th

13 International Conference on
Steel, Space and Composite Structures
31 January - 2 February 2018, Perth, Australia

BEHAVIOUR OF STEEL AND CONCRETE COMPOSITE TRUSSES

Jan Bujnak* and Petra Bujnakova †

*University of Zilina, Univerzitná 8215/1,010 26 Žilina, Slovakia and Opole University of

Technology, Katowicka 48, 45-061 Opole, Poland
e-mail: jan.bujnak@fstav.uniza.sk

Keywords: Composite Structures, Trusses, Connection, Experimental testing

Abstract. The design specifications of composite trusses are only partially

included in the standards. However this construction system can be considered
as one of the most economical for building and bridge structures. To create the
interaction between steel and concrete, it is necessary to prevent the relative slip
at the steel and concrete interface using the shear connectors. But the local
effects of a concentrated longitudinal force and the distribution of the shear force
between steel section and concrete slab, as special task, should be appropriately
examined. The finite element analyses can be used to investigate numerically this
structural system behaviour, exploiting several computer procedures. The
extensive experimental testing has been also executed. The outputs of this
theoretical and experimental research are presented in the paper.

1 INTRODUCTION
Composite steel-concrete trusses can be considered as one of the most economical systems
for building, especially for greater spans, commonly to the 20 m. The structural elements of this
composite type can be used for even greater spans up to the 30 m, which allows better use of
internal space without restricting columns. The trusses are appropriate also to meet the
requirements for building height limitation, the need to run complex electrical, heating, ventilating,
and communication systems.
Also composite steel bridges, whose carriageway deck is supported on a filigree steel truss
structure and slim piers, are particularly preferable especially to ordinary concrete bridges.
Primarily considering the technical and architectural aspects as well as compromi se between
protections of the landscape on the one hand and hard transports necessities on the other. Thus
a composite truss bridge, with its speedy assembly engineering, can be a structural type which is
both economically and aesthetically attractive. To create the interaction between steel parts and
concrete, it is necessary to prevent the relative slip at the steel -concrete interface using the shear
connectors. Nowadays, different types of shear connectors are used. In our investigation, shear
connection is developed using the welded headed studs.

2 LONGITUDINAL SHEAR TRANSFER

In composite trusses, the longitudinal forces are introduced into the concrete slab only locally
at points of the increase of the axial force in the chord, hence in the nodes, where the web
members are connected to the compressed chord. In existing standar ds, there are no particular
1,
rules for the design of composite truss, except the formulas in Eurocode 4 clause 6.6.2.3 for the

†
University of Zilina, Univerzitná 8215/1, 010 26 Žilina, Slovakia

1
 Jan Bujnak and Petra Bujnakova

local effect of a concentrated longitudinal force and the distribution of the longitudinal shear force
into local shear flow between steel section and concrete slab.
Primarily therefore the degree of connection effect has been theoretically studied. Especially
the connector diameter impact, the top chord section influence and the material characteristics of
steel and concrete were analysed pondering over the stiffness and the resistance of the beams
and the shear forces in the connectors.
Figure 1 shows a typical portion of a studied composite sample, taken from the references
2
Machacek & Cudejko , consisting of a concrete deck (1500×80 mm) and a steel truss beam, whose
the top chord was designed as ½ IPE 220, the bottom chord a rectangular hollow section RHS
60×60×4 and the web members were from RHS 50×50×3. During the analytical phase of this
3
composite truss, the finite-element-based software CAST3M was used to evaluate the structural
integrity of the system. Serious considerations had to be given to proper representation of the
geometric characteristics. To ensure a full composite action, shear headed studs connectors (Φ = 19
mm) were provided at the interface between the concrete slab and the truss top chord to resist
interface shear. Therefore, it was important during the initial modelling stages to take into account the
node positions as key locations of interest. Thus the model with a higher degree of refinement could
be assembled such that important results could be obtained at these zones. Proper type and shape of
the elements had also to be considered since different mesh size could sometimes cause significant
variations in results. The resulting computer model was used to evaluate the structural system for
actions, represented by a series of load cases applied to the structural model. An example of
application of the loads is presented also in Figure 1.

