Sie sind auf Seite 1von 14

Construction and Building Materials 163 (2018) 798–811

Contents lists available at ScienceDirect

Construction and Building Materials


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

Deconstructable timber-concrete composite beams with panelised slabs:


Finite element analysis
Nima Khorsandnia a, Hamid Valipour b,⇑, Mark Bradford b
a
BG&E Pty Limited, Sydney, NSW 2000, Australia
b
Centre for Infrastructure Engineering and Safety (CIES), School of Civil and Environmental Engineering, UNSW Sydney, NSW 2052, Australia

h i g h l i g h t s

 Analytical formula for load-slip of deconstructable TCC connections is proposed.


 Finite element models for deconstructable TCC beams are developed and validated.
 Influence of panelised slabs on the behaviour of TCC beams is investigated.
 Behaviour of deconstructable TCC beam under service and ultimate state is evaluated.

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

Article history: This paper deals with the development of efficient and accurate numerical models for the analysis of
Received 28 April 2017 deconstructable timber-concrete composite (TCC) beams with panelised reinforced concrete (RC) slabs.
Received in revised form 6 December 2017 Three different methods, viz. analytical, 1D frame finite element (FE) and 2D continuum-based FE models,
Accepted 25 December 2017
are proposed and validated against experimentally measured data from 4-point bending tests conducted
Available online 4 January 2018
on TCC beams with precast panels. It is shown that the analytical model is the most efficient, whereas the
2D continuum-based model is the most accurate among the three different methods considered. In addi-
Keywords:
tion, a parametric study is carried out using 2D FE models to assess various aspects of the TCC system and
Continuum-based model
Deconstructable
to evaluate the influence of precast panelised RC slabs on the structural response of TCC beams. Lastly,
Finite element the challenges of the practical FE modelling of deconstructable TCC beams are briefly discussed and
Timber-concrete composite accordingly, some guidance on the FE analysis of TCC beams with panelised RC slabs is provided.
Ó 2017 Elsevier Ltd. All rights reserved.

1. Introduction strength and stiffness of timber in the perpendicular to the grain


direction, its low fire rating and the low intrinsic damping of tim-
Apart from being aesthetically pleasing, sustainable and envi- ber associated with respect to vibration are the main design chal-
ronmentally friendly, timber as a natural material can be easily lenges facing timber structures that have hindered its widespread
handled, reused and recycled. During its growth, it also seques- use in large-scale construction and in mid- to high-rise buildings.
trates atmospheric CO2. In addition, timber has a very low density The use of timber in conjunction with other construction materials
to strength ratio, embodied energy and thermal conductivity. such as steel and concrete to produce timber-steel or timber-
These features make timber (particularly innovative engineered concrete composite (TCC) structural systems can address the
wood products) an excellent candidate material for replacing steel design challenges for timber structures adequately and extend
and concrete in building structures. Accordingly, the last two dec- the use of timber in large structures.
ades have witnessed a significant growth in the use of structural Over the last three decades, several studies have been con-
timber and the number of mid-rise timber buildings has increased ducted to investigate the short- and long-term behaviour of TCC
steadily in recent years. Although timber has a good strength-to- beams and connections under service and ultimate loading condi-
density ratio in the parallel to the grain direction, the very low tions [1–8]. In most conventional TCC floors, the composite action
between the timber beam and concrete slab is provided by shear
⇑ Corresponding author at: School of Civil and Environmental Engineering, UNSW
connectors (e.g. nails, screws or notches) permanently embedded
Sydney, Kensington, NSW 2052, Australia. in the cast in-situ concrete slabs. However, the need for propping
E-mail address: H.Valipour@unsw.edu.au (H. Valipour). and formwork for the wet concrete to be cast on top of the timber

https://doi.org/10.1016/j.conbuildmat.2017.12.169
0950-0618/Ó 2017 Elsevier Ltd. All rights reserved.
N. Khorsandnia et al. / Construction and Building Materials 163 (2018) 798–811 799

beams can dramatically reduce the speed of construction in con- 2. Shear-slip models of connections
ventional ‘‘wet” TCC systems [9,10]. Furthermore, the permanent
connections between the timber beams and concrete slabs can The accuracy of non-linear 1D and 2D FE models of composite
hamper future repair and/or upgrade of the structure and recycling members depends largely on the accuracy of the load-slip model
and/or reusing of the structural components. To increase the speed adopted for the composite joints. The load-slip results of identical
of construction and to address some of the disadvantages of con- push-out TCC joints exhibit some variability. Accordingly, in the
ventional TCC systems, much attention has focused in the last dec- first step a non-linear regression was carried out to minimise the
ade on the development of prefabricated TCC systems, in which the error associated with variability of experimental load-slip curves.
concrete slab can be cast and cured beforehand and then con- A 7-parameter representation was considered and fitted to the
nected later to the timber beams by adhesives, mechanical connec- experimental load-slip results of types A to F (see Fig. 1a and
tors or a combination of both; either on- or off-site. Recent studies Table 1) deconstructable TCC joints [23] in the form
have investigated different aspects of prefabricated TCC systems,
ðK 0  K u Þ s ðK u þ K s Þ s
including the development of TCC connections and beams using P¼n h in1 on1 þ n h in2 on1  K s s;
different engineered wood products [11–13], the short- and long- 1 þ ðK 0  K u Þ Ps0
1
1 þ ðK u þ K s Þ P1 P
s 2

