World Academy of Science, Engineering and Technology

International Journal of Structural and Construction Engineering

Vol:13, No:9, 2019

Experimental Investigation of Cold-Formed

Steel-Timber Board Composite Floor Systems
Samar Raffoul, Martin Heywood, Dimitrios Moutaftsis, Michael Rowell

to the performance of a bare steel joist. In another study,

Abstract—This paper comprises an experimental investigation specimens with screw spacing of 150 mm and adhesive
into the structural performance of cold formed steel (CFS) and timber exhibited more than 40% increase in flexural stiffness,
board composite floor systems. The tests include a series of small- indicating that negligible improvements were attained by
scale pushout tests and full-scale bending tests carried out using a decreasing the screw spacing from 150 mm to 100 mm [8].
refined loading system to simulate uniformly distributed constant
load. The influence of connection details (screw spacing and It is well known that the degree of shear connection as well
as the connection load-slip behavior (shear connector
Open Science Index, Structural and Construction Engineering Vol:13, No:9, 2019 publications.waset.org/10010706/pdf

adhesives) on floor performance was investigated. The results are

then compared to predictions from relevant existing models for stiffness) can have a significant influence on the performance
composite floor systems. The results of this research demonstrate the of a composite system [9]. For instance, a lower shear
significant benefits of considering the composite action of the boards connector stiffness would lead to higher deflections in the
in floor design. Depending on connection detail, an increase in system (as a result of slip), making it difficult to achieve the
flexural stiffness of up to 40% was observed in the floor system,
when compared to designing joists individually. “full shear interaction” of the composite components. To
further understand the effect of connection detail on the
Keywords—Cold formed steel joists, composite action, flooring composite floor interaction, push-out tests have been
systems, shear connection. performed, leading to the development of load-slip predictive
equations [7], [10], [11]. Overall, the push-out test results
I. INTRODUCTION indicate significantly higher strengths and connection

L IGHTWEIGHT floors built with CFS joists and structural stiffness, in joints with glue, compared to specimens with only
timber-based deck (referred to in this study as timber mechanical connectors [7], [10]. Nevertheless, despite the
boards or floor boards) have become increasingly popular in apparent benefits of adhesives on composite interaction
both commercial and residential construction due to their behavior, their effect has not been included in predictive
attractive advantages such as high strength-to-weight ratios, models.
speed of build and sustainability [1]. While the benefits of This study aims to examine, experimentally, the interaction
composite construction (e.g. hot rolled steel beam-concrete between CFS joists and floor boards, typically used in
slab systems or engineered timber joists) are well-established residential flooring, with different connection details. The
and exploited [2], the interaction between CFS joists and the potential benefits of considering their composite behavior is
accompanying flooring is still not well understood, with only a assessed, in terms of strength and stiffness, by comparing to
few research studies in the field [3]-[7]. As a result, the control bare steel joists. Finally, the results are compared to
advantageous interaction between the timber boards and CFS analytical equations from the literature and concluding
joists is often ignored, leading to conservative designs. remarks are provided. This research contributes to the
Recently conducted experiments on cold or hot rolled steel- understanding of the behavior of CFS-timber board composite
timber board composite systems show that significant benefits flooring systems, leading to its future implementation and
can be achieved by mobilizing the interaction of the floor standardization. The work presented in this article is part of a
boards with the steel joists [6]-[8]. It has been shown that the continuing collaborative research project between Fusion
composite efficiency is higher in longer spans [6] and smaller Building Systems and Oxford Brookes University, to
steel gauge [7]. It is also highly influenced by connection type investigate more efficient structural systems as part of a
(e.g. screws or bolts [6]), connection spacing and the presence Knowledge Transfer Partnership (KTP), sponsored by
of a structural adhesive [7]. Research by [7] reports that high Innovate UK.
degrees of shear connection could be achieved by using
structural adhesive and a screw spacing of 100 mm leading to II.EXPERIMENTAL PROGRAM
100% and 40% increase in the moment capacity and flexural A. Materials
stiffness of the structural system, respectively, when compared
All joist sections are manufactured from S350 (min yield
strength of 390 MPa) zinc coated Z275g/m2 galvanized steel
S. R. is with Oxford Brookes University and is working with a partner
company, Fusion Building Systems, as a KTP associate, Oxford, OX3 0BP, coils to EN 10326:2004 using roll-formers at the Fusion
UK (phone: +44 (0) 1604 490540; e-mail: [email protected]). Building System production facility. Joists are lipped C-
M. H. and D. M. are with Oxford Brookes University, Oxford, OX3 0BP, sections of depth: 254 mm, flange width: 50 mm; lips:12 mm
UK (e-mail: [email protected], [email protected]).
M. R. is with Fusion Building Systems, Northampton, NN3 6HE, UK
and a thickness of 1.5 mm.
(phone: +44 (0) 7377 864920; e-mail: [email protected]). Floor boards are 22 m thick, 2400 mm x 600 mm P5 grade

