Materials and Structures (2015) 48:1503–1521

DOI 10.1617/s11527-014-0250-6

ORIGINAL ARTICLE

Evaluation of the shear capacity of precast-prestressed

hollow core slabs: numerical and experimental comparisons
E. Brunesi • D. Bolognini • R. Nascimbene

Received: 5 February 2013 / Accepted: 7 January 2014 / Published online: 28 January 2014
 RILEM 2014

Abstract Since eighties, 400 and 500 mm thick of transverse reinforcement. The adopted diffuse
precast-prestressed concrete hollow core slabs, char- smeared fixed cracking constitutive model allows a
acterized by increasingly optimized cross-sections reliable prediction of shear stress distributions and
with non-circular voids, became very common. How- crack patterns for these members in their inelastic
ever, deeper slabs with long spans, which have to resist branch. The presence of a variable inclined strut is
high line loads acting close to the supports, are clearly evident. Peak shear stress is localized at the
subjected to initial web shear cracking and may fail at bottom side of the cross-section, rather than at the
loads less than those predicted by traditional codes level of the centroid. The experienced brittle web-
prescriptions. The shear strength capacity of these shear failure mechanism is governed by hollow core
members without transverse reinforcement is evalu- shapes with circular or non-circular voids, as evi-
ated through a campaign of detailed nonlinear finite denced by the evolution of the principal tensile strain
element analyses, matching experimental test data distributions. Typically, less inclined, more rounded,
collected from past programs. Constitutive models, diagonal crack, controlled by the smooth web width
based on nonlinear fracture mechanisms, are consid- variation along depth, is opposed to a fairly constant
ered to numerically reproduce the experimental variation of the fracture angle inclination, governed by
response of single span, simply supported, isolated the abrupt and irregular web width drop.
hollow core units, highlighting web-shear failure
mechanism, due to short development length and lack Keywords Hollow core slab  Shear strength 
Prestressed concrete  Precast members  Non-
circular voids
E. Brunesi (&)
ROSE Programme, UME School, IUSS Pavia, Pavia, Italy
Abbreviations
e-mail: [email protected] a Shear span
URL: http://www.umeschool.it
a1 Distance between two line loads
D. Bolognini  R. Nascimbene a1 Ratio of distance from end of member
European Centre for Training and Research in Earthquake to transfer length
Engineering (EUCENTRE), Pavia, Italy Alow Area of lower strands
e-mail: [email protected]
URL: http://www.eucentre.it Aup Area of upper strands (if present)
bw Sum of web widths
R. Nascimbene
e-mail: [email protected] D Cross-section depth, assumed as HS
URL: http://www.eucentre.it Di/Di,max Normalized depth
1504 Materials and Structures (2015) 48:1503–1521

Dlow Diameter of lower strands tlow Average concrete thickness below the
Dup Diameter of upper strands (if present) hollow core
Dz Mid-span vertical displacement syz Shear stress
dp,low Average distance from lower strand to Vobs Experimentally observed shear strength
the soffit capacity
dC50 Standard deviation of fC50,mean Vpre Predicted shear strength capacity
E1 Principal tensile strain x Coordinate along the longitudinal axis
E3 Principal compressive strain of the member
Fz Shear force at the support x/D Ratio between longitudinal distance
fc Concrete compressive strength from the support and cross-section depth
fC50,mean Mean concrete compressive strength
by cylinder tests on 50 mm cores
drilled from the tested specimens 1 Hollow core features and criticalities
fctd Design tensile strength of concrete per
EC2 The use of precast concrete in flooring system is a
ft Concrete tensile strength widespread practice in many countries. Precast-pre-
/ Diameter of strand stressed hollow core (PPHC) units are commonly
H Nominal slab depth spread products of precast industry, finding wide
HH Average depth of hollow core applications in building construction, particularly as
HS Average depth of slab cross-section floor and roof components. Short construction times,
I Moment of inertia of the section about high level of quality and consistent reduction of self-
the centroidal axis weight are achievable, allowing manufacturers to
LF Normalized load level cover spans up to 20 m. The upper surface of the
Ls Distance between slab end and centre hollow core slabs is usually levelled with a cast-in situ
of support concrete topping. Normally, for long spans, advantage
Lt Transfer length of prestressing force in is not taken of this topping layer as a composite
strands, assumed as 55/ (slow release) construction because of the uncertainty concerning the
Nlow Number of lower strands bond strength between the precast and cast-in situ
Nup Number of upper strands (if present) concrete. Proper shear strength at the interface is
Ri Ratio between experimentally observed required for the composite action; the horizontal and
and predicted shear strength capacity vertical shear strength of the joint between adjacent
S First area moment of the section about hollow core units has to be high enough to transmit the
the centroidal axis forces due to horizontal diaphragm action and load-
Slow Largest slippage of lower strands sharing of concentrated loads.
Sup Largest slippage of upper strands (if Through the years, 200 and 265 mm thick units,
present) characterized by cross-sections with circular voids, were
rcp Fully effective concrete compressive replaced by deeper units (320, 370, 400 and 500 mm),
stress at the centroid due to prestressing presenting cross sections with non-circular voids. Stan-
rp Prestress in the strands within the dardized cross-section shapes, usually characterized by
transfer region relevant void ratios, mostly depending on the thickness
rp0 Fully effective prestress in the strands of the slab itself, were increasingly optimized. However,
outside the transfer region experimental research works available in literature [9,
rp0,low Initial fully effective prestress in lower 13] evidenced that PPHC slabs, with deeper cross
strands outside the transfer region sections, could be subjected to initial web shear cracking
rp0,up Initial fully effective prestress in upper at the end regions, resulting into critical brittle failure
strands (if present) outside the transfer mechanisms. Comparisons with traditional codes pre-
region scriptions, American Concrete Institute (ACI) [1] and
tb Length of bearing Eurocode (EC2) [5], highlighted that these members
Materials and Structures (2015) 48:1503–1521 1505

a 2.5 b 2.5

200 200
265 265
2 320 2 320
370 370
400 400

Ri = Vobs / Vpre [−]