term behaviour of prefabricated TCC systems under different 0

ultimate and service loading conditions [14–16], applications of ð1Þ


prefabricated TCC systems for multi-storey buildings [17], the
where the input parameters are the serviceability (K0), ultimate (Ku)
use of prefabricated TCC members for refurbishment and strength-
and softening (Ks) stiffnesses, the initial (P0) and subsidiary (P1)
ening [18] and bridge decks [19]. However, only a few studies have
loading capacities and two parameters (n1 and n2) that control the
focused on the numerical modelling and finite element (FE) analy-
curvature of the function (Fig. 1b). Details of the deconstructable
sis of TCC beams with precast RC slab panels [14,19].
TCC connections (thickness of LVL beams, type and size and connec-
In addition to prefabrication, attention has been drawn more
tors) and values of the seven input parameters and the coefficient of
recently to deconstructable structures and development of com-
determination (R2) for the calibrated models are given in Table 1.
posite structural systems (e.g. steel-concrete, steel-timber,
The maximum error of the calibrated models is 13% (R2 = 0.87)
timber-concrete etc.) which can be easily assembled and disman-
and the good correlation between the experimental data and the
tled at the end of a building’s service life. These deconstructable
analytical model is evident from the load-slip curves shown in
structural systems can significantly facilitate recycling and reuse,
Fig. 2 and the R2 values (see Table 1) which are ranging between
reducing the waste of construction materials, thereby enhancing
0.87 and 0.99. It is observable that the calibrated model can ade-
the sustainability of the building industry [20–24]. The feasibility
quately represent the pre- and post-peak behaviour of the timber-
of a deconstructable TCC system comprising of precast concrete
to-precast RC slab panel joints that were tested.
slabs connected to laminated veneer lumber (LVL) beams whose
composite efficiency and structural performance is comparable
to conventional TCC beams (with permanent connections 3. Bending responses of beams
between the concrete slab and timber beams) has been demon-
strated by Khorsandnia et al. [23]. Furthermore, promising results One of the main objectives of this paper is to develop theoretical
and favourable structural performance have been observed in models that can adequately capture the global and local responses
push-out and bending tests conducted on prefabricated TCC con- of the deconstructable TCC beams. The results of Khorsandnia et al.
nections and slabs by Crocetti et al. [25]. However, experimental experiments [23] on deconstructable TCC systems are used as
data needed to develop design provisions for prefabricated decon- benchmark data for evaluating the accuracy of the theoretical
structable TCC beams and floors is very limited, and there is a models. The TCC beams comprised of precast RC slab panels and
need for efficient yet sufficiently accurate numerical models to LVL beams connected by different mechanical connectors
further investigate particular aspects of deconstructable TCC sys- (Fig. 3a). Due to construction tolerances and RC slab panel
tems, including the effect of construction tolerances and pan- misalignments, small gaps between the slab panels were visible
elised RC slabs on the local and global behaviour of this novel and so a filler was used between the RC slab panels. Accordingly,
composite system. it was hypothesised that the segmentation of slabs including the
Recently, detailed FE modelling of conventional TCC beams number of RC slab panels, the size of the gaps and the stiffness
and deconstructable steel-concrete composite beams under ser- of the filler material can affect the response of a TCC member
vice and ultimate loading conditions has been carried out [26– and it should be considered in the FE analysis and parametric study
29]. In addition to deterministic FE analysis of TCC beams and of the TCC system with precast panelised slabs.
connections, nonlinear probabilistic FE analysis of TCC beams In this study, analytical, 1D frame FE and 2D continuum-based
has been carried out to demonstrate the variability of FE predic- FE models are employed and are validated against experimental
tions and local and global structural response with respect to results in the literature [23]. The most accurate model amongst
uncertainties in mechanical properties of timber, concrete and those proposed is selected for a parametric study to elucidate var-
timber-to-concrete connections [30]. However, no comprehensive ious aspects of the structural response of TCC beams with pan-
FE modelling of the prefabricated TCC beams has been reported in elised RC slabs and deconstructable connections.
the literature.
This paper therefore focuses on the development and appli- 3.1. Analytical model
cation of three methods, i.e. an analytical model, and 1D FE
and 2D FE models with different levels of efficiency and accu- An analytical representation can be used for predicting the
racy, for undertaking analyses of deconstructable TCC beams structural behaviour of mechanically jointed composite beams,
with precast panelised RC slabs. Following validation of the the details of which are provided in Annex B of BS EN1995-1
models against the experimental results reported by Khorsand- [31]. This model was developed based on the fulfilment of compat-
nia et al. [23], a parametric study is carried out and the effects ibility and equilibrium conditions for the composite sections, in
of RC slab segmentation and deconstructable timber-to-slab conjunction with a linear elastic assumption for all components.
connections on the structural behaviour of the TCC beams are In the analytical model, the effective flexural stiffness of the
investigated. composite section can be obtained from
800 N. Khorsandnia et al. / Construction and Building Materials 163 (2018) 798–811

Connection Connection
axis axis 2 Bolts 8.8
75 mm 2×60
l = 110 mm
Concrete Concrete Concrete

40
Concrete

25
Timber 1 Bolt 8.8 Timber
25 2 Bolts 8.8
Coach Screw for A4 only Coach Screw 2UA 75×50×5
Timber l = 200 mm l = 100 mm Timber l = 90 mm
l = 200 mm
Type: A joist Type: D joist
th. th.
2 Bolts 8.8 2 Bolts 8.8
2×60 2×45
l = 150 mm l = 110 mm
Concrete Concrete
Concrete Concrete

Timber SH 50×50×5 Timber


l = 180 mm Coach Screw 16 PL150×100×10 2 Coach Screw 10
Timber Timber l = 100 mm
Type: B joist l = 130 mm Type: E joist
th. th.
2 Bolts 8.8 200 Timber
2×60
l = 110 mm 110 40 block
Concrete Concrete

60°

60°
Concrete Concrete

Timber PFC 100 Timber


2 Coach Screw 10 2SPAX Screw 10/
l = 180 mm Timber Timber
l = 100 mm 2 Coach Screw 12
Type: C joist Type: F l = 200 mm joist
th. th.
Elevation View Section View Elevation View Section View
(a)

P1

Ku
P0 1 Ks
n2 1

n1
K0
1
s
(b)
Fig. 1. Outline of (a) deconstructable timber-concrete composite (TCC) joints types A to F tested by Khorsandnia et al. [23] and (b) shear-slip curve with three asymptotes
adopted for TCC joints.

EIef ¼ Ec Ic þ c1 Ec Ac a21 þ Et It þ Et At a22 ; ð2Þ ing assuming that the connections are equally spaced along the TCC
member. The remainder of the input parameters have been defined
where in Khorsandnia et al. [32].
8 In addition, the ultimate loading capacity (Pu) and its corre-
>
> 1 Full composite response
< 1 sponding deflection (du) can be computed by the formulae pro-
c1 ¼ 2 Semi composite response ; ð3Þ
> 1þp Ec2Ac S vided in Khorsandnia et al. [32]. Using the Navier-Bernoulli
>
:
KL

0 Non composite response hypothesis in conjunction with Hooke’s law (i.e. linear elastic
material behaviour), the ultimate strain eu at the mid-span of a
TCC section can be obtained from
c1 Ec Ac ðhc þ ht Þ
a2 ¼ ; ð4Þ
2ðc1 Ec Ac þ Et At Þ ðl  aÞ
eu ¼ b Pu ; ð6Þ
4EIef
hc þ ht
a1 ¼  a2 ; ð5Þ where l is the beam span, a is the distance between the applied
2
loads in a four-point bending test set up, and b is a coefficient
the right subscripts c and t denote concrete and timber respectively, related to the location of the fibre where eu is computed and b is
K is the slip modulus of the connection and S is the connection spac- obtained from the following equation:
N. Khorsandnia et al. / Construction and Building Materials 163 (2018) 798–811 801

Table 1
Details of the deconstructable TCC connections and values of input parameters for the shear-slip curves of the deconstructable connectors.

Type LVL th. (mm) Connector type & size P0 (kN) P1 (kN) K0 kN/mm Ku kN/mm Ks kN/mm n1 n2 R2
A1 45 1 Coach screw 12 mm 50.8 127.1 31.6 0.0 8.0 1.24 2.24 0.91
A2 63 1 Coach screw 12 mm 44.0 707.0 31.4 0.0 53.5 1.23 5.99 0.87
A3 63 1 Coach screw 16 mm 35.1 55.3 60.1 4.2 2.4 2.41 1.85 0.87
A4 63 1 Coach screw 16 mm 43.5 99.1 330.0 14.2 2.6 0.35 2.35 0.92
B1 63 2 Bolts 12 mm 92.4 147.4 30.1 0.0 2.4 2.41 1.87 0.91
B2 63 2 Bolts 12 mm 109.8 147.0 26.0 0.0 2.6 2.14 5.85 0.97
B3 63 2 Bolts 16 mm 102.9 111.0 4714 4685 1.4 4.97 0.82 0.97
B4 63 2 Bolts 16 mm 130.3 140.0 1414 1386 2.8 3.50 1.46 0.99
B5 45 2 Bolts 12 mm 122.9 154.3 75.5 0.0 0.9 0.83 0.76 0.97
C1 45 2 Bolts 12 mm 11.9 112.4 124.3 13.8 1.4 2.48 4.32 0.96
C2 63 2 Bolts 12 mm 3.3 130.8 254.0 16.7 1.5 19.93 4.95 0.96
D1 45 4 Bolts 12 mm 15.9 80.0 81.0 9.9 0.8 4.10 4.25 0.89
D2 63 4 Bolts 12 mm 21.2 104.6 2964 8.2 1.0 0.70 4.85 0.98
E1 45 2 Bolts 12 mm 17.6 47.6 942.5 7.3 0.6 0.76 6.54 0.89
F1 63 2 Coach screws 12 mm 24.1 280.0 20.1 1.4 6.0 3.60 6.11 0.93
F2 45 2 SPAX screws 10 mm 61.6 113.8 250.3 11.5 2.1 0.21 1.39 0.95