Figure 1: Structural geometry of composite truss sample and acting loads

The analysis was performed with the characteristic values of material properties and obvious
assumptions on elastic-plastic stress-strain diagrams of steel (S355) and concrete (C25/30),
commonly used in practice. The non-linear behaviour of the shear connection was modelled using
beam elements uniformly distributed with a regular spacing equal to 100 mm along the span and
4
located between the neutral axis of the top chord and the concrete slab as in Bujnak´ thesis . The
analytical expression of the evolution of the load – slip curve is given by equation:
0.4
Pi = Pmax (1- e- 0.709si ) (1)

5
with the Ollgaard´ formula of for stud shear capacity

Pmax = 0.336Ad (fck Ecm ) (2)

First of all the influence of the connectors size considering numerous theoretical values of
shank diameter varying from 0.1 to 100 mm on truss beam stiffness was analysed. These values
represent the progression of the degree of shear stud connection in the truss from no connection
to full interaction. It was recognized that the usual diameter of 19 mm is quite sufficient to obtain
a full connection. Moreover, the composite effect obtained by the shear connector diameter
variation can increase even twice the stiffness of the truss with no connection in comparison to
the composite truss beam with full connection.
The results of the next investigation focused on the effect of the top chord section on
distribution of the shear forces in the frequently used 19 mm diameter connectors along the
beams are shown in Figure 2a for different top chord sections. This phenomenon is influenced by
the ratio of geometry and resistance between the connector and the top chord section. Thus, it is

2
 Jan Bujnak and Petra Bujnakova

necessary to optimize this ratio. Otherwise, the connectors in the panel area would transfer the
predominant portion of shear forces in comparison to the obvious zones on the chord between
the nodes.

Figure 2: Effect of the top chord section and concrete strength on the shear force distributions
The influence of the material characteristics of concrete and structural steel on the distribution
of shear forces in the connectors was analysed in the additional parametrical study. The concrete
strength is an input value in one of the formulae used to calculate the shear resistance of headed
studs. Therefore, the greater value of concrete strength can provide a better shear force transfer
in the connection. However the concrete strength does not affect significantly the shape of shear
force distribution on connectors as shown in Fig. 2b. Impact of steel strength of truss material on
the shear force distribution in the connectors is small and can be neglected.

3 EXPERIMENTAL INVESTIGATION

3.1 Push-out testing

The shear capacity and the load-slip relations are the most important characteristics for the
design of the composite structure. The standard push-out tests may be obviously used for finding
approximately this relation. To investigate behaviour of connection with different configuration
and spacing of the headed stud connectors, five sets labelled as SP1, SP2, SP3, SP4 and SP5,
each comprising three push-out specimens, shown in Figure 3, were prepared. Thus fifteen
standard tests have been carried out.

Figure 3: Push-out test specimen configuration

They consisted of a steel beam HEB 260, two concrete slabs attached to the flanges of the steel
beam and stud connectors of steel S235J2 with a shank diameter of 10 mm and 50 mm in height.
Slabs of concrete class C25/30 were 620 mm long and 600 mm wide. On ly in the case of the first
specimen SP1, the slab thickness was 150 mm. The other specimens had slabs just 100 mm
thick. In the specimens SP1 and SP2, the steel and concrete elements were attached by four

3
 Jan Bujnak and Petra Bujnakova

stud connectors at each flange using automatic welding procedure. Six studs were welded at
each flange of steel members SP3, SP4 and SP5. The longitudinal distances between the
connectors differed, too. The spacing parallel to the loading direction was 250 mm in SP1 and
SP2, 60 mm in SP3 and SP4, and 40 mm in SP5. Transversal spacing was160 mm, with
exception of SP5 reduced to 32 mm.
1
According to standard procedure of Eurocode 4 ,, the load was applied in increments of 20 kN
from 0 to 100 kN (40% of the expected failure load), then returned to 12 kN ( 5% of the expected
failure load). Later loading was repeated 25 times between 12 kN and 200 kN. At each load
increment, readings of the slip between the steel beam and the concrete were recorded. During
this cycling loading the specimens remained in good condition, the slabs and the steel beam
worked well together, the cracks had not yet been developed at that time. At the end of the 25th
cycle, loading was changed from load control to slip control. The slip controlled load continued up
to the failure at the speed of 1mm/2 min. At the load of 300 kN, the interface between steel and
concrete was delaminated and the slip of 1 mm was achieved. The cracks began to appear in the
slabs. The test ended with the shear of the connectors approximately at this load lev el.
6
For finite element modelling, nonlinear analysis software ATENA was selected. It is capable
to simulate the real behaviour of concrete structures including concrete cracking, crushing and
reinforcement yielding. The three-dimensional linear four-node tetrahedral isoperimetric elements
were used to model the entire specimen. Because of symmetry, only a quarter of the push -out
configuration was modelled. Figure 4 shows the mesh used to represent a quarter of the push-out
test specimen. The element size was 0.05 m for the elements of concrete slab and steel beam
and 0.005 m for the shear connectors head and shank. The circular shapes of the stud
connectors head and shank were approximated by a hexagon.