Ri = Vobs / Vpre [−]

500 500
1.5 Vobs = Vpre 1.5 Vobs = Vpre

1 1

0.5 0.5

0 0
0 100 200 300 400 500 600 700 800 0 100 200 300 400 500 600 700 800
Vpre [kN] Vpre [kN]

c 2.5 d 2.5
200 200
265 265
2 320 2 320
370 370
400 400
Ri = Vobs / Vpre [−]

Ri = Vobs / Vpre [−]

500 500
1.5 Vobs = Vpre 1.5 Vobs = Vpre

1 1

0.5 0.5

0 0
0 100 200 300 400 500 600 700 800 0 100 200 300 400 500 600 700 800
Vpre [kN] Vpre [kN]

Fig. 1 a–d Ri versus Vpre for each considered depth, according to EC2, Yang’s approach, ACI and CSA, respectively

without transverse reinforcement, generally not feasi- predicted one when the voids were non-circular.
ble, can fail, due to web-shear cracking in the ‘‘dis- Furthermore, Yang [23] observed that the maximum
turbed’’ end regions, at loads less than those predicted by tensile stress is not located at the level of the centroid,
conventional design methods. as expected by EC2, which, additionally, ignores the
Although PPHC slabs have been widely adopted, shear stresses due to the prestress force. Recently,
some aspects of the commonly-used design approaches Hawkins and Ghosh [9] have also shown that PPHC
against shear are still a cause of concern; short bearings slabs deeper than 320 mm can fail in shear at loads less
and not fully developed prestressing force influence than those predicted by ACI recommendations. Addi-
their response. Furthermore, their common production tional explanations for the observed mismatch have to
system (i.e. extrusion) may result in features not be researched in the different contributions to shear
properly in agreement with those of the standard transfer mechanism with respect to traditional normal-
concrete strength monitoring procedures. PPHC have strength concretes [7]. Shear failure surfaces in high-
undergone extensive experimental tests [2, 8, 12, 21] to strength concrete members are smoother than in
verify their compliance with the provisions of tradi- normal-strength concrete ones, showing cracks that
tional Codes and Standards and classify the possible can be propagated through coarse aggregate
failure modes. Walraven and Mercx [21] also noticed particles rather than around them, as shown by Cladera
that the observed shear resistance was lower than the and Marı́ [6].
1506 Materials and Structures (2015) 48:1503–1521

Fig. 2 Conceptual experimental test set up: details of wedges, rollers and examples of tested hollow core units (adapted from [13])

Table 1 Cases of study: model and slab type, nominal depth (H), number, diameter, area, initial prestress and largest slippage of
upper (Nup, Dup, Aup, rp0,up, Sup) and lower strands
Model Slab H Nup Dup Aup rp0,up Sup Nlow Dlow Alow rp0,low Slow
type type (mm) (–) (mm) (mm2) (MPa) (mm) (–) (mm) (mm2) (MPa) (mm)

200 Circular 200 – – – – – 7 12.5 93 1100 0.4

265 Circular 265 – – – – – 10 12.5 93 1100 1.1
320a Non-circular 320 – – – – – 13 12.5 93 1100 2.5
320b Non-circular 320 – – – – – 11 12.5 93 1000 1
320c Circular 320 2 9.3 52 900 0.2 11 12.5 93 1000 0.6
370 Non-circular 370 – – – – – 12 12.5 93 1000 0.6
400a Non-circular 400 – – – – – 13 12.5 93 1100 1.3
400b Non-circular 400 – – – – – 13 12.5 93 1000 1.9
500 Non-circular 500 – – – – – 16 12.5 93 1000 0.8

1.1 Experimental test database and accuracy levels of initial prestress. The imposed vertical loads
of current design methods acted at x/D [ 2.4. No cast-in situ filling concrete was
provided in the voids at the end regions. Idealized
Past experimental data, collected by Pajari [13], have pinned ends were mechanically materialized by rollers
been re-examined to obtain an experimental database and wedges, properly tightened to prevent torque.
of single span, simply supported, isolated PPHC units, To quantify their effectiveness, three commonly-
loaded with transverse and uniformly distributed line known Codes prescriptions, EC2, ACI and Canadian
loads, highlighting web-shear failure mechanism, due Standards Association (CSA) [3], as well as a research
to short development length and lack of shear rein- approach [23] against web-shear failure mechanism,
forcement. The 49 specimens were characterized by were adopted to predict Pajari experimental database.
six nominal slab depths, five hollow core shapes with In particular, Eq. 6.4 of EC2, reported below,
circular and non-circular voids, different void ratios, Eqs. 11–12 of ACI and Eqs. 11-6, 11-10, 11-11 and
several prestressing steel strands arrangements and 11-13 of CSA were used.
Materials and Structures (2015) 48:1503–1521 1507

Table 2 Cases of study. length, span, mass, bearing (tb), shear below hollow core (tlow), average distance from lower strand to
span (a) and distance between two line loads (a1), average the soffit (dp,low), mean core strength (fC50,mean) and relative
depth of section (HS), sum of individual web widths (bw), standard deviation (dC50)
average depth of hollow core (HH), average concrete thickness
Model Length Span Mass tb a(/a1) HS bw HH tlow dp,low fC50,mean dC50
type (mm) (mm) (kg) (mm) (mm) (mm) (mm) (mm) (mm) (mm) (MPa) (MPa)