70
40 60
Load (kN), P/2

Load (kN), P/2


50
30
40
20 30
A1
20
10 A2 A3
10
A4
0 0
0 2 4 6 8 10 0 2 4 6 8 10 12
Slip (mm) Slip (mm)

120 120
100 100
Load (kN), P/2

Load (kN), P/2

80 80
60 60
B3
40 40
B1
20 B2 20 B4
0 0
0 5 10 15 20 0 5 10 15 20
Slip (mm) Slip (mm)

120 100
100
80
Load (kN), P/2

Load (kN), P/2

80
60
60
40
40 C1 D1
20 C2 20 D2

0 0
0 5 10 15 20 0 5 10 15 20
Slip (mm) Slip (mm)

100 50

80 40
Load (kN), P/2

Load (kN), P/2

60 B5 30

40 20
F1
20 10
E1 F2
0 0
0 5 10 15 20 0 5 10 15 20
Slip (mm) Slip (mm)

Fig. 2. Fitted shear-slip curves for type A to F deconstructable connections tested by Khorsandnia et al. [23] (thin lines represent experimental results).
802 N. Khorsandnia et al. / Construction and Building Materials 163 (2018) 798–811

(a)

(b)

(c)
Fig. 3. Outline of (a) deconstructable TCC beam set-up, (b) 1-D frame FE model (q and Q are generalised nodal displacements and forces) and (c) 2-D continuum-based model
developed in ABAQUS.

8
> c1 a1  hc =2 Concrete top fibre fibres and 17 composite Simpson’s integration points were used
>
>
< c1 a1 þ hc =2 Concrete bottom fibre over the timber beam and concrete slab cross sections, respec-
b¼ : ð7Þ tively. The adequacy of adopted element formulation, integration
>
> a2  ht =2 Timber top fibre
>
: scheme and number of integration points for capturing the global
a2 þ ht =2 Timber bottom fibre
and local response of the TCC beams has been demonstrated in
The ultimate end slip su at the supports can be calculated by inte- previous studies [6,33]. In the 1D frame FE models, the gap
grating the slip strain (i.e. the difference between the strain at the between the RC slab panels is not considered and a further com-
bottom fibre of the concrete and the top fibre of the timber beam) parative study using 2D FE models is carried out to highlight the
at the interface over the entire beam length as effect of the gap on the structural performance of the TCC systems.
In the 1D frame FE model, Glos [34] and the CEB-FIP model code
2
ð1  c1 Þðhc þ ht  2a2 Þðl  a2 Þ [35] are adopted for the timber and concrete respectively (see
su ¼ Pu : ð8Þ Fig. 4). Uniaxial compression tests were conducted on concrete
32EIef
cylinders [23] and the relationships
8
3.2. 1D frame FE model <f0 2ðe=ec0 Þðe=ec0 Þ 2
Pre  peak response
c 1þðk2Þ ðe=ec0 Þ
r¼ ; ð9Þ
: f 0 eu e Post  peak response
The 1D frame FE model developed by Khorsandnia et al. [33] is c eu ec0

employed for modelling the TCC beams (see Fig. 3b). Each element
(Fig. 4b) were adopted for the pre- and post-peak behaviour of the
in this comprises of two layers representing the timber beam and
compressive concrete in the RC slabs of the TCC beams.
concrete slab in a TCC system, which are connected by horizontal
The non-linear empirical shear-slip models obtained from Eq.
discrete end springs to capture the shear-slip interaction between
(1) and given in Fig. 2 for different types of deconstructable TCC
the layers, whilst the layers are constrained in the vertical direc-
connections are assigned to the end springs (Fig. 3b) representing
tion assuming that no separation occurs between them [33]. The
the shear connectors in the 1D frame FE model.
Navier-Bernoulli hypothesis in conjunction with force interpola-
tion (flexibility-based approach) is adopted in the formulation of
the 1D frame elements. One half of the TCC beam was modelled 3.3. 2D continuum-based model
by 18 elements (maximum element length was limited to 167
mm) with 17 composite Simpson’s integration points along each A versatile 2D (plane stress) FE model was also developed with
element. The section flexibility (stiffness) is calculated by a numer- the focus being the effect of the discontinuity and segmentation of
ical integration scheme rather than discretising the section to the RC slab panels on the structural behaviour and performance of
N. Khorsandnia et al. / Construction and Building Materials 163 (2018) 798–811 803

effect on the tensile failure in parallel to grain directions and can


be taken as a = 0.
The effective stress r^ 31 ¼ ½ r
^ 11 r
^ 22 r
^ 12  T and Cauchy stress
r31 ¼ ½ r11 r22 r12  T vectors are related through diagonal dam-
age matrix x33 ¼ diag ½11xi  ; i ¼ 1; 2; 3 as,

r^ ¼ x r: ð11Þ
Glos’s [34] model The adequacy of damage-plasticity model and Hashin’s damage
model for capturing the failure modes and the non-linear beha-
viour of concrete and timber under multiaxial stress states have
been already verified by Khorsandnia et al. [32].
In the 2D continuum-based FE models, only one half of each TCC
beam is modelled by 4-node CPS4R plane stress quadrilateral ele-
ments with reduced integration and hourglass control. The maxi-
(a) mum element size for both timber beam and concrete slab is 20
Batch 3 (test) Batch 4 (test)
mm. In the absence of a pressure-over closure (soft contact) law
Batch 3 (model) Batch 4 (model) to be used for modelling the gaps and due to numerical instability
45 of soft contact models, the contact between the RC slab panels and
40 f'c= 43.5 MPa the filler material is modelled using hard contact with friction in
35 εc0= 0.003
tangential direction. The stiffness of contact in the direction nor-
Stress (MPa)