International Scholarly and Scientific Research & Innovation 13(9) 2019 509 ISNI:0000000091950263
 World Academy of Science, Engineering and Technology
International Journal of Structural and Construction Engineering
Vol:13, No:9, 2019

randomly oriented chipboard, which follows the BS EN 312- composite test cassettes consisted of two parallel joists spaced
2010 recommendations for flooring products. The mechanical 600-mm apart (to reflect typical construction detail) and
properties of the floor boards (as per manufacturer) are: fastened to a 22-mm timber-based floorboard (1200 mm wide)
Modulus of Elasticity: 2150 MPa; Bending Strength: 14 MPa. using the selected mechanical fixing detail. Fig. 2 presents a
Fasteners used are loose countersunk self-drilling screws schematic diagram of a typical composite system cross-
with reamers of the following dimensions: Head diameter: 7.5 section. The bare frame specimens had similar joist
mm; thread diameter: 4.15 mm; wire diameter: 3.36 mm; arrangement but excluding the floorboards. To allow for the
length: 40 mm. The tensile and shear strengths of the screws, application of the line loads and to allow for global
as per manufacturer are 10 kN and 4.6 kN, respectively. measurements at midspan, thin strips of timber were placed at
To improve board-to-board and board-to-joist bonding, a the underside of the line loads and at midspan. The joists’
structural adhesive was used in some specimens in webs were stiffened locally using short stud lengths (250 mm)
conjunction with the screws. The adhesive used was a class to avoid premature failure at the panel extremities and at the
D4 polyurethane bonding adhesive, typically used in board-to- underside of the line loads positions. In the case of the bare
board and board-to-joist connections in timber construction. frame, angle brackets (lined internally with thin strips of
The mechanical properties of the adhesives are not readily polytetrafluoroethylene) were used to prevent excessive
Open Science Index, Structural and Construction Engineering Vol:13, No:9, 2019 publications.waset.org/10010706/pdf

available from the manufacturers/literature. twisting.

B. Pushout Tests – Specimen Details and Test Setup
Push-out connection tests were performed on small-scale
specimens to acquire load-slip characteristics of the joist-to-
board connections. To reduce eccentricities, the push-out test
specimens were configured symmetrically about both axes.
The setup comprised of two C-joists, arranged back-to-back
(5mm apart) and sandwiched between two floor boards. The
floor boards were mechanically fixed to the joists flanges
using i) two screws per flange spaced 300 mm apart (P1.5-
300), ii) adhesive only (specimen P1.5-A), or iii) screws and
adhesive (P1.5-300-A). Three identical specimens were
fabricated and tested per connection type.
The load protocol was based on the recommendations of BS
EN 26891 [12] aimed for mechanically fastened joints in
timber. Accordingly, the specimens were loaded up to
approximately 40% of the estimated ultimate load capacity,
unloaded, then reloaded until either failure or an average slip
of more than 15 mm was achieved. The tests were performed
at load-controlled rate of 10 kN/m for P1.5-300 specimens and
20 kN/m for the other specimens.
The relative slip between the board and the steel joists was
monitored using four linear variable differential transformers
(LVDTs) mounted onto four different corners of the joists’
webs. This configuration also captures any twisting/bending in
the specimen. The typical specimen dimensions as well as the
position of the instrumentation are presented in Fig. 1.
C.Full Scale Tests - Specimen Details and Test Setup
A total of 15 full scale cassette prototypes were tested in
bending to evaluate the flexural response of the composite
CFS joist – timber board flooring system. The tests comprised
both composite cassettes and bare frame specimens
(i.e. joists without the board). The main parameters Fig. 1 Connection test setup front elevation (a), side elevation (b),
investigated included i) the influence of board-to-joist screw and plan view (c)
spacing (150 mm or 300 mm) and ii) the influence of the use
of adhesives, in addition to the screws, to improve composite The test cassettes were subjected to bending using a refined
performance. loading system to simulate a uniformly distributed load, as
All specimens were simply supported on steel beams shown in Fig. 3. The loads were applied through two actuators
arranged to simulate standard pin and roller boundary and distributed through two spreader beams onto two cross
conditions. The specimens were 5.4 m long, as per [13]. The beams (each) to generate a total of four line loads across the
span of the beam. The cross-beams were positioned apart at