200 6,643 6,603 2000 40 917 200 239 153 19 39 54.8 3.8
265 5,253 5,213 2240 40 724 260 228 175 42 36 71.3 4.2
320a 7,198 7,118 3810 80 1,160 315 309 222 38 43 61.8 2.4
320b 7,995 5,800 3880 80 760 321 289 216 36 38 64.3 2.6
320c 5,985 5,945 3100 40 826 310 263 213 40 51 46.8 2.0
370 7,002 6,942 4060 60 960/300 359 272 237 54 40 72.2 2.1
400a 7,626 7,546 4510 80 1,048 402 286 299 49 38 71.5 4.0
400b 8,390 7,150 4580 100 1,150 390 293 310 38 41 67.3 1.7
500 8,492 8,412 6000 80 1,475/300 503 324 398 40 47 76.4 2.6

Ibw
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 10
Vpre;EC2 ¼ ðfctd Þ2 þ a1 rcp fctd ð1Þ
S 0

As evidenced in Fig. 1, analytical approaches −10

roughly lead to unconservative (EC2 and Yang), or
Stress σ [MPa]

−20
over-conservative predictions (ACI and CSA), accord-
−30
ing to different nominal depths and geometry types,
not directly accounting for the influence of local −40 200
265
hollow core shape. Therefore, the inaccuracy of −50 320a
current codes is reaffirmed, evidencing a particularly 320b
320c
−60
critical scenario, if they are adopted outside of their 370
400a
‘‘implicit field of applicability’’. −70 400b
500
−80
−6 −5 −4 −3 −2 −1 0 1
−3
Strain ε [−] x 10
2 Nonlinear monotonic solid FE analyses
Fig. 3 Calibration of the solid FE models: uniaxial tensile and
The critical scenario depicted by analytical predictions compressive constitutive laws
suggests the need of a detailed campaign of numerical
analyses, as in past research works to assess the influence on the shear strength capacity for members
performance of RC [10] and steel structures [11, 15, with similar depths. Similarly, two 400 mm deep units
22], addressed to investigate the effects of the hollow have been analysed, since they evidence non-circular
core shape on the shear response of such members. To hollow core shapes with various levels of irregularity.
properly reproduce the experimental test set up,
conceptually schematized in Fig. 2, refined nonlinear 2.1 Proposed numerical approach
3D FE models, based on nonlinear fracture mecha-
nisms, were developed and monotonically loaded till To predict brittle web-shear failure mechanism, the
the brittle web-shear failure mechanism is reached. total strain crack model, developed along the lines of
Nine specimens, whose main features are presented the modified compression field theory (MCFT), orig-
in Tables 1 and 2, have been considered. In particular, inally proposed by Vecchio and Collins [19] and then
for 320 mm deep slabs, three specimens have been extended to the three-dimensional case by Selby and
selected: two characterized by non-circular and one by Vecchio [16], was adopted. The classical Von Mises
circular voids, to quantify the hollow core shape yielding criterion with strain hardening was used for
1508 Materials and Structures (2015) 48:1503–1521

Fig. 4 Specimen 320b:

example of detailed solid
FE models developed

Table 3 Total number of bricks and DoF for each specimen

Model type #Bricks #DoF
m1 m2 Strip m1 m2 Strip

200 124,960 88,770 13,640 257,187 181,566 28,927

265 73,920 52,800 9,600 153,800 110,120 20,871
320a 190,060 129,688 30,160 365,533 250,027 59,259
320b 155,100 109,604 25,016 303,727 215,155 50,337
320c 187,872 131,016 29,468 350,891 245,321 56,436
370 253,640 173,072 40,832 461,977 316,015 76,107
400a 247,680 165,120 37,920 467,279 312,299 72,903
400b 224,112 151,984 34,220 427,631 290,711 66,699
500 350,976 230,328 52,416 636,664 418,864 96,714

Table 4 Combinations of prestress distributions and losses of

0 0 x þ Ls  5/
8
prestress >
<  2 
Combination Prestress Loss of rp ¼ 1  Lt þ5/xL
Lt
s
rp0 5/  x þ Ls 5/ þ Lt
>
ID distribution prestress (%)
:
rp0 x þ Ls  5/ þ Lt
m2p1 Parabolic 5 ð2Þ
m2p2 Parabolic 15
Therefore, no interface elements were introduced to
m2p3 Linear 5 represent strands-concrete interaction, since implicitly
m2p4 Linear 15 captured by an equivalent prestress distribution.
According to Pajari [13], two loss of prestress ratios,
prestressed steel strands, represented as embedded 5 and 15 %, were adopted, due to the little level of
elements. Two types of prestress distribution, linear, knowledge about concrete mix and storing conditions,
according to codes provisions (ACI and CSA) and also providing an upper and lower threshold and
parabolic, according to Yang [23], were considered to showing that the shear capacity of PPHC units is not
evaluate their influence on the shear strength. In sensitive to small variations in the losses. Similar
particular, Eq. 2, proven by full-scale tests [17], range is observed by direct computations, as con-
expresses the parabolic curve selected. firmed by Palmer and Schultz [14] for analogous
Materials and Structures (2015) 48:1503–1521 1509