Figure 4: Push-out test and finite element mesh of the model

Figure 5 shows a comparison between the load-slip curves obtained experimentally and
numerically using the finite element method. The load per stud was recorded from fifteen
specimen tests, starting from SP1 to SP5, illustrated by corresponding group of li nes I-1 to V-3,
respectively. The curve obtained from numerical modelling is called FEM in Fig. 5. Theoretical
and experimental values of forces as well as slip confirm a good agreement between the test and
the numerical model in the range of practically applied loading. The supplementary comparison of
results can be judged also as rather good, considering the extremely complex behaviour of
material grouping and welding procedure. The current move towards generating design data by
using numerical models in place of experiments calls for careful attention to random variables. By
considering experiments, random variables are usually represented implicitly, leading to a scatter
in the results. Therefore, there is a need to further review the use of numerical mo del for

4
 Jan Bujnak and Petra Bujnakova

producing reliable design data. Since local connection in truss beams is sensitive to the choice of
the stud location and properties as well as geometric characteristics in these composite structural
elements, it would be particularly important to ensure that the variation of input parameters is
accounted for.

Figure 5: Comparison of test and numerical results

3.2 Composite truss beam tests

To analyse the real behaviour of steel-concrete composite trusses, experimental program is
prepared. Four same steel-concrete composite truss beams of span 3.75 m were prepared.
Shear connection is provided by headed stud shear of diameter 10 mm and height 50 mm located
above the nodes as shown in Figure 6. For this configuration, the hypothesis, that the long itudinal
forces are introduced into the concrete slab only locally in the nodes, was followed. Steel truss
components were made from the steel S235. Upper chord of the beam was made from ½ IPE
160, bottom chord from two welded UPE 120 in box component, outside web members from
square hollow section SHS 70x70x6.3 and the inner diagonals from the square hollow section
SHS 40x40x3. Concrete slab of size 800x100 mm was made with demand on concrete grade
C25/30. Transversal and longitudinal reinforcement was formed from the bars ϕ10. Loading was
applied in the thirds of span above internal nodes as shown in Fig. 7.

Figure 6: Components of composite truss specimen

The experimental measurement on the first truss specimen is illustrated in Figure 7. Strains
are recorded in both chords and web members of the girders as well as concrete slab by system
of sixteen strain gauges. The details of observed locations in cross sections are designated as T1
to T16. During the testing the end slips of concrete slab have been measured using displacement
sensors. The deflection transducers P22 and P24 are placed on the girders ends, in the middle
P23 and near the quarter part of a simple span. Data received from the strain gauge package are

5
 Jan Bujnak and Petra Bujnakova

digitized and sent to the notebook. This computer is used to communicate with the measurement
system for commands regarding data acquisition, calibration, initialization, downloading and
display.

Figure 7: Loading and measurement system

The experimental deflections initially were increasing proportionally to the loading level of 325
kN. This value of an elastic limit load carrying capacity is in correlation with the result of the
1
analyses according to Eurocode 4 . The overall truss deflections under supplementary loading
were growing slightly nonlinearly and develop an ultimate permanent downwards deflection 33
mm at the end of test. Under small load, the stress distribution in strut sections is rather uniform
and progress proportionally. However, with increasing load, the diagram presents new faster
stress development. Especially the upper chord yields rapidly in the mid -span sections, where it
is subjected to bending and compression, because of the beam deflections. Finally the chord
failed by local instability. The comparison of stresses must be judged also as good if the
extremely complex character of composite truss is considered. The experimental load - carrying
capacity of the specimen was 530 kN. Comparison between the numerical and experimental
values shows a good agreement. The slips in shear connection were small in the beginning along
the composite beam span. From a symmetrical distribution of slips between the two beam´ ends
observed at the beginning of test, the slip diagram finally became more unsymmetr ical in
irreversible portion.

Figure 8: Load-Stress in web members and load-slip development.

The extreme final slip at the truss end was 8,1 mm and at the others edge only 5,9 mm. The limit
state can be defined by the load under which the connectors in the support zone failed by
shearing. Thus the bending failure of connectors in span does not need to be taken in
consideration, due to favourable load transmission.