30
Post-peak mal to the interface between timber and concrete and the interface
25
20
f' c= 39 MPa response between concrete and filler was taken knn = 1000 N/mm3 and 500–
εc 0= 0.0025 5000 N/mm3, respectively. Moreover, the coefficient of friction in
15
10 Pre-peak
the tangential direction was taken l = 0.2.
5 response For the sake of simplicity, the filler between the RC slab panels
0 is treated as a linear elastic material. The advantage of 2D
0 0.002 0.004 0.006 0.008 0.01 continuum-based FE models over 1D frame models is their capabil-
Strain (mm/mm) ity to take the effect of the gaps and contacts between the concrete
(b) slab segments into account. In the 2D models, hard and frictionless
contacts are used in the normal and tangential directions respec-
Fig. 4. Stress-strain relationship of (a) timber in tension and compression and (b)
tively for connecting timber to concrete in vertical direction. In
concrete in compression adopted for 1D FE analysis.
the horizontal direction, non-linear springs at the interface of the
timber beam and concrete slab panels are utilised to model the
connectors and to capture the horizontal shear interaction and
TCC beams with panelised RC slabs. The 2D models are developed composite action between the timber beam and concrete slab.
in ABAQUS [36] environment (Fig. 3c) and Hashin’s damage criteria The force-displacement relationship for the horizontal springs is
and concrete isotropic damage plasticity models were employed defined with respect to the type of the shear connector and its cor-
for capturing the onset of failure in timber beam and concrete slab, responding empirical load-slip behaviour provided in Fig. 2. In 2D
respectively. Furthermore, the damage evolution laws for timber FE models, the RC slab panels are modelled with separate elements
are based on the energy dissipation during the damage process (see Fig. 3c) and the interface between two adjacent panels is trea-
and linear material softening. The concrete slab is modelled using ted either as an open gap or as a no gap zone filled with plaster or
concrete damage-plasticity constitutive law available in the ABA- filler material [23]. The contact between the RC slab panels and the
QUS library of materials and steel reinforcements are treated as filler material is modelled using hard contact with friction as
deformable elastic (Es = 200 GPa) bars embedded in the concrete described previously. The details of TCC beams and input parame-
slab. ters (including size of gaps) adopted in the 2D FE models are given
The 2D Hashin damage model comprises of four different dam- in Table 2. The size of gaps between prefabricated RC panels (see
age initiation criteria expressed in terms of effective stress compo- Table 2) was adopted in the FE models with respect to average
nents r ^ ij (i, j = 1, 2), gap size observed in each specimen during the laboratory experi-
8  2  2 ments [23]. Except for specimens B-A5, no significant change in
>
> r^ 11
þ a r^ 12
: Tension=shear ll to grain
>
> T
SL the structural response of the TCC beams tested by Khorsandnia
>
>
X
>
>   2 et al. [23] at the ultimate limit state (ULS) loading condition was
>
> r^
< X22C : Compression ll to grain
observed after dismantling and reassembling the specimens. The
 2  2 ;
>
> r^ 22 ^
þ rS12L
reduction in the structural performance of beam B-A5 was attribu-
>
> : Tension=shear L to grain
>
T
Y ted to looseness of the screw connectors following dismantling and
>
>  2   2    
>
> 2
>
: r^ 22T þ Y T  1 r^ 22C þ r^ 12L : Compression=shear L to grain
C reassembling of the specimen. To take account of this reduction,
2S 2S Y S
the stiffness and the shear-slip behaviour of the connectors was
ð10Þ lowered by introducing a reduction factor (Table 2) under service
T T
limit state (SLS) and ULS loading conditions.
where X and Y denote the tensile strength of timber in parallel The material properties including mean (from three identical
and perpendicular to the grain direction, respectively; X C and Y C specimens) compressive strength f c ; tensile strength f t ; bending
denote the compressive strength of timber in parallel and perpen- strength f b and elastic modulus E of concrete and LVL timber (in
dicular to the grain direction, respectively; SL and ST are the shear parallel to grain direction) adopted in the 2-D FE models are given
strength in parallel and perpendicular to the grain direction, respec- in Table 3. For LVL modulus of rigidity G = 660 MPa, shear strength
tively; and a is a coefficient that takes account of the shear stress fs = 4.6 MPa and compressive strength in the perpendicular to grain
804 N. Khorsandnia et al. / Construction and Building Materials 163 (2018) 798–811

Table 2
Details of TCC beams tested and input parameters in 2-D FE model.

Beam Timber Concrete Connection Test Contact between Input Parameters in 2D FE Model
Section Section No  Type Method Segments
Gap (mm) Contact Stiffness Connector
(N/mm) Reduction
Factor
SLS ULS
B-A1 300 63 600 75 14 A3 ULS Small Gap 0.1 750 – 1.0
B-A2 300 63 600 75 20 A3 ULS Visible Gap 1.7 1000 – 1.0
B-A3 300 63 600 75 20 A3 ULS No Gap + Filler 0.0 Infinity – 1.0
B-A4 400 600 75 30 A3 ULS Small Gap + Filler 0.1 6000 – 1.0
B-A5 300 63 600 75 30 A3 SLS & ULS Small Gap + Filler 0.1 2500 1.0 0.6
B-B1 300 45 600 75 16 B5 SLS & ULS Small Gap + Filler 0.1 Infinity 1.0 1.0
B-D1 300 45 600 75 16 D1 SLS & ULS Small Gap + Filler 0.1 Infinity 1.0 1.0
B-F1 300 63 600 75 16 F1 SLS & ULS Small Gap + Filler 0.1 Infinity 1.0 1.0
B-F2 300 45 600 75 16 F2 ULS Small Gap + Filler 0.1 400 – 1.0

Table 3
Mean material properties for concrete slabs and LVL timber beams (in parallel to grain direction) adopted in the FE models.

Beam Concrete LVL Timber (Mean [CoV])


fc (MPa) ft (MPa) Ec (GPa) Bending Compression Tension
E (GPa) fb (MPa) fc (MPa) ft (MPa)
B-A1 41.6 3.9 32.3 15.3 [5%] 77.0 [13%] 58.4 [4%] 44.4 [2%]
B-A2 41.6 3.9 32.3 15.3 [5%] 77.0 [13%] 58.4 [4%] 44.4 [2%]
B-A3 41.6 3.9 32.3 13.3 [4%] 68.3 [5%] 52.4 [6%] 39.5 [8%]
B-A4 41.6 3.9 32.3 12.1 [1%] 56.5 [4%] 45.8 [4%] 32.0 [1%]
B-A5 41.6 3.9 32.3 13.3 [4%] 68.3 [5%] 52.4 [6%] 39.5 [8%]
B-B1 45.9 4.1 33.2 14.5 [6%] 72.1 [11%] 54.5 [4%] 40.2 [1%]
B-D1 45.9 4.1 33.2 14.5 [6%] 72.1 [11%] 54.5 [4%] 40.2 [1%]
B-F1 45.9 4.1 33.2 13.3 [4%] 68.3 [ 5%] 52.4 [6%] 39.5 [8%]
B-F2 45.9 4.1 33.2 13.6 [2%] 62.7 [ 9%] 51.9 [6%] 39.9 [1%]

direction fp = 12 MPa were adopted, respectively, according to the response is significantly influenced by the segmentation of the RC
manufacturer specifications. slabs (e.g. large gaps between the RC slab panels). It is also observ-
It is noteworthy that the elastic modulus of LVL timber obtained able that the 2D FE model can represent the global response
from bending tests (see Table 3) was used in the FE models to take (deflection and slip) as well as the local behaviour (timber and con-
account of the shear deformation effect in the 1D FE simulations. crete strains) of the TCC beams with an accuracy better than the
The CoV of compressive strength for different concrete batches var- simple analytical and 1D frame FE models; but the 2D FE models
ied between 4.6% and 6.3%. Furthermore, the CoV of timber are computationally more expensive than the other two models.
mechanical properties are provided in Table 3 [23]. Since the simple analytical and 1D frame FE formulations can pro-
duce erroneous results for TCC beams with segmented slabs (par-
3.4. Verification of results ticularly when there are gaps between the slab panels), more
advanced models such as 2D continuum-based FE models are
All nine TCC beams in the experiments [23] are analysed, and required for non-linear analysis of the beams under SLS and ULS
the local and global responses captured by the simple analytical loading conditions. However, when the effect of segmentation is
model, 1D frame FE and 2D FE model are validated against these negligible (i.e. full contact or no gap between the slab panels),
results. The deflection, slip, concrete and timber strain results for the simple analytical and 1D frame FE techniques can adequately
the beams B-A1, B-A2, B-A3 and B-A5 are shown in Figs. 5–8. These capture the behaviour of TCC beams and employing sophisticated
four beams have similar composite cross section and connection 2D FE models only increases the cost of the computations.
type over the beam length, but the number of connections and type The ultimate loading capacity (Pu) and its corresponding mid
of contact between the juxtaposed RC slab panels (Table 2) are dif- span deflection (du), the serviceability stiffness (Ks) at 40% of the
ferent. In Figs. 5–8, the structural responses assuming full and no ultimate load, the ultimate end slip (su) and the strain at the top
composite action are also displayed to demonstrate the composite and bottom fibres of the mid-span cross-section for the slab and
efficiency of the TCC beams with precast panelised slabs at differ- timber beam obtained from the laboratory tests and predicted by
ent stages of loading. It is observable that the simple analytical different models are provided in Tables 4 and 5. The comparison
model can adequately predict structural response of the TCC beams between the experimentally measured and numerically predicted
within the service loading condition and/or when the TCC beams values of Pu, du, Ks, su and the values of the timber and concrete
have high levels of composite efficiency. However, at the ULS con- strain demonstrate the accuracy of the 1D and 2D FE models com-
dition the simple analytical method overestimates the stiffness and pared to the simple analytical technique for predicting the global
strength of the TCC beams, being attributed to the linear elastic and local responses of the deconstructable TCC beams, particularly
behaviour adopted for all the TCC components (i.e. timber, con- when the segmentation of RC slabs does not dominate the struc-
crete and shear connectors). The 1D FE models can predict the tural behaviour. To visualise and compare the accuracy and perfor-
non-linear behaviour of the TCC beams particularly near the col- mance of the strategies adopted, the evaluation of errors with
lapse loads, but the 1D FE models tend to overestimate the stiffness respect to experimental data at global (Pu, du, Ks and su) and local
and strength of the TCC beams (Fig. 5a and b) when the structural (ultimate strains at mid span) levels is carried out and the results
N. Khorsandnia et al. / Construction and Building Materials 163 (2018) 798–811 805