PPHC slabs, even though more conservative levels laws, for each selected specimen, according to CEB-
(20–25 %) may be assumed in the design process of FIP Model Code [4].
such members to safely represent the long-term Confinement effect was neglected, while full shear
effects. The diffuse smeared fixed cracking constitu- retention and lateral crack effect [20] were potentially
tive law assumed for concrete exploits an exponential accounted, as well as inward slippage of the strands.
softening decay curve and Thorenfeldt et al. [18] Newton–Raphson iteration scheme was adopted with
model for uniaxial tensile and compressive behavior, an energy-normalized convergence criterion. Six-
respectively. Figure 3 presents the calibration of the node brick elements were used and mesh sensitivity
adopted uniaxial tensile and compressive constitutive was investigated by doubling the extruding offset. For

Fig. 5 Comparison between experimentally observed and numerically predicted web-shear failure mechanism: a experimental failure
of specimen 320b (adapted from [13] )and b numerical crack pattern at ultimate conditions
1510 Materials and Structures (2015) 48:1503–1521

each specimen, an equivalent strip of the PPHC unit 250

320bm1p1
was modelled to evaluate the effectiveness of this 320bm2p1
320bm2p2
assumption. An example of the FE mesh is depicted in 200 320bm2p3
320bm2p4
Fig. 4, while Table 3 summarizes the total number of HypS320bm2p1

Shear Force Fz [kN]

brick elements and degrees of freedom employed for 150
each specimen. Finally, considering the four combi-
nations of prestress distributions and losses, detailed in
100
Table 4, a total of 54 models were created.

2.2 Results and discussion 50

In this section, FE results are synthesized to outline the 0

−6 −4 −2 0 2 4 6
influence of the hollow core shape on the web-shear
Vertical Displacement Dz [mm]
failure mechanism and, hence, on the shear strength of
such members. First, the proposed numerical approach Fig. 6 Specimen 320b: capacity curves, according to mesh
is validated by focusing on a single PPHC unit, taken sensitivity, strip modeling assumption and the four combina-
as reference. Therefore, the accuracy of the FE tions of prestress distributions and losses
predictions obtained for the nine specimens analysed
is quantified in comparison with experimental data. vertical displacement, Dz) are graphed in Fig. 6.
Finally, the observed trends are justified by consider- Independency with respect to adopted mesh refine-
ing the evolution of local quantities, such as principal ments is achieved through a proper calibration of the
tensile strain and shear stress, in order to depict crack bandwidth in each discretization zone, since the
behavioral changes as consequence of geometrical observed response is essentially governed by tensile
variations in the cross-section shape of such slabs. behavior. The choice of parabolic or linear prestress
distribution in strands leads to negligibly small global
2.2.1 Model validation: specimen 320b differences in terms of shear resistance (\0.1 %),
while larger influence is expected to occur if local
The adopted modeling approach evidences a consis- parameters, as stresses and strains distributions, are
tent match with experimental test results; Fig. 5 considered. More appreciable difference is due to the
presents a comparison between the experimentally loss of prestress ratio. The adoption of 15 % loss
observed and numerically predicted web-shear failure implies a reduced shear strength capacity, of about
mechanism, for a 320 mm deep PPHC unit, charac- 5 %, if compared to 5 % loss; furthermore, the
terized by non-circular voids (specimen 320b). Prin- capacity curves obtained for 15 % evidences a
cipal tensile and compressive strains, crack pattern and smoother nonlinear response for lower levels of shear
shear stress distribution at failure are in close agree- force at the support, being also characterized by a
ment with the collapse mechanism evidenced by the reduced ultimate mid-span vertical displacement
experimental damage pattern at the end of the test. A capacity of about 5 %. In this case, the strip modeling
smooth, inclined crack, whose width reduces as the assumption, referred as ‘‘HypS’’, cannot be accounted
imposed displacement monotonically increases, either for design purposes, as the convergence of the
emerges from both principal tensile strains and analysis failed when the capacity curve came into its
numerical crack pattern. Simultaneously, an inclined nonlinear branch, resulting in an over-conservative
compressive diagonal strut arises according to the estimation of the shear strength capacity. The intrinsic
principal tensile strain distribution, resulting in the lack of ductility reserves is confirmed by an evidently
undergone failure mode, also confirmed by a cut-off in low displacement ductility capacity, up to 1.3,
the shear stress flow. achieved for a maximum mid-span vertical drift of
According to mesh sensitivity analysis, strip *0.2 %.
modeling assumptions and the considered combina- The evolution of the shear stress distributions in the
tions of prestress distributions and losses, six capacity end region of the specimen is presented in Fig. 7.
curves (i.e. shear force at the support, Fz, vs. mid-span Specimen 320b, whose non-circular hollow core shape
Materials and Structures (2015) 48:1503–1521 1511

a 350 b 350
LF=19% LF=19%
LF=37% LF=37%
300 LF=48% 300 LF=48%
LF=59% LF=59%
LF=70% LF=70%
250 LF=78% 250 LF=78%
LF=85% LF=85%
Depth [mm]

Depth [mm]
LF=93% LF=93%
200 LF=96% 200 LF=96%
LF=100% LF=100%

150 150

100 100

50 50

0 0
−7 −6 −5 −4 −3 −2 −1 0 1 2 3 −7 −6 −5 −4 −3 −2 −1 0 1 2 3
Shear Stress τYZ [MPa] Shear Stress τYZ [MPa]

c 350
320bm2p1
320bm2p2
300 320bm2p3
320bm2p4

250
Depth [mm]

200

150

100

50

0
−7 −6 −5 −4 −3 −2 −1 0 1 2 3
Shear Stress τYZ [MPa]

