Journal of Constructional Steel Research 95 (2014) 56–70

Contents lists available at ScienceDirect

Journal of Constructional Steel Research

Stress transfer mechanism investigation in hybrid steel trussed–concrete

beams by push-out tests
Piero Colajanni a,⁎, Lidia La Mendola b, Alessia Monaco b
a
Dipartimento di Ingegneria Civile, Università di Messina, Contrada Di Dio 1, 98166 Messina, Italy
b
Dipartimento di Ingegneria Civile, Ambientale, Aerospaziale, dei Materiali, Università di Palermo, Viale delle Scienze, 90128 Palermo, Italy

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

Article history: Results of push-out tests carried out on Hybrid Steel Trussed–Concrete Beams (HSTCBs) before and after the con-
Received 17 July 2013 crete casting are presented and interpreted. Firstly, in order to check the ability of weldings before casting, tensile
Accepted 23 November 2013 tests were performed on specimens reproducing different types of welded joints. Simplified design formulae
Available online xxxx
were used to predict their ultimate strength. Secondly, results obtained by push-out tests on specimen represen-
tative of the beam before and after the concrete casting are presented and discussed. Finally, simplified analytical
Keywords:
Hybrid steel trussed–concrete beams
models proposed by the current European building code were adapted to the specific typology to roughly predict
Welded joints the ultimate strength obtained by push-out tests on specimens complete with concrete casting.
Push-out tests © 2013 Elsevier Ltd. All rights reserved.
Stress transferring mechanism
Analytical modelling

1. Introduction The mechanical behaviour of HSTCBs was usually modelled by

Hybrid Steel Trussed–Concrete Beams (HSTCBs) are a typical Italian
structural typology constituted by a precast steel truss embedded into a
concrete core generally cast in situ. The adoption of HSTCBs within rein-
forced concrete frames is a structural solution for light industrialization
in use in the Italian construction industry since 50 years due to high
construction speed with minimum site labour, the possibility of covering
wide spans with low depths as well as economical convenience.
In spite of the widespread use of this beam typology due to both the
easy manufacturing operations and the possibility of covering wide
spans, to date few studies are present in the literature about this topic
[1–9]. Some of these studies refer to a beam type similar [4,5,7] or identi-
cal [6,9] to that considered in this paper.
Among several existing HSTCB typologies, this study concerns beam
types made up of:—a steel plate placed at the bottom of the beam that
acts as the bottom chord of the truss (it also plays the role of a mould
during the concrete casting);—coupled steel bars constituting the
upper chord;—steel inclined web bars (V-reverse) welded to the two
chord elements. In the first phase of the realization process (phase I),
only the bearing capacity of the steel truss is often utilized in order to
reduce or to eliminate the supports before and during casting. After
cure and maturation of the concrete (phase II), bearing capacity is eval-
uated by considering the composite beam made of steel and concrete.
Reinforcement is provided by the steel truss, and reinforcing bars are
added at the end zone of the beam to guarantee the stress transfer to
the beam-to-column joint panel.
⁎ Corresponding author. Tel.: +39 0903977161; fax: +39 0903977480.
E-mail address: pcolajanni@unime.it (P. Colajanni).
employing the classical models and rules developed for reinforced con-
crete or for steel–concrete composite beams [10,11]. Notwithstanding,
due to the widespread use of HSTCBs in practice, current Italian building
code makes a clear reference to this structural typology, requiring the
adoption of specific design criteria and calculus methods derived by
means of design by testing procedure. Mainly in the last decade a signif-
icant research activity has begun. The main topics covered by the scien-
tific community involved in such research program have regarded
evaluation of flexural and shear strength of the beam [8], behaviour of
the steel to concrete connection [7,9], the study of the buckling of
steel trusses before, during or after the concrete cast, the cyclic behav-
iour of beam-to-column joints, and the issues related to the creep.
Concerning to HSTCB behaviour after the concrete casting, the beam
can be viewed as steel–concrete hybrid beam with deformable shear con-
nection represented by the web members of the steel truss welded to the
steel plate at the bottom [7]. Thus, in most of the problem related to be-
haviour in phase II a decisive role is played by the transfer mechanism
of the stresses from the steel truss to the concrete core. As a matter of
fact, experimental tests on shear critical HSTCB showed that the collapse
can arise from three different mechanisms: collapse for diagonal tension,
or shear compression [8], or failure of the shear connection [7,9].
The paper herein presented mainly deals with this last topic of
paramount importance for the behaviour of HSTCBs both under service-
ability and under ultimate loads.
Particularly, the paper is focused on the investigation of the stress
transfer mechanisms in HSTCBs occurring in experimental push-out
tests carried out on specimen representative of the aforementioned
beam typology before and after the concrete casting. As far as there is
not a standardised experimental procedure to evaluate the resistance