6
 Jan Bujnak and Petra Bujnakova

Figure 9: Shear failure of connectors at the edge zone and in less stressed interior span area

4 INVESTIGATION OF ADVANCE DAMAGE PROPAGATION IN TRUSS SLAB

4.1 Vibration analysis system

Consecutive investigation in several stages may permit to recognize advanced propagation of
damages in structural elements as a function of rising loading. Figure 10 represents increasing
and dropping loading steps. They started from the theoretically initial untouched first state without
loading and going to the final stage of beam resistance exhausting. As it can be seen, time
dependent loading grow was slightly nonlinear from the third step, because loading started to be
introduced by imposed deflection to avoid sudden truss failure. Potential technique for
identification of structural component damages can be based on dynamic properties modification
7
according to principles in reference . In every unloaded stage, the truss free vibrations have been
excited using a steel hammer two kilograms weight. Generated vertical accelerations have been
recorded by four sensors, situated according to Figure 10. A knock was applied at the upper
concrete deck side and in the vicinity of accelerometers. Each of measurements was reiterated
three times.

Figure 10: Subsequent stages of measurements, acceleration sensor locations

Accelerograms processing by Fourier´s transformation can provide proper frequencies of free
8
vibrations and eliminate extreme amplitudes of waves from recorded signals as in paper .
Supposing that low damping of trusses can influence only marginally frequencies of ideal free
vibrations.

Figure 11: Exemplary accelerograms of the free vibrations of truss

7
 Jan Bujnak and Petra Bujnakova

Forms of natural vibration and corresponding frequency values were determined . The dynamic
analysis was executed using Autodesk finite numerical model. As example, the first five bending
shapes of the structure are shown in Figure 12. The corresponding numerical values are listed in
Table 1. Primarily theoretical frequencies in this table correspond to the infinitely stiff connection
at the steel concrete interface. Then, the more realistic values relate to the experimentally
determined actual rigidity, specified by a factor k = 54 MN/m are given in the third column. Also, it
can be concluded that frequencies are decreasing progressively due to gradual damage
propagation in the truss concrete deck.

Figure 12: Forms of free flexural vibrations in the vertical direction

At the end of loading, this frequency depletion varies from 8 to 20 %. Evidentially, the decrease
ratio to the initial values for higher free vibrations is greater, probably due to damping and
boundary effects.

FEM Experiment
Form Virgin state Virgin 10% 20% 40% 60% 80%
kv ® Ą k v = 54MN / m state [55kN] [100kN] [200kN] [400kN] [500kN]
1 53,4 39,31 39,0 37,56 37,46 36,12 36,12 36,12
2 79,4 71,02 72,0 66,45 65,0 62,12 62,12 62,0
3 108,2 114,77 111,2 108,4 107,0 102,6 102,6 101,1
4 148,2 152,06 158,9 153,1 144,5 143,0 141,6 132,0
5 180,7 182,13 195,0 193,6 193,6 189.3 186,4 177,7
6 223,3 225,47 216,7 216,7 216,7 212,4 206,6 204,0
7 326,8 346,44 302,0 300,5 300,0 294,7 283,2 273,0
8 429,3 406,13 388,0 387,2 385,7 375,6 355,4 335,2
9 518,6 537,40 502,7 501,3 501,3 494,1 485,4 466,6
10 630,2 653,00 585,0 573,5 573,5 567,8 524,4 505,0
Table 1: Comparison of free vibration frequencies from finite numerical model and test

4.2 Ultrasonic deck cracking investigation

Propagation time and velocity of waves are reliable parameters in ultrasonic diagnostic of
structural materials, especially by means of longitudinal type of waves with highest transmission
velocity. An average value of velocity can be obtained by measuring wave time propagatio n
between sending and receiving points. Usually sender and receiver are located perpendicularly at
opposite side of a tested specimen providing a shortest and quickest running route. A micro-
cracking in the concrete can modify the original velocity c L0 using scalar damage parameter at the
9
reduced value c L as follows
cL = 1 - w cL 0 (3)

8
 Jan Bujnak and Petra Bujnakova

In every unloaded stage, similarly as during dynamic testing, velocity propagation of ultrasonic
waves with frequency of 54 kHz among selected concrete deck points was recorded as shown in
Figure 13. This measurement was repeated three times. Beside two points at the opposite deck
surfaces, the adjacent points and in addition also two postponed points were taken into
consideration. The average velocity determined from registered times are listed in the Table 2.
The results indicate the velocity decrease about 30% especially for skew wave movement. For
the shortest transversal wave paths between points on opposite sides, the average velocity
variations were less significant. The main reason is that cracks started developing and then rising
in the crosswise direction of the truss, so matching the wave paths as shown in Figure 13.