International Scholarly and Scientific Research & Innovation 13(9) 2019 510 ISNI:0000000091950263
 World Academy of Science, Engineering and Technology
International Journal of Structural and Construction Engineering
Vol:13, No:9, 2019

quarter span lengths and at a distance of an eighth of the span displacement, but failed at much lower loads than the other
length from the support. To ensure the vertical application of tested connections. Specimens prepared with adhesive only
the load at high deformations, four rollers were placed (P-A) failed predominantly by shearing in the adhesive. The
between the spreader beams and the cross-beams. failure was brittle, with very little steel to timber relative
displacement. The detached steel and timber board had
significant adhesive residue on the surface, indicating strong
glue-to-timber and glue-to-steel adhesion (see Fig. 4).

TABLE I
PUSH OUT TEST RESULTS
Fmax δmax ks1
Specimen
(kN) (mm) (kN/m)
P-300-1 29.5 9.67 8.9
P-300-2 30.9 9.67 10.6
P-300-3 27.8 9.78 8.2
Fig. 2 Cross-section of composite test floor cassette
P-A-1 56.2 0.29 260.7
Open Science Index, Structural and Construction Engineering Vol:13, No:9, 2019 publications.waset.org/10010706/pdf

P-A-2 78.5 0.45 189.1

P-A-3 46.7 0.57 242.9
P-300-A-1 101.2 1.21 109.7
P-300-A-2 155.8 1.13 177.1
P-300-A-3 88.7 1.75 -*

Fig. 3 Typical load configuration for the bare frame (a) and
composite cassette (b)

The specimens were subjected to initial load cycling to

ensure the appropriate seating of the specimen as well as the
proper function of instrumentation. The load protocol was as
follows: Fig. 4 Timber board and steel joist interface upon failure in
1) Load to about 60% of the live-load serviceability loading, specimens with adhesive only (P1.5-A)
then unload at a rate of 0.05 kN/sec
2) Load to full serviceability load, then unload at a rate of Specimens with screws and glue (P-300-A) exhibited an
0.05 kN/sec intermediate failure mode, including tilting in the screws (see
3) Load to failure at a displacement-controlled rate of 0.1 Fig. 5), but failure at a much lower overall displacements
mm/sec compared to specimens with screws only.
To measure the vertical displacements of the cassette, two
linear variable transducers were placed at the underside of the
joists and at the underside of the floor board at specimen
midspan. In the case of the bare frame, the latter
measurements were taken from a thin timber strip fixed to the
specimen at midspan. Two LVDTs were also placed at the
underside of the supports to measure any support vertical
displacement.

III. RESULTS AND ANALYSIS

A. Pushout Tests
The maximum load (Fmax), displacement at maximum load
(
) and slip modulus (ks1), which were estimated as per BS Fig. 5 Failure in specimens with screws and adhesive (P-300-A)
EN 2689, are summarized in Table I.
Specimens with screws only (P-300) underwent significant No visible crushing was observed in the timber,
tilting and bending in the screws, until the screws sheared in nevertheless, in some specimens, crushing was visible in the
two. As shown in Table I, these experienced the highest top part of the steel joists (at the underside of the loading