120 120

100 100

Total Load (kN)

Total Load (kN)


80 80

60 60

40 40
Test (ULS) Test (ULS)
Analytical Analytical
20 FE (1D) 20 FE (1D)
FE (2D) FE (2D)
0 0
0 20 40 60 80 100 0 20 40 60 80 100
Mid span deflection (mm) Mid span deflection (mm)
(a) (b)
120
120
100
100
Total Load (kN)

Total Load (kN)


80
80
60
60
40 Test (SLS)
Test (ULS) 40 Test (ULS)
Analytical Analytical
20 FE (1D) 20 FE (1D)
FE (2D) FE (2D)
0 0
0 20 40 60 80 100 0 20 40 60 80 100
Mid span deflection (mm) Mid span deflection (mm)
(c) (d)
Fig. 5. Load-deflection experimental and theoretical results of TCC beams tested for (a) B-A1, (b) B-A2, (c) B-A3 and (d) B-A5.

120 120
Full composite

Full composite

100 100
Total Load (kN)

Total Load (kN)

80 80

60 60

40 Test (ULS) 40 Test (ULS)


Analytical Analytical
20 FE (1D) 20 FE (1D)
FE (2D) FE (2D)
0 0
0 2 4 6 8 0 2 4 6 8
End slip (mm) End slip (mm)
(a) (b)
120
Full composite

Full composite

120
100
100
Total Load (kN)

Total Load (kN)

Test (SLS)
80 Test (ULS)
80
Analytical
60 FE (1D)
60
FE (2D)
40 Test (ULS) 40
Analytical
20 FE (1D) 20
FE (2D)
0 0
0 2 4 6 8 0 2 4 6
End slip (mm) End slip (mm)
(c) (d)
Fig. 6. Load-slip experimental and theoretical results of TCC beams tested for (a) B-A1, (b) B-A2, (c) B-A3 and (d) B-A5.b) B-A2, (c) B-A3 and (d) B-A5.
806 N. Khorsandnia et al. / Construction and Building Materials 163 (2018) 798–811

120 120
Top fibre Bottom fibre Top fibre Bottom fibre

Full composite

Full composite
100 100

Total Load (kN)

Total Load (kN)


80 80

60 60

40 40
Test (ULS) Test (ULS)
Analytical Analytical
20 20
FE (1D) FE (1D)
FE (2D) FE (2D)
0 0
-1000 -500 0 500 1000 -1000 -500 0 500 1000
Strain (µε) Strain (µε)
(a) (b)
120
Top fibre Bottom fibre Top fibre Bottom fibre
120
100
100 Test (ULS)
Total Load (kN)

Total Load (kN)


80 Analytical
80 FE (1D)
60 FE (2D)
60
40 40
Test (ULS)
Analytical
20 20
FE (1D)
FE (2D)
0 0
-1000 -500 0 500 1000 -1000 -500 0 500 1000
Strain (µε) Strain (µε)
(c) (d)
Fig. 7. Load-concrete strain experimental and theoretical results of TCC beams tested for beams (a) B-A1, (b) B-A2, (c) B-A3 and (d) B-A5.

120 120
Top fibre Bottom fibre Top fibre Bottom fibre
Full composite

Full composite

100 100
Total Load (kN)

Total Load (kN)

80 80

60 60

40 40
Test (ULS) Test (ULS)
Analytical Analytical
20 20
FE (1D) FE (1D)
FE (2D) FE (2D)
0 0
-5000 -2500 0 2500 5000 -5000 -2500 0 2500 5000
Strain (µε) Strain (µε)
(a) (b)
120
Full composite

Full composite

Top fibre Bottom fibre Top fibre Bottom fibre


120
100
100
Total Load (kN)

Total Load (kN)

80
80
60
60
40 40
Test (ULS) Test (ULS)
Analytical Analytical
20 20
FE (1D) FE (1D)
FE (2D) FE (2D)
0 0
-5000 -2500 0 2500 5000 -5000 -2500 0 2500 5000
Strain (µε) Strain (µε)
(c) (d)
Fig. 8. Load-timber strain experimental and theoretical results of TCC beams tested for beams (a) B-A1, (b) B-A2, (c) B-A3 and (d) B-A5.
N. Khorsandnia et al. / Construction and Building Materials 163 (2018) 798–811 807

Table 4
Comparison of experimental and analytical results for global responses (Pu, du, Ks and su) of all TCC beams tested.

Beam Pu (kN) du (mm) Ks (kN/mm) su (mm)


Test Anal FE (1-D) FE (2-D) Test Anal FE (1-D) FE (2-D) Test Anal FE (1-D) FE (2-D) Test Anal FE (1-D) FE (2-D)
B-A1 107 130 109 106 96 59 91 79 1.7 2.2 2.3 1.7 6.8 1.8 7.7 3.2
B-A2 94 132 120 104 65 55 85 70 1.4 2.4 2.5 1.4 1.6 1.4 6.4 1.3
B-A3 113 119 111 115 83 55 79 84 2.2 2.2 2.3 2.1 5.7 1.3 5.4 6.0
B-A4 211 187 204 214 56 39 51 57 4.6 4.8 5.1 4.5 1.8 1.0 2.8 2.6
B-A5 121 121 136 122 86 51 63 95 1.9 2.3 2.5 2.0 5.6 0.9 1.6 5.7
B-B1 95 90 108 94 61 54 63 57 1.9 1.7 1.9 1.7 2.4 1.6 1.7 1.3
B-D1 99 88 103 98 70 60 75 72 1.8 1.5 1.9 1.7 2.5 2.5 3.4 3.1
B-F1 93 109 105 104 86 78 98 94 1.6 1.4 1.7 1.6 6.5 4.4 7.7 7.0
B-F2 73 80 80 74 81 67 75 81 0.9 1.2 1.4 1.0 5.0 3.3 4.7 2.4

Table 5
Comparison of experimental and analytical results for local responses (ultimate strain values at mid span) of all TCC beams tested.