Fig. 7 Evolution of the shear stress distributions at a x/D = 0.4 and b x/D = 0.7; c comparison between the four combinations of
prestress distributions and losses (x/D = 0.4)

presents an evidently abrupt and irregular web width evident increment is detected in the upper discontinu-
drop along the depth (Fig. 4), highlights the presence ity. At LF roughly equal to 50 %, the peak shear stress
of a strongly-disturbed zone. High shear stress overall the depth rises up from the strand to the web
concentrations are observed in the first LFs, since width drop level. This concentration remains pro-
the section shown in Fig. 7a is taken in correspon- nounced up to LF = 85 %, when the first cracking
dence of a strand, longitudinally, and, transversally at appears. For further LF increments, the ratio between
x/D = 0.4, where the prestressing force develops at a the current and previous peaks increases at a lower
high force variation per length. This concentration is rate, in accordance with the reduced slope of the
mostly due to the prestressing force and is mitigated as capacity curve into its nonlinear branch and the related
x/D increases (see Fig. 7b). As LF increases, the shear shear stress redistribution. Finally, at LF = 100 %,
stress develops in the linear branch of the capacity the peak is further uplifted, in accordance with the
curve, evidencing a second discontinuity below the shear stress cut-off experienced. This behavior reveals
mid-depth, where the web width drop is localized. For the complex interaction between cross-section geo-
further LF increments, the shear stress peaks at the metric features, in terms of hollow core shape, and the
strand level slightly increase, while a much more transfer mechanism of the prestressing force. The two
1512 Materials and Structures (2015) 48:1503–1521

discontinuities of the indented, fairly parabolic shear a 500

200m2p1
stress distributions are particularly visible for sections 450 265m2p1
320am2p1
close to the support (x/D \ 0.56). Their evolution, 400 320bm2p1
according to the experienced web-shear failure mech- 320cm2p1

Shear Force Fz [kN]

350 370m2p1
anism, results into an inclined crack. Their peak is 400am2p1
300 400bm2p1
forced to rise up, increasing both longitudinal distance 500m2p1

from the support and imposed vertical displacement, 250

according to hollow core shape and inherent web 200

width drop. In fact, sections particularly close to the 150

support (x/D \ 0.56), are characterized by a localiza- 100

tion of the shear stress peak evidently below with 50
respect to the centroidal axis, about 25 %, while for
0
sections, placed at 0.56 \ x/D \ 1.68, this mismatch −0.5 0 0.5 1 1.5
Vertical Drift [%]
tends to reduce, reaching a difference of *12 %.
Furthermore, at x/D = 2, the maximum shear stress is b 2

localized at the level of the centroid of the PPHC unit. 1.8 EC2
Yang
If shear stresses at ultimate conditions are consid- 1.6 ACI
ered (see Fig. 7c), the comparison between the two 1.4
CSA

Ri = Vobs / Vpre [−]

FEM
prestress distributions, for the same prestress loss, 1.2 Vobs = Vpre
leads to negligibly small differences in terms of peak
1
shear stress experienced overall the whole unit (2 %),
0.8
while the comparison between the two loss of prestress
0.6
ratios, for the same prestress distribution, evidences
slightly more appreciable mismatch (6 %). 0.4

0.2

2.2.2 Accuracy of FE predictions 0

0 100 200 300 400 500 600 700 800
Vpre [kN]
Figure 8a presents the capacity curves determined for
the nine specimens, assuming 5 % loss of prestress; Fig. 8 Shear capacities of the nine specimens selected: a shear
shear force at support versus mid-span vertical drift force at the support versus mid-span vertical drift curves and
b experimentally observed versus numerically (FEM) and
rather than displacement curves are shown to compare
analytically predicted shear strength
units with different lengths. The effectiveness of the FE
models developed, accounting both for geometrical and Similar trends are obtained for 15 % loss of
material nonlinearities, is reaffirmed. An accurate prestress; in fact, the adoption of 15 % loss confirms
agreement with experimentally observed shear strength the previously mentioned reduction of shear resis-
capacities is achieved, since 0.96 \ Ri \ 1.05, with a tance, approximately ranging from 4 to 10 %, accord-
mean value of 0.998, characterized by a reduced ing to specimen type. The capacity curves, obtained
standard deviation of 2.8 %, as shown in Tables 5 and for 15 % loss, are again characterized by a smoother
6 and graphed in Fig. 8b. Herein, a comparison with nonlinear branch for lower levels of shear force and, in
analytical predictions (EC2, Yang, ACI and CSA) is some cases, lower ultimate displacement capacity, of
presented as well. As a general tendency, slightly about 10–13 %, is even noticed. However, a weak
unconservative predictions are observed for PPHC units sensitivity of the failure mode with respect to the
with circular voids and small nominal depths (200 and prestress losses emerges, as, for all the specimens, the
265 mm), while, as the slab depth increases ([300 mm), web-shear failure mechanism is experienced at a
the accuracy of the FE results tends to increase. The level distance from support, which varies between one and
of non-conservatism (EC2 and Yang) or over-conser- two cross-section depths, where the prestressing force
vatism (ACI and CSA) disappears and an evidently is not yet fully developed.
lower standard deviation demonstrates the robustness of For mid-span vertical drifts, ranging from 0.1 to
the proposed modeling approach. 0.4 %, the considered capacity curves reach their
Materials and Structures (2015) 48:1503–1521 1513