International Scholarly and Scientific Research & Innovation 13(9) 2019 511 ISNI:0000000091950263
 World Academy of Science, Engineering and Technology
International Journal of Structural and Construction Engineering
Vol:13, No:9, 2019

plate), which may have led to some specimen bending and serviceability DL+LL loading, where the load-deflection
variability in the test results. This could be due to curves presented a linear-elastic behavior.
unintentional eccentricity in some of the tested specimens. The specimens are identified according to the type of
specimen (BF: bare frame or C: composite), following by the
B. Full-Scale Tests
screw spacing in mm (300 or 150), followed by specimen
The test cassettes failed predominantly in plane. In the case number. In specimens where an adhesive was used, the
of the bare frame (BF) specimens, the failure consisted of a specimen number was preceded by the letter “A”.
combination of local buckling at the top flanges at midspan
and a global bending failure (see Fig. 6). Such local buckling
mode may have been influenced by the timber strip at midspan
used for placing displacement instrumentation.
Open Science Index, Structural and Construction Engineering Vol:13, No:9, 2019 publications.waset.org/10010706/pdf

Fig. 8 Load displacement curves for a representative specimen from

each tested parameter
Fig. 6 Interaction bending/local buckling failures in BF-2
TABLE II
In the case of the composite cassettes, the failure mode was AVERAGE LARGE-SCALE TEST RESULTS
influenced by the shear connection detail. Specimens with Increase Failure load
δGlobal# δLocal# EIeff,exp
Specimen relative to kN
screws only tend to experience local buckling at the underside mm (CoV*) mm (CoV*) kN.m2
BF (%) (CoV*)
of the line loads position (see Fig. 7 (a)), whereas specimens BF 18.9 (5.6) 19.4 (6.0) 1963.5 - 33.4 (9.4)
with screws and adhesive failure occur predominantly by C-300 16.3 (2.3) 16.2 (3.7) 2302.4 17.3 51.9 (1.5)
shear (see Fig. 7 (b)). As the specimens approached high C-150 16.0 (2.2) 16.3 (2.0) 2356.4 20.0 58.9 (3.1)
displacements, minor twisting was observed in the steel joists; C-300-A 14.2 (1.7) 14.4 (2.0) 2572.4 31.0 64.9 (0.2)
nevertheless, this was effectively controlled in the case of bare C-150-A 13.6 (1.8) 13.8 (2.5) 2733.8 39.2 73.3 (7.5)
#
frame specimens using the angle brackets. *
Deflections measured at DL+LL serviceability loads
CoV: coefficient of variation in %

The results in Fig. 8 and Table II indicate a significant

increase in stiffness when the timber board-to-joist interaction
was mobilized. For instance, compared to the bare frame
response, the composite cassette stiffness increased by a
maximum of 39%, respectively, for specimens with screws at
150 mm and adhesive. Additionally, the average failure load
of the composite C-150-A increased by up to 120%.
As expected, the lowest increase in flexural stiffness was
a b observed for C-300 specimens with screws only (at 300 mm),
which were about 17% stiffer than the bare frame specimens.
Fig. 7 Typical failure in specimens with a) screws only (C-150-2) and The use of a smaller screw spacing of 150 mm did not
b) screws and adhesive (C-300-A-1)
significantly increase this flexural stiffness, which reached
20% increase, compared to the bare frame specimens. The
Fig. 8 presents the load displacement measurements (taken
increase in flexural stiffness almost doubled, when adhesives
at midspan) for a typical specimen from each tested parameter,
were incorporated at the timber-to-steel joint interface, rather
while Table II presents average global and local measurements
than using screws only. This could be attributed to the much
at serviceability live load (LL) + dead load (DL), as well as
higher slip modulus of specimens with screws and glue, as
the experimental flexural stiffness (EIeff,exp) and the failure
opposed to specimens with screws only, as presented in Table
load. The % increase in flexural stiffness relative the bare
I. Compared to the bare frame, the increase in stiffness was
frame experimental stiffness was also presented in Table II.
31% to 39% for composites with adhesive and screw spacings
The experimental flexural stiffness EIeff,exp was derived based
of 300 mm and 150mm, respectively. The use of a tighter
on global mid-span deflections between 30% and 60%
screw spacing led to a more significant increase in composite