Figure 13: Localization of measuring points and visible cracks during testing

The presented procedure may identify damaged concrete zones due to stress actions. Moreover,
the obtained results allow predicting degree and cracking location in the truss deck before
appearing at its surface. In the case of tested truss, the cracks became observable only after the
third loading stage as shown in Figure 13.

The average speed of ultrasound waves propagation in various stages [m/s]

I
II stage III stage IV stage V stage VI stage
stage

to point i 4208,3 4181,9 4160,2 4160,2 4155,8 4054,7

from
to point i-1 and i+1 4044,6 4003,1 3979,6 3957,9 3820,8 3594,9
point i
to point i-2 and i+2 4581,9 4531,5 4298,0 4124,7 3670,7 3425,1

Table 2: Average wave velocity among selected points

5 CONCLUDING REMARKS

The influence of different parameters of the steel-concrete composite truss on its behaviour
was investigated by elastic and plastic analysis. The parametric studies showed that the top
chord sections have no significant effect on the flexural stiffness and the load carrying capacity of
the composite trusses, because they are usually located very near to the neutral axis of the
composite member. However, the top chord section has an important influence on values of the
shear forces in the connectors. In fact, the ratio between the characteristics of the shear
connector and the top chord section governs the distribution of shear force along the beam. It is
necessary to optimize this ratio and to develop the rules for predicting the distribution of shear
forces in the connectors for various ratios of shear connectors and top chord sections. In this way
a more efficient use of the connectors can be achieved. The influence of material characteristics
of the structure components presented the next subject of the study. It was found that concrete
strength affect the connector resistance, but has no significant effect on the redistribution of
shear forces in the connectors. Similarly, the steel strength of the truss has littl e influence on the
shear forces in connection.
The experimental program of standard push-out examinations as well as bending test of
composite truss beam were executed to investigate real behaviour of the composite truss with
connection only above the nodes. The promising finite element models were developed, using
solid elements and local damage evolution of concrete. The aim is to take account of the local

9
 Jan Bujnak and Petra Bujnakova

phenomena such as the plastic deformation between the connectors and the top chord on all the
length, including the panel points.
Dynamic and ultrasonic non-destructive investigation methods represent effective tools for
identification of inner non-evident imperfection also in the case of composite trusses. Deck
cracking could be predicted in selected concrete part from beginnin g of truss loading and started
to be observable after the third stage.

ACKNOWLEDGMENT
The paper presents results of the research activities supported partly by the Slovak Cultural
and Educational Grant Agency; grant No. 019ŽU-4/2016.

REFERENCES

[1] Eurocode 4, Design of composite steel and concrete structures. Part 2: General rules and
rules for bridges. Brussels: CEN; 2005.

[2] J.Machacek and J. Cudejko, Shear connection in steel and concrete composite trusses,
Proc.:SDSS’Rio 2010 Stability and Ductility of Steel Structures, Rio de Janeiro, Brazil, (2010).
[3] Cast3m, an Object oriented Finite Element Method framework, Clermont-Ferrand (2007).

[4] J.Bujnak, Global analysis of steel-concrete composite beams - Analytical approach and
nonlinear modelling, PhD Thesis, Blaise Pascal University, Clermont-Ferrand II, France, (2007).

[5] J.G. Ollgaard, R.G.Slutter and J.W. Fisher, Shear strength of stud connectors in lightweight
and normal-density concrete, AISC Engineering Journal, Vol. 8, n°2, p. 55-64, (1971).

[6] Advanced Tool for Engineering Nonlinear Analysis, Software for Analysis of Concrete and
Reinforced Concrete Structures, ATENA, http://www.compsolas.com/index.htm.

[7] D. Lam and E. El-Lobody, Behavior of Headed Stud Shear Connectors in Composite Beam.
Journal of Structural Engineering, ASCE, p. 96-107, (2005).

[8] S. Berczyński and T.Wróblewski, Experimental Verification of Natural Vibration Models of

Steel-concrete Composite Beams, Journal of Vibration and Control, n.16 ( 2010).

[9] A.C. Kak and M.Slaney. Principles of Computerized Tomographic Imaging, IEEE Press, New
York, (1999)

[10] J. Lemaitre, A continuous damage mechanics model for ductile fracture, Journal, Engineering
Materials and Technology, 107, 83-89, 1985

10