nonlinear branches, with different slope and post- displacement ductility of about 1.3. Therefore, the
yielding stiffness. Evidently low displacement ductil- non-circularity of the voids, the inherent web width
ity levels, approximately ranging from 1.3 to 2.3, variation along depth and concrete thickness above
according to different hollow core shapes and cross- and below the hollow core are observed to govern the
section features, are observed. Each specimen col- displacement ductility capacity of PPHC units, and,
lapses by brittle web-shear mode, thus confirming the hence, the fragility of the collapse mechanism
intrinsic lack of ductility reserve for this type of experienced.
member. According to FE results, PPHC units with
circular voids typically exhibits a relatively more 2.2.3 Influence of hollow core shape: 400b
ductile response, reaching a displacement ductility of versus 400a and 320c versus 320b
about 2, while a less uniform trend is detected in
deeper cross-sections characterized by non-circular This trend is corroborated by the evolution of principal
voids. In fact, displacement ductilities up to *2 are tensile and compressive strain distributions, presented
determined for slabs presenting a relevant web width in Figs. 9 and 10, for specimen 400b. The principal
and concrete thickness above and below the hollow tensile strain directly governs the failure mode and the
core itself (specimens 500 and 370) or an almost presence of a variable inclined strut and tie, whose
regular web width variation along the hollow core width and inclination angle evolves during the mono-
depth (specimen 400a). By contrast, specimens char- tonic analysis, is clearly evident. As shown in Fig. 9,
acterized by an abrupt and irregular web width drop tie width reduces as the imposed vertical displacement
(specimen 400b) or reduced web width, if compared to increases, reproducing the actual shear stress flow,
concrete thickness above and below the hollow core characterized by peaks placed in correspondence of
(specimens 320a and 320b), ensure evidently lower the bottom side of the cross-section, rather than at the

Table 5 Comparison between experimental, analytical and FEM results

Model type Loss of prestress = 5 %
EC2 Yang ACI CSA FEM
Vobs Vpre Ri Vpre Ri Vpre Ri Vpre Ri Vpre Ri
(kN) (kN) (–) (kN) (–) (kN) (–) (kN) (–) (kN) (–)

200 90.2 145.7 0.619 128.0 0.704 98.5 0.916 114.8 0.786 94.3 0.956
265 219.4 223.8 0.980 223.7 0.981 155.8 1.408 158.1 1.388 223.2 0.983
320a 297.1 378.7 0.785 322.9 0.920 266.8 1.114 240.4 1.236 290.8 1.022
320b 232.5 359.7 0.646 317.2 0.733 253.1 0.919 236.5 0.983 221.4 1.050
320c 253.9 246.6 1.030 239.7 1.059 174.8 1.453 166.0 1.530 252.8 1.004
370 306.2 396.2 0.773 373.8 0.819 277.7 1.103 255.4 1.199 315.4 0.971
400a 291.3 475.4 0.613 444.0 0.656 363.3 0.802 309.0 0.943 293.9 0.991
400b 444.0 500.9 0.886 405.7 1.094 348.8 1.273 278.9 1.592 443.3 1.002
500 471.2 702.5 0.671 565.7 0.833 550.7 0.856 415.6 1.134 469.8 1.003
Vobs, Vpre and relative Ri for each specimen, according to EC2, ACI, CSA provisions, Yang’s approach and FEM

Table 6 Mean and standard deviation of Ri, according to EC2, Yang, ACI, CSA and FEM approaches
Model type Loss of prestress = 5 %
EC2 Yang ACI CSA FEM
Mean SD Mean SD Mean SD Mean SD Mean SD

All 0.778 15.7 % 0.867 15.7 % 1.094 24.1 % 1.199 27.1 % 0.998 2.8 %
1514 Materials and Structures (2015) 48:1503–1521

Fig. 9 Evolution of the principal tensile strains E1 for monotonically increased vertical displacement steps imposed (specimen 400b)
Materials and Structures (2015) 48:1503–1521 1515