0143-974X/$ – see front matter © 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.jcsr.2013.11.025
 P. Colajanni et al. / Journal of Constructional Steel Research 95 (2014) 56–70
and the stiffness of the steel–concrete shear connection, the slip–load
curve is usually characterized by means of push-out tests inspired to the
classical tests developed for steel–concrete composite beams according
to Eurocode 4 [10]. Even if the load conditions of such a test do not repro-
duce the load condition of the beam under combined shear and flexure
action, as in steel–concrete composite beams, these results are needed
to evaluate strength and stiffness of the connection in order to recognize
if the full flexural strength [7,9] or the shear strength [8] can be reached in
the beam.
The tests were carried out at the University of Palermo, Italy, in the
framework of the abovementioned research program addressed to vali-
dating methods and criteria for structural design of this beam typology.
Firstly results of tensile tests performed on specimens reproducing
different types of welded joints are presented in order to investigate
their bearing capacity which is a requirement of paramount importance
in the phase before casting, when self-bearing capacities are exploited.
The joints are different from the ones which are normally used due to
the coupling of different steel types and complex geometry involving
rebar bent with large curvature. Simplified design formulae of structural
codes [12] are used to predict the ultimate strength of welded joints for
each different typology, according also to results and discussions of a
larger extensive experimental test survey on strength of this typology
welded joint presented in Colajanni et al. [13].
Secondly, results obtained by push-out experimental tests on speci-
men representative of the beam before and after the concrete casting
are presented and discussed. Push-out tests are carried out first in speci-
mens before concrete casting, and then on specimens completed with
concrete, allowing local problems regarding the stress transferring mech-
anism between different materials to be investigated. More precisely, the
stress transferring between steel elements (of different steel types also) in
phase I, and the stress transferring between steel truss elements and con-
crete in phase II are analysed. Furthermore, the behaviour of the steel
truss before concrete casting was investigated with the focus on buckling
phenomena occurring in both web bars and the upper chord [14]. Finally,
some simple analytical models proposed by the current European build-
ing code, Eurocode 2 [11], were adapted to the specific typology aiming
at interpreting the strength values obtained by experimental tests. All
these results are part of a more extensive experimental and theoretical re-
search carried out in order to investigate the seismic behaviour of HSTCBs
in framed structures [6].
2. Experimental program
In Fig. 1a) a photo showing the beam typology to which this study is
addressed is presented before concrete casting, and in Fig. 1b) the spe-
cific slab truss beam used in this experimental program is reported; for
a)
Fig. 1. Typology of hybrid steel trussed–concrete beam: a) photo before concrete casting in r.c. framed structure; b) steel truss type utilized in experimental investigation.
57
this beam typology the self-bearing capacity of the steel truss in phase I
is not exploited.
The experimental tests related to phase I investigate the bearing ca-
pacity of welded joints, verifying their overstrength among steel ele-
ments by means of tensile tests on welded joints (Fig. 2). Moreover,
the efficiency in phase I of inclined web bars is investigated by push-
out tests on bare steel trusses. Tests were carried out on specimens of
beam segments similar to those usually utilized for traditional steel–
concrete composite beams, according to Eurocode 4 [10] (Fig. 3).
Afterwards, push-out tests on specimens completed with concrete
representing the beam in phase II were carried out in order to investi-
gate the stress transfer mechanism between steel and concrete, follow-
ing again the procedure suggested in Eurocode 4 [10].
In all the examined cases the plate located at the bottom is consti-
tuted by S355 steel having nominal yield stress 355 N/mm2 and the
rebars by steel B450C with nominal yield stress 450 N/mm 2 .
Weldings are made according to recommendations in the ISO Code
[15].
Preliminarily, the material strength values of coupling elements
were evaluated by axial tensile tests on steel bars and steel plates ac-
cording to code provisions [16]. Three tests on specimens extracted
from the 5 mm thick steel plate of HSTCB and six tests on pieces of
bars, three for each diameter used in HSTCB (12 mm and 16 mm
for web bars and upper chord respectively) were performed.
Table 1 shows the mean values of test results in terms of stress and
corresponding strain at yielding (fy, εy) and the ultimate values at
peak (fu, εu).
To carry out the tensile tests a universal testing machine with a
600 kN load carrying capacity was utilized.
In the next sections, first tests performed on welded joints, and sec-
ondly push-out tests in phase I and phase II are presented. Specifically
twelve tensile tests on welded joints, two push-out tests on bare truss
specimens and three push-out tests on complete truss specimens
were carried out.
3. Welded joints
In this section the results of experimental tests on weldings together
with a simplified calculus of their bearing capacity are reported. Butt
welded joints of web bar to bottom steel plate and fillet welding joints
of inclined web bars to top chord bars were tested. Test on these joint
configurations is performed since they are different from the ones
which are normally involved in concrete or hybrid structures due to
the coupling of different types of steel and complex geometry involving
rebar bent with large curvature. Such investigation is of paramount im-
portance when the steel truss is employed exploiting its self-bearing
b)
Upper chord (3 bars with dup=16 mm)
Web steel bars (reverse V with dw=12
mm)
Bars for welding guidance
Bottom steel plate 5 mm thick
58 P. Colajanni et al. / Journal of Constructional Steel Research 95 (2014) 56–70

a)
Type B

Type A

b)
5
F F

100

12-16
300 300

c)
F F
auxiliary bar
18 auxiliary bar
steel plate
steel plate bar-to-plate
welding
31
6

2 bar-to-plate
22

90

welding

steel bar
(upper chord)
16

steel bar
(upper chord) reverse V

12
reverse V
F
F

Fig. 2. Welded joints (dimensions in millimetres): a) types; b) specimen representing joint type A; c) specimen representing joint type B.

capacity. In phase I, in fact, weldings are required to be able to transfer
loads to the steel elements that are mutually connected each other.
The results presented in the following sections show the adequate be-
haviour of joints under loads.
An auxiliary bar and two stiffening steel plates were added in
order to apply the tensile load F.
All tensile tests were carried out using the same testing machine as
used for the mechanical characterization of different steel elements
composing HSTCBs.

3.1. Specimens and test execution 3.2. Experimental results and resistance evaluation

Tensile tests in displacement control were performed for two
different types of welded joint, as shown in Fig. 2a). The first one,
named type A, is a butt welding between web bar and steel plate;
specimens are made up of pieces of bar butts welded to a steel
plate inclined at an angle α of 60°, chosen in such a way as to repro-
duce the joint in the steel truss. An auxiliary bar was added in order
to apply the tensile load F, as shown in Fig. 2b). Specimens with full
penetration welding were prepared between steel plate of 5 mm
thickness and web bar of 12 mm diameter.
The second one, named type B, is a fillet welding between the
web bar of 12 mm of diameter and the web bars of upper chord of
16 mm of diameter; in this case the specimen consists of a segment
of bar, bent as a reverse V, jointed by corner welding to two longi-
tudinal bars representing the upper chord, as shown in Fig. 2c).
Table 2 shows the experimental results in terms of ultimate load
value Fexp and failure mode.
Specimens marked by superscript ()* in Table 2 were obtained from
trusses manufactured in a factory and ready to be completed by con-
crete casting.
The same table contains the ratio between the experimental ulti-
mate load value Fexp and the one obtained by simplified expressions
for welding strength, Fu.
In the case of type A butt welding, the ultimate load value is cal-
culated by assuming that the section is monolithic and the tensile
stress value is that of the filler welding material. The obtained
value is then multiplied by the coefficient 0.85 (class II in [17]), giv-
ing 49.03 kN; in all cases, as can be seen in the table, the calculated
values are precautionary.
 P. Colajanni et al. / Journal of Constructional Steel Research 95 (2014) 56–70 59

a) b) P

75
140

300

675
1060
750
bottom bars

Steel plate

200
700
three coupled steel bars
dup = 16 mm

300

two web steel bars

210 dw = 12 mm P/2 P/2

Fig. 3. Push-out test in phase I (dimensions in millimetres): a) specimen S1; b) specimen S2.

For evaluation of the bearing capacity of the type B welded joint, the The predicted analytical value Fu = 47.54 kN, calculated according
simplified method proposed in Eurocode 3 [12] is utilized, in which the to Eq. (2), versus experimental values, Fexp, is reported in Table 2 for
shear strength fvw is: the cases in which the collapse occurs due to welding failure. In all
cases analytical assessment is precautionary.
pffiffiffi
f u= 3 From Table 2 in which the failure mode is reported for all tests, it has
f vw ¼ ð1Þ to be noted that in some cases the collapse occurs due to the failure of
βw
steel bars for higher load values. This event mostly concerns A speci-
where fu is the nominal ultimate stress of the filler metal and βw is the mens in which the full penetration butt weldings between the bottom
correlation factor for the fillet welds, which for the grade S355 steel plate and a 12 mm diameter bar is well manufactured, as it is also con-
(EN 10025) is equal to 0.9. firmed by the low scatter of the experimental results (CVr = 0.076). For
The strength value of the fillet welding can be calculated by referring the specimens in which the welding failure occur, the mean value of
to the throat section of welding bead corner a, equal to 0.3 times the Fexp/Fu is 1.253 with coefficient of variation CVr = 0.08 for type A joint.
smaller diameter dw between the two welded bars (being the minimum For type B joints, three different failure modes were observed, name-
value recommended in ISO Code [15]), and the effective length of the fil- ly welding or bar rupture or delamination of the bar in the part subject-
let leff, which is assumed to be equal to (l-2a), in which l indicates the ed to large bent curvature and heat input during weld thermal cycle.
overall length of the fillet welding. Thus, experimental welding strength values are more scattered
According to the aforementioned assumptions and specimen geom- (CVr = 0.132). However, in this case also, analytical assessment is al-
etry one obtains: ways precautionary, and the mean value of Fexp/Fu is 1.137 for type B
joint. In conclusion, for both welding types, the analytical prediction is
 