Beam Concrete top fibre Concrete bottom fibre Timber top fibre Timber bottom fibre
Test Anal FE (1-D) FE (2-D) Test Anal FE (1-D) FE (2-D) Test Anal FE (1-D) FE (2-D) Test Anal FE (1-D) FE (2-D)
B-A1 300 956 963 711 89 280 724 378 2857 995 2439 1698 4505 3948 4309 3650
B-A2 860 937 984 658 47 220 422 179 762 746 1541 968 3236 3882 4084 3214
B-A3 760 899 938 776 103 244 362 251 664 639 1144 1099 3099 3934 4057 4021
B-A4 968 813 947 827 345 1 11 59 536 745 1009 1023 3871 3583 4099 4145
B-A5 691 881 1034 853 184 196 168 427 1047 430 369 1783 4669 3878 4439 4570
B-B1 611 784 929 667 99 338 314 137 120 724 476 280 3998 3766 4495 3703
B-D1 643 832 955 738 175 432 467 291 22 1164 1076 907 4685 3891 4614 4176
B-F1 566 1021 994 796 151 615 913 604 1760 2195 2788 2470 3323 4350 4837 4393
B-F2 493 856 836 642 137 537 624 511 1655 1624 1636 1803 3810 3947 4202 3864

are illustrated in Fig. 9. It is observable that 2D FE model has the of this specimen and its experimental results were quite similar
lowest error (the most accurate) with minimum coefficient of vari- to a TCC beam with a single RC slab (i.e. no segmentation in the
ation for most cases considered, whereas the simple analytical and slab).
1D frame FE models have comparable accuracy for predicting the
behaviour of deconstructable TCC beams. 3.5. Gap between RC slab panels
The mean of difference between the experimental and numeri-
cally predicted peak load Pu is 8.5% and 2.8% for 1D and 2D FE mod- The effect of the gap size on the global response of the TCC
els, respectively. For the mid-span deflection du corresponding to beams was evaluated by assuming three different values (1, 2
ultimate load, the mean of difference between experimental and and 4 mm) for the gap between the RC slab panels. The load-
1D and 2D FE results are 15.0% and 9.5%, respectively and the mean deflection and load-slip response of specimen B-A3 with these
of difference between experimental and 1D and 2D FE results for adopted gap sizes are shown in Fig. 10a and b, respectively. In
service stiffness Ks is 14.0% and 4.8%, respectively. This is demon- these analyses, a hard contact between the RC slab panels (and fil-
strative of the better accuracy of 2D FE models compared to 1D ler) was considered. It is observable that the size of the gaps
FE models. Furthermore, it is observed that the variability in the between the panels has negligible influence on the ultimate load-
TCC beam peak load capacities predicted by 2D FE models is smal- ing capacity of the beams, but the load-deflection response and
ler than the variability observed in the bending strength fb of tim- stiffness of the beams under service loading conditions is reduced
ber (average CoVs = 8.5%). However, the variability in the significantly by increasing the gap size between slab panels. The
displacement du and stiffness Ks is significantly bigger than the hardening behaviour (i.e. increasing stiffness) at the initial stages
variability of LVL timber elastic modulus (average CoVs = 4.1%). of the load-deflection evolution (see Fig. 10a) can be attributed
The relatively large variability in the numerically predicted dis- to the closing gaps with the increasing curvature that brings the
placement du and stiffness Ks can be attributed to large variability furthest top fibre of the adjacent slabs into direct contact. As is evi-
(uncertainty) in the connectors’ behaviour (stiffness) as evident dent from Fig. 10b, the slip that occurs at end supports of the TCC
from R2 values provided in Table 1. These observations are consis- beams with large gaps between the RC slab panels is very small.
tent with the results of probabilistic FE analyses carried out by The non-linear analyses of the TCC beams show that the structural
Zona et al. [30]. response (load-deflection) of the TCC beams with precast panelised
Since the values of the input parameters with respect to the slabs tends to a lower bound as the size of gaps between the RC
segmentation of the slabs (i.e. number of slab panels and the gap slab panels increases (Fig. 10a). This lower bound (load-
between the panels) along the TCC beams have not been reported deflection) is representative of the response of a TCC beam with
in the experimental program and these input parameters cannot be no contact between the RC slab panels (Fig. 10a).
measured easily during a test, a parametric study is carried out to
understand how four different characteristics of the prefabricated 3.6. Modulus of elasticity (MOE) of filler
deconstructable TCC beams, the number of slab panels, the gap
between the panels and the stiffness of materials used for filling Due to significant effect of the gaps on the structural behaviour
the gaps in conjunction with stiffness of shear connectors, can of TCC beams under SLS loading conditions, Khorsandnia et al. [23]
affect the structural response of the TCC beams. All the analyses recommended filling of the gaps with plaster or a filler materials
in the parametric study are conducted on the B-A3 beam using with strength and elastic modulus comparable to or higher than
2D FE models, because there was no gap between the slab panels concrete. However, different fillers (e.g. grout, mortar or plaster)
808 N. Khorsandnia et al. / Construction and Building Materials 163 (2018) 798–811

Fig. 9. Error evaluation of different theoretical methods compared to experimental results for (a) global (Pu, du, Ks and su) and (b) local (ultimate strain values at mid-span)
characteristics.

120 120

100 100

Total Load (kN)


Total Load (kN)

80 80

60 60
Test Test
40 40
Full contact
Full contact
Gap = 0.1 mm Gap = 0.1 mm
20 20 Gap = 1 mm
Gap = 1 mm
Gap = 2 mm Gap = 2 mm
0 0
0 20 40 60 80 100 0 1 2 3 4 5 6
Mid span deflection (mm) End slip (mm)
(a) (b)
Fig. 10. Sensitivity of (a) load-deflection response and (b) load-slip response, with respect to different gaps between slab segments.

120 120

100 100
Total Load (kN)

Total Load (kN)

80 80

60 60
Test Test
40 Full contact 40 Full contact
E = 20 GPa E = 20 GPa
E = 2 GPa E = 2 GPa
20 20
E = 0.2 GPa E = 0.2 GPa
Lower Bound Lower Bound
0 0
0 20 40 60 80 100 0 1 2 3 4 5 6
Mid span deflection (mm) End slip (mm)
(a) (b)
Fig. 11. Sensitivity of (a) load-deflection response and (b) load-slip response, with respect to different elastic modulus of filler used between slab segments.

with different mechanical properties can have significantly differ- sidered and the filler material is modelled by a linear elastic consti-
ent influences on the SLS and ULS response of TCC beams with pan- tutive law. The load-deflection and load-slip response of the TCC
elised RC slabs. beams with MOEs of 0.2, 2.0 and 20 GPa are shown in
The effect of the filler’s elastic modulus on the overall structural Fig. 11a and b, respectively. Furthermore, the lower and upper
behaviour of the TCC beams is studied by assuming three different bounds of load-deflection and load-slip diagrams are provided in
values for its MOE, i.e. 0.2, 2.0 and 20 GPa. The upper bound of the Fig. 11a and b. These lower and/or upper bound responses repre-
MOE values (i.e. 20 GPa) for filler material was adopted with sent scenarios in which there is a gap without filler and/or no
respect to MOE of concrete (contacting surfaces) to represent a sce- gap with full contact between adjacent RC slab panels, respec-
nario in which the slab along the TCC beam is nearly continuous. tively. Regarding Fig. 11, it can be concluded that the ultimate
The lower bound of 0.2 GPa was used to represent a scenario in loading capacity of the TCC beams with prefabricated slab panels
which the gaps between precast slabs were filled by a very weak is not sensitive to stiffness of the filler material. However, the
filler compared to concrete. In the TCC beam adopted for this part load-deflection response as well as stiffness of TCC beams at SLS
of the parametric study, a 2 mm gap between the RC panels is con- conditions is significantly influenced by the MOE of the filler. For
N. Khorsandnia et al. / Construction and Building Materials 163 (2018) 798–811 809