level of the centroid. A brittle web-shear failure To further validate this trend, a comparison
mechanism results from an inclined diagonal crack, between the evolution of the principal tensile strain
which, for this specimen, with a sharp web width distributions, according to the monotonically increased
variation along the depth, almost linearly propagates vertical displacements imposed, is presented in
from the support to zones upper than the centroidal Fig. 12, for units with circular (specimen 320c) and
axis at a distance of about two cross-section depths. non-circular (specimen 320b) hollow core shapes.
Accordingly, a diagonal strut, whose width increases Specimen 320b evidences a failure mode in consistent
as the imposed vertical displacement increases, con- agreement with that previously discussed for specimen
nects the transverse line load and the support, as 400b, being the two PPHC units characterized by
depicted in Fig. 10. This failure mode results into an similar cross-section geometric features, particularly
incipient change in the shear stress distribution, in terms of non-circularity of the voids. High principal
localized at a distance roughly equal to the hollow tensile strains concentrations are again depicted in the
core depth, where the maximum tensile strain overall inner constant part of the web width, causing a fairly
the whole member is undergone. linear crack, which expands from the lower part of the
In addition, the evolution of the principal tensile hollow core, in the proximity of the support, to upper
strain distributions and inherent damage pattern ones, in sections at about two slab depths. The fracture
experienced by non-circular PPHC units with different angle observed visibly differs from those experienced
regularity of web width variation along depth is by PPHC units with circular voids, e.g. specimen
compared, by taking specimens 400a and 400b as 320c. According to its capacity curve, a more rapid
reference. Specimen 400a, more regular than 400b, and brittle failure mechanism takes place in specimen
presents very highly concentrated tensile strains in 320b, highlighting a more inclined, almost linear,
correspondence of the minimum of the web width, as fracture angle, whose spread follows the sharp and
shown in Fig. 11. For this type of cross-section prominent web width drop. On the contrary, in
configuration, an inclined crack appears at the support specimen 320c, principal tensile strain peaks are
and, almost linearly, extends to approximately twice forced to occur in accordance with the gradual and
the depth of the considered PPHC slab, since its regular web width variation along the depth. In fact,
propagation along the longitudinal axis finds no their evolution, according to the inherent crack pattern
significantly abrupt web width variation. Conversely, at ultimate conditions, reveals a more manifest
in specimen 400b, the principal tensile strain was variation of the fracture angle at further distances
shown to be more evidently lumped at the level of the from the support, resulting in a clearly more rounded
sharp and irregular web width drop. As the distance and less inclined diagonal crack, which links the
from the support increases, the experienced peak rises support and the centroidal axis at a distance of about
up from the bottom of the cross-section, evidently two hollow core depths.
below with respect to the centroid, for sections close to Therefore, in Fig. 13, the shear stress distribu-
the support, to zones upper than the centroidal axis, for tions at ultimate conditions, obtained by assuming
sections at two depths. The evolution of the fracture parabolic prestress distribution along strands, com-
angle significantly differs from that shown by speci- bined with 5 % loss of prestress, are compared, for
men 400a. Even though the inclination of both remains each specimen, at commonly selected distance from
fairly constant, at increased distances from the the support. The nine PPHC slabs analysed experi-
support, specimen 400b highlights a more prominent ence their peaks overall the whole unit at different
brittle web-shear failure mechanism, if compared to sections, anyway all included within 0 \ x/D \ 0.5.
400a, since the shear fracture more suddenly appears The fairly parabolic shear stress distributions col-
and more rapidly evolves, as confirmed by their lected reaffirm the behavior previously discussed, in
capacity curves (Fig. 8a). As a result, it occurs a less terms of fracture angle inclination. Cross-sectional
inclined and slightly more rounded fracture, whose geometric features, interacting with the transfer
propagation along the longitudinal axis of the member mechanism of the prestressing force, govern their
and evolution, according to the imposed monotonic evolution and shape, confirming that the web-shear
loading history, is controlled by a more abrupt and failure mode of such PPHC units depends on the
irregular change in the hollow core shape. irregularity of the voids. Slabs characterized by
1516 Materials and Structures (2015) 48:1503–1521

Fig. 10 Subsequent evolution of the principal compressive strains E3 for monotonically increased vertical displacement steps
(specimen 400b)
Materials and Structures (2015) 48:1503–1521 1517

Fig. 11 Different evolution of E1 in specimen 400a, still characterized by non-circular voids, but with increased regularity in terms of
web width variation along depth
1518 Materials and Structures (2015) 48:1503–1521

Fig. 12 Comparison
between the evolution of E1
for different hollow core
shapes, characterized by
a circular (specimen 320c)
and b non-circular voids
(specimen 320b)
Materials and Structures (2015) 48:1503–1521 1519

a 1 b 1
200m2p1 200m2p1
0.9 265m2p1 0.9 265m2p1
Normalized Depth Di / Di,max [−]

Normalized Depth Di / Di,max [−]

320am2p1 320am2p1
0.8 320bm2p1 0.8 320bm2p1
320cm2p1 320cm2p1
0.7 370m2p1 0.7 370m2p1
400am2p1 400am2p1
0.6 400bm2p1 0.6 400bm2p1
500m2p1 500m2p1
0.5 0.5

0.4 0.4

0.3 0.3

0.2 0.2

0.1 0.1

0 0
−8 −6 −4 −2 0 2 −8 −6 −4 −2 0 2

Shear Stress τYZ [MPa] Shear Stress τYZ [MPa]

c 1 d 1
200m2p1 200m2p1
0.9 265m2p1 0.9 265m2p1
Normalized Depth Di / Di,max [−]

Normalized Depth Di / Di,max [−]

320am2p1 320am2p1
0.8 320bm2p1 0.8 320bm2p1
320cm2p1 320cm2p1
0.7 370m2p1 0.7 370m2p1
400am2p1 400am2p1
0.6 400bm2p1 0.6 400bm2p1
500m2p1 500m2p1
0.5 0.5

0.4 0.4

0.3 0.3

0.2 0.2

0.1 0.1

0 0
−8 −6 −4 −2 0 2 −8 −6 −4 −2 0 2

Shear Stress τYZ [MPa] Shear Stress τYZ [MPa]

Fig. 13 Shear stress distributions at ultimate conditions versus normalized slab depth for increased distance from the support,
assuming parabolic prestress distribution and 5 % loss of prestress

circular voids show a nearly parabolic slope, with 3 Conclusions

peaks closer to the level of the centroid, while more
prominent and irregular distributions are observed A modeling approach, based on nonlinear fracture
for sections presenting non-circular voids, in accor- mechanisms, was proposed to predict the shear
dance with a different web width variation along response of PPHC slabs. Numerical observations were
depth. Furthermore, discontinuities at the pre- discussed to investigate the failure mode of the units
stressed strands level are particularly evident for analysed and to depict behavioral changes as conse-
deeper cross-sections (specimens 400a, 400b and quence of geometrical variations in their cross-section
500) which locally evidence strongly-disturbed and shape. The following observations and conclusions
indented slopes characterized by maxima localized were drawn from the numerical study:
clearly below the centroidal axis of the member. For
• In contrast of the mismatch shown by conventional
further longitudinal distances from the support, the
codes, a satisfying fit with experimental results
local disturbance is visibly mitigated, resulting in
was achieved in terms of shear capacity, faithfully
peaks close to the level of the centroid.
reflecting the behavior of such high-strength
1520 Materials and Structures (2015) 48:1503–1521