Fu ¼ 0:7 f vw 2 a leff ð2Þ
Table 2
Experimental values of tests on joints.
in which the coefficient 0.7 takes into account the overturning of the Joint type Ultimate load Fexp (kN) Fu (kN) Fexp/Fu Failure mode
throat section, dw = 12 mm is the diameter of the web bars and the
A 70.81 49.03 – Bar
length l is assumed to be equal to 3 dw according to experimental eval-
66.27 – Bar
uations. The nominal ultimate stress, fu, of the filler metal appearing in 66.95a – Bar
Eq. (1) was set at fu = 510 N/mm2. 60.71a 1.24 Welding
57.03a 1.16 Welding
66.60a 1.36 Welding
Mean 64.73 1.253
CVr 0.076 0.080
Table 1 B 62.63 47.54 1.32 Welding
Tensile tests on steel specimens: stress and strain mean values. 51.52 1.08 Welding
54.27 – Delamination
Steel bar Steel bar Steel plate
47.90a 1.01 Welding
dw = 12 mm dup = 16 mm 5 mm thick
55.14a – Delamination
fy [N/mm2] 555 518 367 68.28a – Bar
εy [%] 0.277 0.260 0.183 Mean 56.62 1.137
fu [N/mm2] 632 635 519 CVr 0.132 0.143
εu [%] 11.67 9.12 15.00 a
Specimens extracted from the HSTCB prepared in plant.
60 P. Colajanni et al. / Journal of Constructional Steel Research 95 (2014) 56–70
always precautionary and it approximates the experimental results
satisfactorily.
4. Behaviour of steel truss
4.1. Push-out tests in phase I
The behaviour of the beam in phase I concerns with the transient
time of structure construction. In this phase the main problem is buck-
ling involving the compressed web bars or the upper chord of the steel
truss. In Vincenzi and Savoia [14] three mainly different buckling modes
for the steel truss are presented: 1) the instability of the single rods con-
stituting the web truss or the upper chord; 2) a wider instability of the
compressed upper chord in which the buckling length involves more
than one mesh of the truss; 3) the coupled flexural–torsional instability
of the entire truss. The classical criteria for the investigation of instabil-
ity can be employed only in the first case that is when the buckling
length can be defined from the distance between the transversal re-
straints of the upper chord or from the length of rods constituting the
web truss. On the other hand, when the transversal restraints have
not stiffness enough to avoid the displacements of nodes belonging to
the upper chord, the latter could collapse involving a buckling length
which includes more than one mesh of the truss. Finally flexural–
torsional instability could involve the entire beam and mainly occurs
in the case of high depth cross-sections. It has to be noted that in the
latter cases the critical load value is significantly lower than the Eulerian
critical load of single rods. Nevertheless, it has been proved that the
critical load inducing instability increases as the steel truss is a spatial
truss rather than a planar truss, so that both the axial and the flexural
strength of diagonal web bars can be exploited in the response
mechanism.
Aiming to better understand the behaviour of HSTCBs in phase I,
monotonic push-out tests in displacement control were carried out on
specimens realized as the scheme in Fig. 3 shows, similar to that sug-
gested in Eurocode 4 [10] for push-out tests on classical composite
steel–concrete beams. Two tests were carried out on specimens S1
and S2. Specimen S1 consists of two pieces of truss symmetrically
coupled along the bottom plate middle lines by means of a connecting
steel plate 8 mm thick, placed perpendicularly to the plates of steel
trusses. Each truss beam is constituted by an upper chord composed
a)
Strain gauges
Fig. 4. Test-set-up for push-out test in phase I: a) testing machine; b) instrumentation.
by three coupled 16 mm diameter rebars, a bottom chord constituted
by 5 mm S355 steel plate and 12 mm diameter web rebars formed at
reverse V. The height of the truss beam is 210 mm. At the top of the
specimen an 8 mm thick plate is placed orthogonally to the specimen
axis, and adequately stiffened to apply the load through the press. Spec-
imen S2 (Fig. 3b) is different from specimen S1 (Fig. 3a) only for two
longitudinal plates, 40 mm width and 5 mm thick, welded to web
bars placed in order to reduce web bar slenderness to avoid buckling
phenomena in the web bars. The load was applied by the press with a
4000 kN load carrying capacity shown in Fig. 4a) used in displacement
controlled mode by means of an electronic device recording the reaction
force and applied displacements. The strain rate utilized for the tests
was 0.1 mm/min. Two electric strain gauges were put on the upper
chords of the two coupled trusses as it is shown in Fig. 4b, in order to
evaluate the top chord strain during the test.
4.2. Experimental results
In Fig. 5 load–displacement curves obtained by push-out tests on S1
and S2 specimens are shown; the displacement is the one recorded by
the testing machine. Both curves in Fig. 5 show a similar response in
the first linear branch; then the nonlinear part preceding the peak
value (328 kN for S1 and 333 kN for S2) is characterized by an increas-
ing loss of stiffness due to geometric nonlinearity and buckling phenom-
ena, the latter arising in compressed bars subjected both to axial stress
and bending moment. In specimen S1 buckling occurred in the com-
pressed web bars, while in the S2 specimen the presence of the added
plates made it possible to delay the instability of the web bars, modify-
ing the failure mechanism. The peak load values are similar in both
specimens, while the post-peak branch of specimen S2 shows slower
strength reduction than in specimen S1. In specimen S2 buckling
occurred in the upper chord rebars which have showed better post-
critical behaviour than that of the web rebars, since they made it possi-
ble to achieve larger deformations in the plastic range with a slower
decrease in strength. In both the two tests the weldings did not undergo
failure. Fig. 6a–d) shows the specimen configurations at the end of the
tests.
The results recorded by strain gauges, reported in Fig. 7, underline
that during both the tests, the upper chord remained in the elastic
field with a maximum stress of roughly 165 N/mm2. In specimen S2
b) Force P/displacement
 P. Colajanni et al. / Journal of Constructional Steel Research 95 (2014) 56–70 61

350 experimental maximum load for the specimen S1, P = 328 kN, shows
P [kN]

that the simplified evaluation provides an effective assessment of the

300 experimental load.
A more accurate evaluation has to take into account the stiffness of
250
the truss joint connections. Therefore due to the presence of the bend-
ing moments in the truss members, the M–N plastic domain has to be
considered. At this aim a simplified expression of the domain proposed
200
by Millard and Johnson [19] is utilized:
!
150 d3 N2
Mu ¼ f y 1− u2 ð4Þ
6 Ny
100
in which d indicates the diameter of the cross-section of the circular
50 member. This simplified curve is reported in Fig. 8 and compared to an-
Specimen S1
Specimen S2
other curve which represents the correct form proposed by Campione
0
et al. [20]:
0 3 6 9 12 15 ( " sffiffiffiffiffiffiffiffiffiffi # " sffiffiffiffiffiffiffiffiffiffi!#)
displacement [mm] 1 2 1 3 6M u 1 3 6Mu
Nu ¼ f y d π−2 arcsin þ sin 2 arcsin ð5Þ
4 d fy d fy
Fig. 5. Load–displacement curves for push-out test in phase I.