example, reducing this from 20 GPa to 0.2 GPa leads to a 50% non-linear FE models for capturing the global and local response
reduction in the serviceability stiffness of the TCC beam considered of the TCC beams under SLS and ULS conditions.
(Fig. 11a). Accordingly, the use of stiffer materials for the filler is
recommended to enhance the structural performance (particularly
3.8. Number of segments
the stiffness under SLS conditions) of the TCC system with prefab-
ricated RC slab panels.
Another factor that may affect the structural response of TCC
beams with segmented slabs is the number of panels over the
3.7. Stiffness of shear connectors beam length. Assuming 2 mm gaps between the slab segments
and a MOE of 2 GPa for the filler material within the gaps, beam
Deconstructable TCC beams with precast panelised RC slabs can B-A3 with 4, 6 and 8 slab segments was analysed using the 2D
be easily dismantled and reassembled. However, disassembling FE formulation and the load-deflection and load-slip results are
and reassembling the TCC beams involves loosening, removing shown in Fig. 13. It is seen that the number of slab panels has neg-
and fastening the screw shear connectors, that can reduce the stiff- ligible influence on the peak load carrying capacity of the TCC
ness of the TCC connections as reported in Khorsandnia et al. [23]. beam, but the stiffness of the TCC beams at SLS conditions can be
The stiffness reduction (due to unscrewing and screwing the shear significantly influenced by the number of slab panels over the
connectors) largely depends on screw type and size, timber type beam length. In the B-A3 beam, increasing the number of segments
and installation method. Accordingly, TCC beam B-A3 is analysed from 4 to 8 leads to a 30% reduction in the initial stiffness of the
assuming four different stiffness reduction factors (ranging from TCC beam (Fig. 13a). In addition, it is observable that the ultimate
40% to 100%) for shear connection A3 and the load-deflection and end slip in the TCC beams decreases as the number of slab panels
load-slip results are shown in Fig. 12a and b, respectively. The sig- increases, because part of the slip is diminished within the gap
nificant influence of the stiffness reduction factor on the structural between the slab panels.
response (i.e. load-deflection, peak load carrying capacity and ser- Finally, the adequacy and capabilities of the models adopted for
viceability stiffness) of the TCC beams is evident from Fig. 12a. It capturing different aspects of the behaviour of a deconstructable
can be also concluded that when the stiffness of the connectors TCC beam with prefabricated RC slab panels are summarised and
is noticeably reduced (e.g. 40–60% reduction factor) and/or the compared in Table 6. Amongst the analysis options considered,
number of shear connectors over the beam length is inadequate the analytical formulation is the simplest and most efficient one,
(e.g. under-designed TCC beams), significant non-linearities can whereas the continuum-based FE model is the most complex and
be observed in both the load-deflection and load-slip responses computationally expensive option. The continuum-based FE for-
of the TCC beams, and such situations warrant development of mulations are also the most versatile models for capturing various

120 120

100 100
Total Load (kN)

Total Load (kN)

80 80

60 60
Test Test
40 40
100% A3 connector 100% A3 connector
80% A3 connector 80% A3 connector
20 20
60% A3 connector 60% A3 connector
40% A3 connector 40% A3 connector
0 0
0 20 40 60 80 100 0 2 4 6 8
Mid span deflection (mm) End slip (mm)
(a) (b)
Fig. 12. Sensitivity of (a) load-deflection response and (b) load-slip response, with respect to different connector stiffness.

120 120

100 100
Total Load (kN)

Total Load (kN)

80 80

60 60
Test Test
40 Full contact 40 Full contact
4 segments 4 segments
20 6 segments 20 6 segments
8 segments 8 segments
0 0
0 20 40 60 80 100 0 1 2 3 4 5 6
Mid span deflection (mm) End slip (mm)
(a) (b)
Fig. 13. Sensitivity of (a) load-deflection response and (b) load-slip response, with respect to different number of slab segments.
810 N. Khorsandnia et al. / Construction and Building Materials 163 (2018) 798–811

Table 6
Comparison of predictive models with respect to different characteristics of deconstructable TCC beams.

Characteristic Analytical 1D frame FE 2D continuum-based FE


Full-designed TCC beams (adequate no. of connections) U U U
Under-designed TCC beams (inadequate no. of connections) ✗ U U
Service loads U U U
Ultimate loads ✗ U U
Minor segmentation effect U U U
Significant number of segments ✗ ✗ U
Negligible gap between segments U U U
Significant gap between segments ✗ ✗ U
Hard material as filler between segments U U U
Soft material as filler between segments ✗ ✗ U
Negligible change in connector stiffness after reassembling U U U
Significant change in connector stiffness after reassembling ✗ U U
Simplicity Easiest Moderate Most complex
Efficiency Highest Moderate Lowest

aspects of the structural behaviour of deconstructable TCC systems  The results of the parametric study show that the gaps between
with precast panelised slabs. The comparison of different models in the RC slab panels, the use of soft filler materials (with low elas-
Table 6 can serve as a preliminary guide for choosing simplest yet tic modulus) between the slab segments and/or increasing the
sufficiently accurate models for the analysis of TCC beams with number of RC slab panels over the beam length has a minor
precast panels. influence on the ultimate loading capacity of TCC beams with
precast segmented slabs. However, the structural performance
4. Conclusions under service loading conditions can be significantly influenced
by the number of slab panels, size of gaps or elastic modulus of
The simple analytical model of BS EN 1995-1 [31], a non-linear filler between the RC slab panels.
1D frame FE and a 2D continuum-based FE model were proposed  Dismantling and reassembling TCC beams by unscrewing and
and employed for the analysis of deconstructable TCC beams with re-screwing the shear connectors can lead to a noticeable
precast panelised RC slabs. The proposed formulations were veri- reduction in the stiffness of shear connection, that in turn can
fied against experimental data from 4-point bending tests con- significantly reduce the structural performance of the decon-
ducted previously by the authors, and the adequacy and accuracy structable TCC beams with panelised slabs under service and
of the models for capturing various aspects of the structural beha- ultimate limit state loading conditions.
viour of TCC beams with segmented slabs were discussed. The non-  Practical design of conventional TCC floors/beams (particularly
linear 2D FE models are the most accurate (among the analysis the long-span floors) is significantly influenced and mostly gov-
methods considered) and the 2D FE models can adequately capture erned by the short- and long-term serviceability limit state
the local response (strain in timber and concrete) and the global design requirements. Accordingly, further study is recom-
response (load-deflection, load-slip, stiffness and peak/ultimate mended to shed light on long-term behaviour of decon-
load carrying capacity) of TCC beams with panelised slabs. Accord- structable TCC joints and beams under variable environmental
ingly, a parametric study was carried out using the 2D FE models to conditions.
evaluate the structural behaviour of TCC beams with respect to the
number of slab panels, the size of the gaps between the panels, the
References
stiffness of the material used for filling the gaps between the pan-
els and the reduction in the stiffness of the shear connectors due to [1] A. Frangi, M. Fontana, Elasto-plastic model for timber-concrete composite
dismantling and reassembling of the beams. The main conclusions beams with ductile connection, Struct. Eng. Int. 13 (2003) 47–57.
[2] M. Fragiacomo, A finite element model for long-term analysis of timber-
of the analytical and numerical modelling and parametric study
concrete composite beams, Struct. Eng. Mech. 20 (2) (2005) 173–190.
are as follows. [3] B. Faggiano, A. Marzo, F.M. Mazzolani, L.M. Calado, Analysis of rectangular-
shaped collar connectors for composite timber-steel-concrete floors: push-out
 The analytical model of Eurocode 5 can only predict the tests, J. Civ. Eng. Manage. 15 (1) (2009) 47–58.
[4] A.M.P.G. Dias, A.R.D. Martins, L.M.C. Simões, P.M. Providência, A.A.M. Andrade,
response if the beam behaviour is predominantly linear up to Statistical analysis of timber–concrete connections–mechanical properties,
failure. This is the case when an adequate number of shear con- Comput. Struct. (2015).
nectors is used over the beam length and when the effect of slab [5] M. Stepinac, V. Rajcic, J. Barbalic, Influence of long term load on timber-
concrete composite systems, Gradevinar 67 (2015) 235–246.
segmentation is negligible. [6] N. Khorsandnia, J. Schänzlin, H. Valipour, K. Crews, Time-dependent behaviour
 The 1D frame FE formulation can efficiently capture the non- of timber–concrete composite members: numerical verification, sensitivity
linear behaviour of TCC beams in which the effect of RC slab and influence of material properties, Constr. Build. Mater. 66 (2014) 192–208.
[7] M. Fragiacomo, Long-term behaviour of timber-concrete composite beams. II:
segmentation and/or size of gaps between slab panels is mini- numerical analysis and simplified evaluation, J. Struct. Eng. (ASCE) 132 (1)
mal. If this is not the case, a more versatile 2D continuum- (2006) 23–33.
based FE model is required to predict the structural behaviour [8] M. Fragiacomo, A. Ceccotti, Long-term behavior of timber-concrete composite
beams. I: Finite element modelling and validation, J. Struct. Eng. (ASCE) 132 (1)
of TCC beams with panelised RC slabs. It should be noted that (2006) 13–22.
1D and 2D FE models are suitable for capturing the global beha- [9] E. Lukaszewska, M. Fragiacomo, H. Johnsson, Laboratory tests and numerical
viour of TCC beams and some aspects of local behaviour (e.g. nnalyses of prefabricated timber-concrete composite floors, ASCE J. Struct. Eng.
136 (1) (2010) 46–55.
strain distribution over the timber beam and concrete slab
[10] E. Lukaszewska, H. Johnsson, M. Fragiacomo, Performance of connections for
depth), particularly when the load-slip behaviour of the TCC prefabricated timber–concrete composite floors, Mater. Struct. 41 (9) (2008)
connections are available from push-out tests. But, predicting 1533–1550.
the local behaviour and failure mode of TCC connections require [11] D. Yeoh, M. Fragiacomo, A. Buchanan, C. Gerber, Preliminary research towards
a semi-prefabricated LVL-concrete composite floor system for the Australasian
versatile 3D FE models. market, Aust. J. Struct. Eng. 9 (3) (2009) 225–240.
N. Khorsandnia et al. / Construction and Building Materials 163 (2018) 798–811 811