concrete members (0.96 \ Ri \ 1.05, with mean 2. Becker RJ, Buettner DR (1985) Shear tests of extruded
of 0.998 and standard deviation of 2.8 %). hollow core slabs. PCI J 30(2):40–54
3. Canadian Standards Association (2001). CSA standard
• The complex contribution due to the transfer A23.3-04 design of concrete structures
mechanism of prestressing force, which interacts 4. CEB-FIP (1990) CEB-FIP Model Code. Comité Euro-
with the hollow core features, was highlighted by International du Béton, 1993
the detailed FE analyses performed in a nonlinear 5. Commission for the European Communities (2005) Euro-
code 2: design of concrete structures—Part 1: general rules
fashion. Furthermore, the influence of prestress and rules for buildings, EN 1992-1-1
distributions and losses was quantified. 6. Cladera A, Marı́ AR (2005) Experimental study on high-
• Intrinsic lack of ductility reserves was confirmed for strength concrete beams failing in shear. Eng Struct 27(10):
these units without transverse reinforcement and a 1519–1527
7. Duthinh D (1999) Sensitivity of shear strength of reinforced
trend was established, according to the regularity of concrete and prestressed concrete beams to shear friction
the voids. Displacement ductility of about 2 was and concrete softening according to modified compression
observed for specimens with circular voids, while field theory. ACI Struct J 96(4):495–508
levels slightly larger than unity were determined in 8. Girhammar UA (1992) Design principles for simply sup-
ported prestressed hollow core slabs. Struct Eng Rev,
deep cross-sections with pronounced flat webs that Oxford, England, 4(4): 301–316
create discontinuities in the shear stress path. 9. Hawkins NM, Ghosh SK (2006) Shear strength of hollow
• The brittle web-shear failure mechanism experi- core slabs. PCI J 51(1):110–115
enced was proven to be controlled by hollow core 10. Mpampatsikos V, Nascimbene R, Petrini L (2008) A critical
review of the RC frame existing building assessment pro-
shapes and related non-circularity of the voids, cedure according to EUROCODE 8 and Italian seismic
inherent web width variation along depth and code. J Earthq Eng 12(S1):52–82
concrete chords above and below the void, as 11. Nascimbene R, Rassati GA, Wijesundara KK (2011)
confirmed by the evolution of the principal tensile Numerical simulation of gusset-plate connections with
rectangular hollow section shape brace under quasi-static
and compressive strains, developing a variable cyclic loading. J Constr Steel Res 70:177–189
inclined strut and tie mechanism. 12. Pajari M (1989) Design of prestressed hollow core slabs.
• The presence of a less inclined, more rounded, Research Report 657, VTT Building and Transport, Finland
diagonal crack, controlled by the smooth web width 13. Pajari M (2005) Resistance of prestressed hollow core slabs
against web shear failure. Research Notes 1292, VTT
variation along depth, is generally opposed to a fairly Building and Transport, Finland
constant variation of the fracture angle inclination, 14. Palmer KD, Schultz AE (2011) Experimental investigation
governed by the abrupt and irregular web width drop. of the web-shear strength of deep hollow-core units. PCI J
• Shear stress distributions confirmed the key role 56(4):83–104
15. Santagati S, Bolognini D, Nascimbene R (2012) Strain life
played by cross-section features. Deeper specimens, analysis at low cycle fatigue on concentrically braced steel
whose cross-sections are characterized by non- structures with RHS shape braces. J Earthq Eng 16(S1):
circular hollow core shapes, reveal more prominent, 107–137
irregular and indented shear stress distributions, 16. Selby RG, Vecchio FJ (1993) Three-dimensional constitu-
tive relations for reinforced concrete, Technical Report
whose peak is localized evidently below the centroid, 93–02, University of Toronto, Department Civil Engineer-
at the level of the sharp change in their web width. ing Toronto, Canada
• The numerical outcome of this research may be 17. Shahawy MA, Issa M, Batchelor BD (1992) Strand transfer
directly applicable to develop alternative design lengths in full scale AASHTO prestressed concrete girders.
PCI J 37(3):84–96
methods against web-shear failure of PPHC units, 18. Thorenfeldt E, Tomaszewicz A, Jensen JJ (1987) Mechan-
saving the inaccuracy of previous proposals, ical properties of high-strength concrete and applications in
especially the various codes, by explicitly account- design. In: Proceedings symposium utilization of high-
ing for the influence of the hollow core shape. strength concrete, Trondheim, Norway, Tapir
19. Vecchio FJ, Collins MP (1986) The modified compression
field theory for reinforced concrete elements subjected to
shear. ACI Struct J 83(22):219–231
References 20. Vecchio FJ, Collins MP (1993) Compression response of
cracked reinforced concrete. J Struct Eng ASCE 119(12):
1. ACI Committee (2005) Building code requirements for 3590–3610
structural concrete (ACI 318–05) and commentary (ACI 21. Walraven JC, Mercx WPM (1983) The bearing capacity of
318R–05). American Concrete Institute, Farmington Hills prestressed hollow core slabs. Heron, 28(3), p. 46
Materials and Structures (2015) 48:1503–1521 1521

22. Wijesundara KK, Bolognini D, Nascimbene R, Calvi GM 23. Yang L (1994) Design of prestressed hollow core slabs with
(2009) Review of design parameters of concentrically reference to web shear failure. J Struct Eng ASCE 120(9):
braced frames with RHS shape braces. J Earthq Eng 13(S1): 2675–2696
109–131