The figure shows how the two formulas provide almost coincident
result.
the load–deformation curve is linear until failure, while in specimen S1
Moreover, for compressed bars the reduction due to buckling has to
the trend was nonlinear for load values greater than 200 kN due to the
be taken into account. To this aim the M–N buckling domain proposed
buckling occurring in the web bars that modified the stiffness of the
by Eurocode 3 [18] can be considered:
stress transfer mechanism between upper chord and steel plate.
N M
4.3. Interpretation of experimental results þ k 0 ≤1 ð6Þ
Nb Mu
In order to interpret the experimental behaviour, an approximate
in which k is an interaction factor, and M0u = Wpl fy is the plastic bending
evaluation of specimen strength is made in this section.
moment corresponding to a zero axial force (Wpl = plastic section
According to Eurocode 3 [18] the buckling strength of a compressed
modulus).
member, Nb, is expressed as follows:
The k factor can be evaluated, for plastic cross-sectional properties,
class 1, as suggested in Eurocode 3 (Method 1 in Annex A) [18] as:
Nb ¼ χ Ny ; Ny ¼ A f y ð3Þ
μ 1
k ¼ Cm ð7Þ
with A the cross-section area of the member and χ the reduction factor N C
1−
for the buckling mode. Ncr
In order to evaluate the reduction factor χ for the bars in which buck-
ling occurs, the non-dimensional slenderness is evaluated and the buck- with
ling curve is chosen.
For the web bar having diameter d w = 12 mm, and length N
1−
lw = 250 mm in the specimen S1, the buckling length is assumed Ncr N
μ¼ ; C m ¼ 0:79 þ 0:21ψ þ 0:36ðψ−0:33Þ ; ψ ¼ −1;
to be 175 mm, by considering a semi-rigid connection (reduction N Ncr
1−χ
coefficient 0.7), the slenderness 58.3 and the non-dimensional Ncr 
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

slenderness λ ¼ A f y =Ncr ¼ 0:96, being the bar cross-section of 1:6 2 1:6 2 2 N W W pl
C ¼ 1 þ ðw−1Þ 2− C m λ− Cmλ ≥ el ; w ¼ ≤ 1:5;
class 1, and the elastic flexural buckling force Ncr = 67.6 kN. Refer- w w Nu W pl W el
ring to the buckling curve c, the imperfection factor is 0.49 and the
 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2 in which ψ takes the shape of the moment diagram into account and Wel
reduction factor χ ¼ 1= Φ þ Φ2 −λ ¼ 0:56 , in which Φ ¼ 0:5 is the elastic section modulus. In this case ψ = −1 is chosen consider-
h   2
i ing the same restraint conditions at the bar ends. The interaction
1 þ 0:49 λ−0:2 þ λ ¼ 1:15:. Thus, the reduced axial strength
curve obtained is also represented in Fig. 8 for the web bar with
in Eq. (3) is Nb = 35.2 kN. d = dw = 12 mm.
By approximating specimen S1 behaviour to that of a simple truss in With the aim of representing simplified linear domains considering
which shear forces and bending moments are neglected at the joints, an buckling effects in compression, in the same figure, also the curves cor-
approximate estimation of the maximum load can be obtained taking responding to the following equation are reported:
into account the contribution of 16 web bars only (8 of them com-
pressed and the other 8 in tension), assuming that the web bars at the N M
þ ≤1 ð8Þ
bottom are unloaded (see Fig. 3). In this way the maximum load can N M
be evaluated by assuming linear specimen behaviour up to the buckling
of the compressed web rebars, by adding the contribution of 16 bars where N⁎ = Ny in tension and N⁎ = Nb in compression, M⁎ = Wy fy
with axial strength Nb calculated as in Eq. (3). Thus the strength load (where Wy is the elastic modulus) has to be read as My in the case of
P = 16 × 35.2 cos(α*) = 306.74 kN is obtained, α* = 57° being the the yielding domain while it has to be read as M0u in the case of the plas-
angle of inclination of the web bar axis with the longitudinal axis of tic domain. Furthermore, in Fig. 8 points C and T correspond to the Nb
the specimen (see Fig. 16). In Table 3 the comparison with the and Ny values respectively.
62 P. Colajanni et al. / Journal of Constructional Steel Research 95 (2014) 56–70

All the statements and expressions in the following refer to one half
of the steel truss belonging to the plane which contains the axis of the
web bars of the truss itself.
Aiming at estimate the maximum load value taking into account the
axial force redistribution in the web bars allowed by shear forces and re-
lated flexural moments in the web bars, the equilibrium equations of
the planar system represented in Fig. 9 can be written assuming the fol-
lowing hypotheses:
e) axial force in bar n°5 vanishes, and therefore the bending moment is
the plastic one;
f) the bending moment in the rod BC vanishes at a distance of one third
of the bar length from the node C.
On the bases of the aforementioned assumptions and referring to the
symbols introduced in Fig. 9, the following equilibrium equations are
derived:

a) the upper chord bar remains elastic, as it had been observed by the 2 2ðM1 þ M2 Þ þ M3 þ M4
 
measurements recorded by the strain gauges during the test (Fig. 7); ðN1 −N 2 Þ sinα − ðM1 −M2 Þ cosα − ¼ 0 ð9Þ
lw 2lc
b) the web bars have reached their maximum capacity, that is in the
tensile bars the flexural moments and axial forces are linked by
Eq. (4), while in the compressed bars by Eq. (8) (by assuming a sta-  2  3M5 ρCB
ble post-buckling behaviour); ðN3 −N 4 Þ sinα − ðM3 −M4 Þ cosα −
lw lc ρCB þ ρCD
c) the distribution of internal forces in node B (see Fig. 9a) of the upper 2ðM 1 þ M 2 Þ þ M 3 þ M 4
chord is anti-symmetric; thus bending moments MBC = MBA; þ
2lc
d) the bending moment diagrams in web bars are anti-symmetric; ¼0 ð10Þ

a) b)

c) d)

Fig. 6. Specimens of push-out tests in phase I at the end of the test: a) specimen S1; b) zoom of web bars in specimen S1; c) specimen S2; d) zoom of upper chord in specimen S2.
 P. Colajanni et al. / Journal of Constructional Steel Research 95 (2014) 56–70 63