[12] D. Yeoh, M. Fragiacomo, The design of a semi-prefabricated LVL-concrete [25] R. Crocetti, T. Sartori, R. Tomasi, Innovative timber-concrete composite
composite floor, Adv. Civ. Eng. 2012 (2012). structures with prefabricated FRC slabs, ASCE J. Struct. Eng. 141 (9) (2015)
[13] N. Jacquier, U.A. Girhammar, (2012) Tests on shear connections in 04014224.
prefabricated composite cross-laminated- timber and concrete elements, in: [26] A. Ataei, M. Bradford, H. Valipour, Moment-rotation model for blind-bolted
World Conference on Timber Engineering 2012: Architecture and Engineering flush end-plate connections in composite frame structures, J. Struct. Eng.
Case Studies, WCTE, 2012, Auckland. (2015) 04014211.
[14] M. Fragiacomo, E. Lukaszewska, Time-dependent behaviour of timber– [27] A. Ataei, M.A. Bradford, H.R. Valipour, Finite element analysis of HSS semi-rigid
concrete composite floors with prefabricated concrete slabs, Eng. Struct. 52 composite joints with precast concrete slabs and demountable bolted shear
(2013) 687–696. connectors, Finite Elem. Anal. Des. 122 (2016) 16–38.
[15] M. Fragiacomo, E. Lukaszewska, Development of prefabricated timber- [28] X. Liu, M.A. Bradford, Q.J. Chen, H. Ban, Finite element modelling of steel-
concrete composite floor systems, Proc. Inst. Civ. Eng. Struct. Build. 164 (2) concrete composite beams with high-strength friction-grip bolt shear
(2011) 117–129. connectors, Finite Elem. Anal. Des. 108 (2016) 54–65.
[16] T. Sartori, R. Crocetti, Prefabricated timber-concrete composite floors, Eur. J. [29] C. Bedon, M. Fragiacomo, Three-dimensional modelling of notched
Wood Wood Prod. (2016) 1–3. connections for timber-concrete composite beams, Struct. Eng. Int.: J. of the
[17] P. Kuklík, P. Nechanický, A. Kuklíková, Development of Prefabricated Timber- Int. Assoc. Bridge Struct. Eng. (IABSE) 27 (2) (2017) 184–196.
Concrete Composite Floors, RILEM Bookseries, 2014. [30] A. Zona, M. Barbato, M. Fragiacomo, Finite-element model updating and
[18] P. Nechanický, P. Kuklík, Development of prefabricated timber-concrete probabilistic analysis of timber-concrete composite beams, J. Struct. Eng.
composite floors, in: 2nd International Conference on Structures and (ASCE) 138 (7) (2012) 899–910.
Architecture, ICSA 2013. Guimaraes, 2013. [31] B.S. Institution, (2004) BS EN 1995-1-1, Eurocode 5, Design of Timber
[19] C.Y. Adam, H.R. Milner, Wood-based prefabricated composite-acting bridge Structures. Common Rules and Rules for Buildings, BSI, London, 2004.
deck, J. Bridge Eng. 17 (2) (2012) 363–370. [32] N. Khorsandnia, H. Valipour, K. Crews, Structural response of timber-concrete
[20] A. Ataei, M.A. Bradford, H.R. Valipour, Experimental study of flush end plate composite beams predicted by finite element models and manual calculations,
beam-to-CFST column composite joints with deconstructable bolted shear Adv. Struct. Eng. 17 (11) (2014) 1601–1621.
connectors, Eng. Struct. 99 (2015) 616–630. [33] N. Khorsandnia, H. Valipour, S. Foster, K. Crews, A force-based frame finite
[21] H. Valipour, A. Rajabi, S.J. Foster, M.A. Bradford, Arching behaviour of precast element formulation for analysis of two- and three-layered composite
concrete slabs in a deconstructable composite bridge deck, Constr. Build. beams with material non-linearity, Int. J. Non-Linear Mech. 62 (2014)
Mater. 87 (2015) 67–77. 12–22.
[22] A. Ataei, M.A. Bradford, H.R. Valipour, X. Liu, Experimental study of sustainable [34] Glos, P., Zur Modellierung des Festigkeitsverhaltens von Bauholz bei Druck-,
high strength steel flush end plate beam-to-column composite joints with Zug- und Biegebeanspruchung. Laboratorium für den konstruktiven
deconstructable bolted shear connectors, Eng. Struct. 123 (2016) 124–140. Ingenieurbau (LKI), TU München, 1981 – Forschungsbericht.
[23] N. Khorsandnia, H. Valipour, J. Schänzlin, K. Crews, Experimental [35] CEB-FIP, CEB-FIP model code 2010, Switzerland, International Federation for
investigations of deconstructable timber-concrete composite beams, J. Structural Concrete (fib), 2010.
Struct. Eng. (2016) 04016130. [36] ABAQUS, Ver. 6.11. Documentation ed., Ó Dassault Systèmes, 2011.
[24] X. Liu, M.A. Bradford, M.S.S. Lee, Behavior of high-strength friction-grip bolted
shear connectors in sustainable composite beams, J. Struct. Eng. (United
States) 141 (6) (2015).

Das könnte Ihnen auch gefallen