Swiss
impulsive loads. ACI Struct J 1991;88(5):538e45. Federal Institute of Technology; 1988.
[26] Miyamoto A, King MW, Fujii M. Nonlinear dynamic analysis of RC slabs under [51] Muttoni A. Punching shear strength of reinforced concrete slabs without
impulsive loads. ACI Struct J 1991;88(4):411e9. transverse reinforcement. ACI Struct J 2008;105(4):440e50.
[27] Saito H, et al. Loading capacities and failure modes of various reinforced- [52] Muttoni A, Ruiz MF. Shear strength of members without transverse rein-
concrete slabs subjected to high speed loading. Nucl Eng Des 1995;156(1e forcement as function of critical shear crack width. ACI Struct J 2008;105(2):
2):277e86. 163e72.
[28] Delhomme F, et al. Simulation of a block impacting a reinforced concrete slab [53] Ruiz MF, Muttoni A. Applications of critical shear crack theory to punching of
with a finite element model and a mass-spring system. Eng Struct reinforced concrete slabs with transverse reinforcement. ACI Struct J
2007;29(11):2844e52. 2009;106(4):485e94.
[29] Zineddin M, Krauthammer T. Dynamic response and behaviour of reinforced [54] Sagaseta J, et al. Non-axis-symmetrical punching shear around internal col-
concrete slabs under impact loading. Int J Impact Eng 2007;34(9):1517e34. umns of RC slabs without transverse reinforcement. Mag Concr Res
[30] Cotsovos D, Stathopoulos N, Zeris C. Behaviour of RC beams subjected to high 2011;63(6):441e57.
rates of concentrated loading. ASCE J Struct Eng 2008;134(12):1839e51. [55] Hillerborg A, Modéer M, Petersson PE. Analysis of crack formation and crack
[31] Chen Y, May IM. Reinforced concrete members under drop-weight impacts. growth in concrete by means of fracture mechanics and finite elements. Cem
Proc ICE e Struct Build 2009;162(1):45e56. Concr Res 1976;6(6):773e81.
[32] Izatt C, et al. Perforation owing to impacts on RC slabs. Proc ICE e Struct Build [56] Hordijk DA. Local approach to fatigue of concrete. Delft: Delft University of
2009;162(1):37e44. Technology; 1991.
[33] Saatci S, Vecchio FJ. Effects of shear mechanisms on impact behaviour of [57] Guidotti R. Poinçonnement des planchers-dalles avec colonnes superposées
reinforced concrete beams. ACI Struct J 2009;106(1):78e86. fortement sollicitées. Lausanne: École Polytechnique Fédérale de Lausanne;
[34] Abbas AA, Pullen AD, Cotsovos D. Structural response of RC wide beams under 2010.
low-rate and impact loading. Mag Concr Res 2010;62(10):723e40. [58] Walraven JC. Aggregate interlock: a theoretical and experimental analysis.
[35] Cotsovos DM. A simplified approach for assessing the load-carrying capacity Delft: Delft University of Technology; 1980.
of reinforced concrete beams under concentrated load applied at high rates. [59] Lui LM. Shear strength of concrete joints under dynamic loads. Hong Kong:
Int J Impact Eng 2010;37(8):907e17. University of Hong Kong; 1977.
[36] Bhatti AQ, Kishi N, Tan KH. Impact resistant behaviour of RC slab strengthened [60] Muttoni A, Ruiz MF. The levels-of-approximation approach in MC 2010:
with FRP strips. Mater Struct; 2011. application to punching shear provisions. Struct Concr 2012;13(1):32e41.
[37] Ozbolt J, Sharma A. Numerical simulation of reinforced concrete beams with [61] Ruiz MF, Muttoni A, Gambarova PG. Analytical modeling of the pre- and
different shear reinforcements under dynamic impact loads. Int J Impact Eng postyield behaviour of bond in reinforced concrete. ASCE J Struct Eng
2011;38(12):940e50. 2007;133(10):1364e72.
[38] Jiang H, Wang X, He S. Numerical simulation of impact tests on reinforced [62] UFC 3-340-02-structures to resist the effects of accidental explosions. USA
concrete beams. Mater Des 2012;39(0):111e20. Department of Defense; 2008.
[39] Mokhatar SN, Abdullah R. Computational analysis of reinforced concrete slabs [63] TM 5-1300-structures to resist the effects of accidental explosions. Depart-
subjected to impact loads. Int J Integr Eng 2012;4(2):70e6. ment of the Army; 1990.