350 On the basis of the reduced strength, and taking into account that
P [kN]

each of the two symmetrical beam top chords is made of 3 rebars, an

300 upper bound of the expected specimen strength related to the top
chord rebar buckling can be predicted as P = 6 × Nb = 399 kN. A
250 more effective prediction of the specimen strength should be obtained
by reducing the latter value taking into account the presence of the
bending moment on the top chord rebars transmitted by the web
200
bars. However in this case it is not easy to achieve this goal in a simpli-
fied way, since both the web and the upper bar not involved in the buck-
150 ling remain in the elastic field. Finally it is worth noting that, even if in
the top chord buckling represents one of the most relevant failure
100 modes for the trussed beam in phase I, the experimental value obtained
for specimen S2 is influenced by the setup geometry, that is not able to
50 effectively reproduce the restrain condition of the actual beam during
Specimen S1
the cast phase.
Specimen S2
0
0 0.0002 0.0004 0.0006 0.0008 0.001 5. Composite beam after concrete casting
ε
5.1. Push-out tests in phase II
Fig. 7. Load–strain curves of strain gauge put on the upper chord.
As aforesaid in the previous paragraphs, push-out tests on speci-
mens representing HSTCBs in phase II were developed in order to inves-
tigate the stress transfer mechanisms between the steel elements
! !
X
5
 2 X
5

constituting the truss of the beam and the surrounding hardened con-
S¼ Ni cosα þ Mi sinα ð11Þ crete. Although there is a large variety of HSTCB typologies that are
i¼1
lw i¼1
used in different kind of structures, such as buildings, bridges, big-
span roofs, park buildings, etc., only few authors have presented exper-
being ρCB and ρCD the flexural stiffness of CB and CD rods respectively; lw imental results [7–9,21–23] addressed to study this issue of paramount
and lc the length of the web and the upper chord bars respectively. Par- importance for the behaviour of HSTCBs both under serviceability and
ticularly, with reference to the coordinate system in Fig. 9 b), under ultimate loads.
Eqs. (9)–(11) represent the equilibrium of the node A in the y- Experimental investigation by push-out tests on HSTCBs was per-
direction, the equilibrium of node B in the y-direction and the equilibri- formed on specimen typologies and test modes adopted by Eurocode
um of the entire system in the x-direction respectively. 4 [10] for classical composite beams. In all the cases examined in this
In order to find the maximum value of the external force S that the paper tests were carried out on beams realized with the same steel
system is able to withstand, the lower bound theorem of plasticity is truss used for tests in phase I. Again, two pieces of beams were coupled
used, according to the equilibrium Eqs. (9)–(11) and the admissible with a central steel plate conveniently strengthened at its ends where
plastic and buckling conditions expressed by Eqs. (4) and (8). Thus, the load is applied, as shown in the photo in Fig. 10a) and in the scheme
the estimated analytical value of the maximum load, P, is obtained in Fig. 10b). Results obtained by tests carried out by the authors are uti-
from the maximum value of S considering the couple of two trusses con- lized in this section together with some of the results available in the lit-
stituting the beam. In this way, the value P = 326.64 kN is obtained erature, in order to define simple analytical models able to predict the
(this value with respect to the experimental one is reported in maximum load exhibited by the HSTCB during the push-out test.
Table 3), corresponding to the internal forces values indicated in Fig. 8 Three identical specimens P1, P2 and P3 were cast at the same time
by the point C for compressed web bars n°1 and 3, points T2 and T4 for push-out tests with the geometry shown in Fig. 10b). Specimens P1
for the tensile bar n°2 and 4 and the point M5 for the bar n°5 in pure and P2 were tested 1 month after casting, when concrete was charac-
bending. terized by three compressive tests on cylindrical specimens having di-
For the S2 specimen, due to the presence of the stiffening plate, a dif- ameter 100 mm and height 200 mm that provided a mean strength
ferent collapse mechanism is expected, involving buckling of the more value fcm = 27.77 N/mm2. Specimen P3 was tested about three months
compressed top chord bar, i.e. bar BC. The buckling load of the top later.
chord bar having diameter 16 mm and length 300 mm is evaluated as In order to evaluate the stress in the web bars, electric strain gauges
follows: the buckling length is assumed to be 210 mm by considering were placed before concrete casting as shown in Fig. 10c) and 10d). The
a semi-rigid connection (reduction coefficient 0.7), the slenderness position of the strain gauges is depicted in Fig. 10b). In all specimens one
52.5 and the non-dimensional slenderness λ  ¼ 0:84; being the bar strain gauge (E1) was placed on the compressed upper chord. More-
cross-section of class 1 and the elastic flexural buckling force over, in specimens P1 and P2 a strain gauge was placed on the web
Ncr = 148.3 kN. Referring to the buckling curve c, the imperfection fac- bar subjected to a tensile load (E2) and another one on the compressed
tor is 0.49 and the reduction factor χ = 0.64, calculated assuming for web bar (E3) (see Fig. 10c); in specimen P3, as a substitute, two strain
the coefficient Φ the value 1.01. Thus, the reduced axial strength is gauges for each web bar were placed (E2.1, E2.2 and E3.1, E3.2) in
Nb = 66.5 kN. order to measure possible stress gradient in web bars (see Fig. 10d).
After cure and maturation, in all specimens two other strain gauges
(E4 and E5) were also placed on the steel plate on two sides of the
truss bottom joint, in order to evaluate the stress transfer in the truss
Table 3 joint.
Comparison between theoretical and experimental results (phase I): Pu,theo/Pu,exp. Tests were carried out in displacement control by using the same
testing machine as utilized for tests on bare trusses. According to the re-
ID Truss model Frame model
quirements of Eurocode 4 for push-out tests on classical steel–concrete
S1 0.935 0.996 beams, 25 load cycles were preliminarily applied with a maximum
S2 1.198 –
value of 200 kN for specimen P1 and 400 kN for specimens P2 and P3.
64 P. Colajanni et al. / Journal of Constructional Steel Research 95 (2014) 56–70

200
M [kNmm] plastic domain Eq. (5)
M5
simplified plastic domain
150 Eq. (4)
T2
100
T4

50

C 0 T
-70 -50 -30 -10 10 30 50 70
N [kN]
-50

-100
simplified yielding buckling
plastic buckling EC3
Eq. (8)
Eq. (6) -150
simplified plastic buckling
Eq. (8)
-200

Fig. 8. M–N domains for web bar with dw = 12 mm.

5.2. Experimental results
In Fig. 11 load–displacement curves are reported for the three tests.
The displacement values are those recorded by the test machine and
concern the slipping between the concrete and the steel plate, the latter
assumed to be rigid in its own plane.
The maximum load values recorded for P1 and P2 specimens are
equal to 1051.6 kN and 1004.9 kN respectively while for specimen P3
a higher value was achieved, equal to 1257.5 kN. The curves for the
three specimens show a similar slope up to a load value of 800 kN,
confirming the reliability of the results obtained. The increment in con-
crete stiffness of specimen P3 due to longer maturation interval was
found to be negligible. Appreciable cracks began to appear at a load
value of 1000 kN; they propagated until the maximum loads corre-
sponding to the maximum concrete tensile strength were achieved. In
the descending branches the curves show the progressive strength re-
duction that during the test was associated with a typical noise of break-
ing of metallic parts; it can be observed that the specimens exhibited a
somewhat ductile branch to which a rapid increase in vertical cracking
of the concrete corresponds; the configuration of specimen P1 at the
end of the test can be observed in photos in Fig. 12. Particularly in
Fig. 12a) side views of the specimen are shown, while in Fig. 12b) a
view of the top face of the specimen is represented (similar failure
type was observed for the other specimens).

a) b)
C B A
D
5 4 3 2 1
S = P/2

c)

S VCD VCB VBC VBA VAB

B A
C
MCD MCB MBA MAB
MBC
V5 V3 V1
V4 V2

N5 M4 N3 M2 N1
N4 N2
M1
y M5 M3
α∗
x
Fig. 9. Simplified model: a) real system; b) planar system; c) equilibrium of the upper chord at each joint.
 P. Colajanni et al. / Journal of Constructional Steel Research 95 (2014) 56–70 65

a) b) P
E - strain gauges

140

E2.1
E4
E2
E1 E2.2
E3.2 675
E3
790
E3.1 E5

s 100
250 200 250
700
3 210
50

300

2 250
c) e)

d)

Fig. 10. Specimen for push-out test in phase II: a) photo of a specimen; b) geometry and instrumentation (dimensions in millimetres); c) strain-gauges in specimens P1 and P2; d) strain-
gauges in specimen P3; e) specimen P2 collocated into the test machine.

For all three specimens failure was due to the achievement of tensile
1400
strength in concrete with formation of large amplitude cracks in the lon-
1200
gitudinal direction of the specimen. The cracking caused a progressive
transfer of tensile stresses from concrete to tensile web bars of steel
1000 trusses. The latter underwent large plastic strains near the connection
to the plate, as was seen after removing the concrete from the specimen.
In particular by analysing the photos in Fig. 13 relating to specimen P1, it
P [kN]

800
emerges that the compressed bars exhibited large deformations at a dis-
600 tance of about 50 mm from the plate, presumably corresponding to the
longitudinal crack. Five tensile bars progressively failed causing partial
400 detachment of trusses from plates (see Fig. 13c). A progressive reduc-
Specimen P1 tion in the strength of the specimen was revealed, corresponding to
200 Specimen P2 the progressive rupture of tensile web bars near to the plate. In all
Specimen P3
three tests the welding remained undamaged.
0
0 5 10 15 20 Recording in electrical strain gauges located as shown in Fig. 10 is
displacement [mm] depicted in Fig. 14a and b. The curves represent the strain measured
by the strain gauges put into diagonal bars respectively for specimens
Fig. 11. Load–displacement curves for push-out test in phase II. P1 and P2. Specifically, strain gauge E2 measured strain in the tensile
66 P. Colajanni et al. / Journal of Constructional Steel Research 95 (2014) 56–70
a)
b)
Fig. 12. Specimen P1 at failure: a) lateral views; b) top view.
web bar and strain gauge E3 in the compressed one. The curves clearly
show that compressed web bars remained in the elastic range during
the phase of test before the concrete collapse, while the tensile bars
overcame the yielding strain. The curves relating to compressed web
bars were interrupted when the growing damage to the specimen
made strain gauge measurements unreliable.
a)
Fig. 13. Deformed steel truss of specimen P1 at end of test: a) global view; b) zoom of deformed bars; c) zoom of failure zone near the plate to bar welding in a tensile bar.
Nevertheless, it should be noted that welding between web bars and
upper chord as well as web bars and bottom steel plate can be consid-
ered as fixed restraints; therefore bending moment and axial force de-
velop in the bars, and therefore the end sections of bars are the most
stressed. For this reason, in specimen P3 two strain gauges for each
web bar were placed close to the bar ends.
In Fig. 14c) deformations recorded by strain gauges placed in the
web bars of specimen P3 are plotted. Specifically, strain gauges E2.1
and E3.1 were placed near to the upper chord and strain gauges E2.2
and E2.3 were placed near to the steel plate. Deformation pertaining
to the former has a similar pattern to those of the tests above reported,
in which the strain gauges were placed at the middle of the web bars,
while those placed near to the steel plate recorded larger deformations
both in the tensile and the compressed bar, stressing that plastic defor-
mations are exhibited in both tensile and compressed web bar sections
close to the plate.
In all three specimens, the yielding strain was reached in the tensile
bar and plastic hinges formed in the compressed bar also after the at-
tainment of the full strength, as a consequence of a through crack in
the concrete (as can be seen in Fig. 13b for specimen P1).
Fig. 15 shows the curves of the recording of strain gauges E4 and E5
placed on the steel plate of all three specimens. Low strain values show
that the plate remains elastic; the progressive stress transfer from con-
crete to beam truss plate is underlined by the differences in the readings
of the strain gauges; however, it should be noted that the presence of
the additional steel plate orthogonal to bottom steel plates of the testing
specimens, and connecting the two pieces of HSTCBs renders very com-
plex the evaluation of the stress state of truss steel plate due to stress
transferring from concrete. Finally, Fig. 16 shows the plot of the curves
representing the deformations recorded for all specimens by strain
gauges E1 placed on the upper chord which remains elastic in all tests.
5.3. Analytical model for strength evaluation
In order to evaluate the strength of the steel to concrete connection
in HSTCBs, results of push-out tests obtained in the present research de-
scribed in the previous section, together with some results present in
the literature [21–23], were interpreted through a typical truss model,
according to indications present in the Italian code [17] as well as in
Eurocode 2 [11] for R.C. beams. To this aim the HSTCBs were considered
as belonging to the typology of R.C. structures. Thus, an equivalent pla-
nar truss model (see Fig. 17) was defined, consisting of a compressed
concrete top chord, tensile bottom steel plate, web tie made up of
b)
c)
 P. Colajanni et al. / Journal of Constructional Steel Research 95 (2014) 56–70 67

a) b)
1200 1200

1000 1000

800 800

P [kN]

P [kN]
600 600

400 400

200 200
E2
E2
E3 E3
0 0
-0.002 -0.001 0 0.001 0.002 0.003 0.004 -0.002 -0.001 0 0.001 0.002 0.003 0.004
ε ε

c)
1400

1200

1000
P [kN]

800

600

400 E2.1
E2.2
200 E3.1
E3.2
0
-0.004 -0.002 0 0.002 0.004
ε

Fig. 14. Load–strain curves of strain gauges put on web bars: a) specimen P1; b) specimen P2; c) specimen P3.

diagonal steel bars and steel and concrete struts inclined at an angle α should be noted that, approximately, it is possible to refer to the statical-
and ϑ respectively, the latter chosen so that it makes the system ly determinate truss system obtained neglecting the contribution of
strength maximum, fulfilling the condition that 1 ≤ ctgϑ ≤ 2.5. It steel compressed web bars and, using the expressions present in the

a) b)
1200 1200

1000 1000

800 800
P [kN]

P [kN]

600 600

400 400

200 200 E4
E4
E5 E5
0 0
-0.001 -0.0005 0 0.0005 -0.001 -0.0005 0 0.0005
ε ε

c)
1400

1200

1000
P [kN]

800

600

400

200 E4
E5
0
-0.001 -0.0005 0 0.0005
ε

Fig. 15. Load–strain curves of strain gauges put on steel plate: a) specimen P1; b) specimen P2; c) specimen P3.
68 P. Colajanni et al. / Journal of Constructional Steel Research 95 (2014) 56–70

1400
a)
1200

1000

800
P [kN]

600

400
b)
Specimen P1
200
Specimen P2
Specimen P3
0
-0.0003 -0.0002 -0.0001 0 0.0001
ε

Fig. 16. Load–strain curves of strain gauge E1 put on the upper chord.

Italian code and in Eurocode 2, properly adapted to the current problem

(see Fig. 18). Thus it is possible to calculate the failure load as the min- c)
imum value corresponding to collapse of compressed concrete or ten-
sile steel.
With reference to the symbols in the single mesh in Fig. 17 b), the
slippage force is P/n, n being the number of meshes which can form
in relation to the choice of the angle ϑ and to the geometry of the
specimen, and considering the ultimate strength in the compressed con-
crete strut (whose cross-section area is equal to bw t, with t = s sinϑ)
Sc,u = bw t f′cu or the ultimate strength in the tensile web bars Ss,u = Asw
fsu, the following expressions can be obtained respectively:
Fig. 18. Truss model: a) Eurocode 2 (Model 1); b) maximum number of web struts (Model
0   2); c) concrete struts with different slope (Model 3).
2
P Rc;u ¼ n bw s f cu ðctgα þ ctgϑÞ= 1 þ ctg ϑ ð12aÞ
presence of steel compressed web bars inclined α renders possible the
P Rs;u ¼ n Asw f su ðctgα þ ctgϑÞ sinα ð12bÞ existence of meshes in the steel truss inclined at an angle ϑ so that
ctgα ≤ ctgϑ; thus when ctgα = ctgϑ is assumed, Eq. (12b) gives the
where s is the distance between successive web bars, bw the width of the resistance of the bare steel truss, instability of the compressed bar
cross-section, Asw the area of the web steel reinforcement, fsu and f′cu being prevented.
respectively the values of the ultimate strength in steel and the reduced In this section the model described will be indicated as Model 1; be-
strength in the concrete (f′cu = 0.5 fcu). It can be observed that the cause of the geometry, in some cases this model cannot be applied, so

a)

b)

Fig. 17. Truss planar model for push-out specimen.

 P. Colajanni et al. / Journal of Constructional Steel Research 95 (2014) 56–70 69

two other models are considered with different inclinations of the strut Table 5
concrete element, called Model 2 and Model 3. All the models are based Specimen features for push-out tests and experimental results. Aiello 2009.

on the truss model but with different inclinations of the strut concrete ID Web bars Upper chord Ultimate load [kN] Failure
element, as shown in Fig. 18. Model 1 in Fig. 18a), is based on the appli-
S12 No.2 bars No.1 bar 829.6 Concrete
cation of Eq. (12) as usually employed for reinforced concrete. The ϑ (dw = 12 mm) (dup = 18 mm)
angle is chosen in order to make the steel and concrete strength equal, S14 No.2 bars No.1 bar 818.8 Concrete
and having modified the lower limit of ctgϑ, leading to the following as- (dw = 14 mm) (dup = 18 mm)
B12 No.2 bars No.1 bar 1001.8 Concrete
sumptions:
(dw = 12 mm) (dup = 18 mm)
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
0
B14 No.2 bars No.1 bar 1139.0 Steel
(dw = 14 mm) (dup = 18 mm)
sbw f cu
ctgα ≤ctgϑ ¼ −1 ≤2:5: ð13Þ
Asw f su sinα

In some cases the angle obtained by applying the condition in

Eq. (13) is not compatible with the geometry of the specimen because
the geometry of the specimens does not permit the formation of the
equivalent truss mesh. A second option, namely Model 2, is obtained
by choosing the angle ϑ = ϑ* in order to achieve the largest number
of strengthening meshes of the steel–concrete truss, taking advantage
of the presence of the diagonal compressed steel bar by taking into ac-
count the lower limit for ϑ as in Eq. (13). It leads to the structural
scheme represented in Fig. 18b). A further increase in strength predict-
ed by the model can be obtained with Model 3, allowing each mesh to
have a different inclination of the concrete strut as indicated in
Fig. 18c), compatibly with the limitation in Eq. (13) and those imposed
by the specimen geometry.
5.4. Experimental test versus analytical model results
The ratio between the strength predicted by the models and that re-
corded by experimental tests is reported in Table 6, both for experimen-
tal tests obtained by the authors and the experimental tests in [18,19] in
which an analogous failure mode was verified in the concrete strut or in
the steel web bar. The features of the specimens, ultimate load value and
failure modes of tests described in the abovementioned papers [21,22]
are presented in Tables 4 and 5 respectively.
In the specimens tested by Puhali and Smotlack [21] two types iden-
tified as A and B in Table 4 are considered: the one with a single planar
(2D) truss (group A) and the other with a double spatial (3D) truss
(group B). For each typology three specimens were manufactured; the
steel plate utilized in all specimens is 6 mm thick and the nominal yield-
ing strength of the steel is 355 N/mm2; other features are presented in
Table 4, where dw indicates the diameter of the web bars and dup the di-
ameter of the bars constituting the upper chord, both realized with steel
having nominal yielding strength equal to 355 N/mm2 (S355); the con-
crete used had Rck = 26–30 N/mm2.
An experimental investigation carried out by Aiello [22] presents the
results of nine experimental tests carried out on specimens of four ty-
pologies as indicated in Table 4. All specimens contained three meshes
with double spatial (3D) steel truss. It can be observed that, in all spec-
imens, the bars constituting the upper chords were made of steel having
nominal yielding stress 450 N/mm2, while for web reinforcement the
same type of steel was utilized for specimens B12 and B14, and S355
type steel for S12 and S14. All specimens had a S355 steel plate
of 6 mm thickness and double web bars. The concrete had
Rcm = 42.6 N/mm2. The ultimate load value used in this section for in-
terpretation of the results is the mean value of each of four specimen
typologies.
The experimental investigation in Tullini et al. [7,23] concerned
three identical specimens whose truss was realized with S355 steel
plate 4 mm thick, double web bars of 12 mm diameter and ribbed
steel upper chord with diameter 18 mm; the bars were made up of
steel with nominal yielding stress 440 N/mm2 and the concrete had
Rcm = 42.24 N/mm2. The results have not been considered here be-
cause the ultimate load was achieved prematurely due to welding fail-
ure. The result obtained in [21] concerning specimen B-P6 has also
been excluded because the ultimate load was achieved prematurely be-
cause of steel plate instability.
It can be observed in Table 6 that Model 1 only furnishes a value in a
few cases; in other cases the specimen geometry is not compatible, as
explained above. In the same table the mean value, rm, and the coeffi-
cient of variation of the ratio between the theoretical ultimate load
and the experimental one are shown.
Model 2 provides an estimation of the mean value of the specimen
strength alongside the conservative experimental value, unlike Model
3, that gives a mean value of Pu,teo/Pu,exp = 1.027. The results
from Model 2 are also characterized by a higher coefficient of variation
(CVrMod.2 = 0.434) than those of Model 3 (CVrMod.3 = 0.277). Even if a
mean value of the ratio between model and experimental strength close
to the unity is found for Model 3, the still large value of the CVr stresses
that the model cannot be considered fully reliable.

Table 6
Comparison between theoretical and experimental results (phase II): Pu,theo/Pu,exp.

Table 4 ID Model 1 Model 2 Model 3

Specimen features for push-out tests and experimental results. Puhali and Smotlack 1980. A-P1 – 0.593 0.593
A-P2 – 1.247 1.247
ID Web bars Upper chord Ultimate load [kN] Failure
A-P3 – 1.702 1.702
A-P1 No.1 bar No.2 bars 463.0 Steel B-P4 – 0.747 0.747
(dw = 14 mm) (dup = 14 mm) B-P5 – 1.371 1.371
A-P2 No.1 bar No.2 bars 647.5 Concrete P1a 0.624 0.696 0.783
(dw = 24 mm) (dup = 24 mm) P2a 0.653 0.728 0.820
A-P3 No.1 bar No.2 bars 843.5 Concrete P3a 0.522 0.582 0.655
(dw = 32 mm) (dup = 32 mm) S12 – 0.643 1.182
B-P4 No.2 bars No.3 bars 735.5 Steel S14 – 0.591 0.843
(dw = 14 mm) (dup = 14 mm) B12 – 0.675 1.241
B-P5 No.2 bars No.3 bars 1177.5 Concrete B14 0.674 0.673 1.441
(dw = 24 mm) (dup = 24 mm) Average value rm 0.618 0.854 1.027
B-P6 No.2 bars No.3 bars 1324.5 Steel plate buckling CVr 0.109 0.434 0.277
(dw = 32 mm) (dup = 32 mm) a
Present investigation.
70 P. Colajanni et al. / Journal of Constructional Steel Research 95 (2014) 56–70
6. Summary and conclusions
The results obtained by an experimental investigation concerning
stress transfer mechanisms among steel elements and concrete in
HSTCBs are presented.
With reference to the bare steel truss (phase I), the results have
shown that failure is governed by buckling of compressed web or
upper chord bars. In the first test failure occurred in diagonal bars (spec-
imen S1), with an extremely brittle collapse. When stiffeners for web
bars were placed in order to inhibit web bars instability (specimen
S2), failure was due to the buckling of the upper chord bars; only a slight
increase in the ultimate load was obtained, but better post-peak behav-
iour. In none of the tests was the welding damaged.
The results of push-out tests carried out after the concrete casting
(phase II) have shown that the specimens exhibit almost brittle failures
due to the collapse of the concrete in tension, with the steel tensile web
rebars that developed large inelastic deformations, mainly concentrated
at the ends close to the bottom steel plate. In none of the tests in phase II
was the welding damaged.
As a consequence it can be also concluded that the usual dimensions
of welding ensures system resistance in both phases I and II.
HSTCB strengths in push-out tests were predicted through suitably
adapting the truss model with variable angle of the concrete strut, typ-
ical for R.C. elements. The scattering resulted in the strength predictions
shows that these models are not enough accurate to interpret the
HSTCB failure mode, showing that the failure mechanism in this struc-
tural beam typology has to be represented specifically taking into ac-
count the stress transfer mechanisms between steel and concrete.
References
[1] Hsu HL, Hsieh JC, Juang JL. Seismic performance of steel-encased composite mem-
bers with strengthening cross-inclined bars. J Constr Steel Res 2004;60:1663–79.
[2] Johnson RP. Composite structures of steel and concrete. 1: beams, slabs, columns
and frames for buildings. London: Blackwell Scientific Publication; 1994.
[3] Ju YK, Kim JY, Kim SD. Experimental evaluation of new concrete encased steel com-
posite beam to steel column joint. J Struct Eng ASCE 2007;133(4):519–29.
[4] Amadio C, Macorini L, Suraci G. Structural performance of a new hybrid RC-encased
steel joist system. Advances in reinforced concrete and precast constructions. In: di
Prisco Marco, editor. Starrylink; 2008. p. 19–29 [Milan, Italy].
[5] Amadio C, Macorini L, Sorgon S, Suraci S. A novel hybrid system with RC-encased
steel joists. Eur J Environ Civ Eng 2011;15(10):1433–63.
[6] Badalamenti V, La Mendola L, Colajanni P. Seismic behaviour of hybrid steel trussed
concrete beams. Proceedings of 14th European Conference Earthquake Engineering
2010, 1527. Curran Associates, INC.; 2010. p. 1–8 [Ohrid, Macedonia, CD-rom].
[7] Tullini N, Minghini F. Nonlinear analysis of composite beams with concrete-encased
steel truss. J Constr Steel Res 2013;91:1–13.
[8] Tesser L, Scotta S. Flexural and shear capacity of composite steel truss and concrete
beams with inferior precast concrete base. Eng Struct 2013;49:135–45.
[9] Cancelliere N, Colajanni P, La Mendola L. On bottom steel plate to concrete anchor-
age in hybrid steel trussed concrete beams. STESSA 2012 Behaviour of Steel Struc-
tures in Seismic Areas, 9–12 January. Santiago del Chile: CRC Press/Balkema
Publishers; 2011. p. 243–8.
[10] Eurocode 4. Design of composite steel and concrete structures—Part 1–1: general
rules and rules for buildings; 2005.
[11] Eurocode 2. Design of concrete structures—Part 1–1: general rules and rules for
buildings. European Committee for Standardization (CEN); 2005.
[12] Eurocode 3. Design of steel structures—Part 1–8: design of joints. European Commit-
tee for Standardization (CEN); 2005.
[13] Colajanni P, La Mendola L, Recupero A. Experimental test results vs. analytical
prediction of welded joint strength in Hybrid Steel Trussed Concrete Beams
(HSTCBs). European Journal of Environmental and Civil Engineering
2013;17(8):742–59. http://dx.doi.org/10.1080/19648189.2013.815135.
[14] Vincenzi L, Savoia M. Stabilità di tralicci PREM in prima fase (Stability of PREM-beams
in the first phase). Proceeding of 18th C.T.E. Congress, (in Italian) Brescia, 11–13 Nov
2010; 2010. p. 849–58 [Falciano—San Marino, Imready Edition].
[15] ISO 17660-1. Welding – welding of reinforcing steel – Part 1: load-bearing welded
joints; 2007.
[16] UNI EN ISO 6892-1. Metallic materials – tensile testing – Part 1: method of test at
ambient temperature; 2009.
[17] D. M. LL. PP. Norme Tecniche per le Costruzioni (Construction Technical Codes).
Gazzetta Ufficiale; 14 Jan 2008 [(in Italian), 04 Feb 2008].
[18] Eurocode 3. Design of steel structures—Part 1–1: general rules and rules for build-
ings. European Committee for Standardization (CEN); 2005.
[19] Millard SG, Johnson RP. Shear transfer across cracks in reinforced concrete due to
aggregate interlock and to dowel action. Mag Concr Res March, 1984;36(No.126).
[20] Campione G, Fossetti M, Minafò G, Papia M. Influence of steel reinforcements on the
behavior of compressed high strength R.C. circular columns. Eng Struct Jan.
2012;34:371–82.
[21] Puhali R, Smotlack I. Relazione sulle prove di push-out atte a determinare le leggi di
carico-scorrimento delle travi in sistema composto tipo“REP”. (Report on the
push-out tests fit for the determination of load–slip laws of REP composite truss
beams)Science of Constructions Institute Acts, University of Trieste; 1980 [in Italian].
[22] Aiello MA. Analisi sperimentale della connessione acciaio-calcestruzzo nelle travi
reticolari miste (Experimental analysis of steel–concrete connection in hybrid
truss beams). Proceeding of 7th Italian Workshop on Composite Structures. Aesse;
2009. p. 33–42 [(in Italian). Benevento (Italy)].
[23] Tullini N, Reato P, Cappellozza M. Indagini sperimentali su travi miste acciaio-
calcestruzzo con connessione a traliccio (Experimental investigations on hybrid
steel–concrete beams with truss connection). Proceeding of 16th CTE Congress;
2006 [(in Italian), Parma (Italy)].