International Scholarly and Scientific Research & Innovation 13(9) 2019 512 ISNI:0000000091950263
 World Academy of Science, Engineering and Technology
International Journal of Structural and Construction Engineering
Vol:13, No:9, 2019

flexural stiffness in specimens with adhesives, than in specimens with adhesive, the full length of the flange was
specimens excluding adhesives. The above increase in flexural used, rather than screw spacing, as in [14].
stiffness is reflected by a decrease in the deflection of the
specimens at a particular loading, indicating that the TABLE III
PREDICTIONS OF FLEXURAL STIFFNESS
composite section can fulfill more serviceability load
EI predicted Error (%) EI predicted Error (%)
deflection requirements, when the board action is mobilized. Specimen (kN.m2) Relative to (kN.m2) Relative to
Overall, the data in Table II are very consistent with an [5] experiment [14] experiment
average coefficient of variation <3.7% for deflection in the C-300 1983 -16% 2056 -12%
composite test cassettes and <6% in the bare frame specimens. C-300-A 2057 -25% 2065 -24%
A slightly higher variability was observed in the ultimate
failure loads achieved, i.e. a maximum of 7.5% for the The above comparisons indicate that existing models do not
composite specimens and 9.4% for the bare frame specimens. capture very well the advantageous performance of composite
The higher variability in the bare frame specimens and failure cassettes, particularly those with structural adhesives. Further
load can be due to the higher effect of localized buckling. It tests including connection tests using other screws and
must be noted that the failure load does not influence flexural adhesive detailing may be useful to calibrate these models to
Open Science Index, Structural and Construction Engineering Vol:13, No:9, 2019 publications.waset.org/10010706/pdf

stiffness and serviceability considerations, which are the suit various connection properties.
purpose of this study.
V.CONCLUSION
IV. ANALYTICAL CALCULATIONS An experimental program comprising of a total of 15 large-
A. Fully-Composite Section scale tests and nine push-out tests were implemented to
investigate the effect of connection detail on the flexural
The effective flexural stiffness of a fully composite system
performance of a timber board-CFS joist composite flooring
is calculated using (1)
system. Two screw spacings, namely 150 mm and 300 mm
were investigated as well as the influence of a structural
  
  
  (1)
adhesive, when applied at the joist to board interface.
where Is is the moment of inertia of the steel about its major Among the parameters investigated, the application of a
axis, As is the gross area of the steel section, ds is the distance structural adhesive (in accompaniment to the screws)
from the centroid of steel section to the centroid of the presented the highest benefits to the overall performance of
composite, Ib is the moment of inertia of the transformed the floor (in terms of flexural stiffness and failure strength)
board section about its major axis, Ab is the area of the while reducing the screw spacing from 300 mm to 150 mm,
transformed board section, and db is the distance from the did not seem to have a significant influence. This could be due
centroid of the board section to the centroid of the composite to the much higher slip modulus achieved in specimens with
The analytical flexural stiffness of the bare frame specimens glue and screws, as opposed to specimens with screws only.
comprising two 5.4 m steel joists is calculated as follows: The results indicate a significant improvement in the
Es:210 GPa; Is: 469.4 cm4; EsIs: 1971.5 kN.m2. Based on (1), structural performance of the composite flooring system when
the flexural stiffness of the section is shown in Fig. 2, the board-to-steel interaction was mobilized. Compared to the
assuming that a fully composite system is 2838 kN/m2. bare steel frame, specimens with screws (at 150 mm) and
It can be observed that the analytical flexural stiffness of the adhesives experienced about 40% increase in flexural stiffness
bare frame specimens almost concurs with the flexural (reaching 96% of the fully composite section stiffness) and
stiffness obtained experimentally (see Table II). It can also be 120% increase in failure load. Such improvements in floor
observed that the experimental flexural stiffness (EIeff,exp) for performance, if implemented in design, can either lead to
specimens with the highest stiffness (C-150-A) reaches almost reducing the steel requirements for a specific span length or
96% of the flexural composite section, indicating that the increasing the span lengths capabilities of a specific steel
specimen is nearly fully composite and that high shear load section, leading to invaluable material and embodied carbon
transfer occurs at such connection detail. savings.
Predictive equations available in the literature do not appear
B. Flexural Stiffness Predictions to capture well the performance of composite floors including
Few models are proposed in the literature to predict the adhesives. The results of this study, including the composite
effective flexural stiffness of timber board-steel joist cassette test data and load-slip relationships from push-out
composite systems [5], [14]. These models account for the slip tests are fundamental for the development of analytical
modulus of the shear connector, mainly consisting of equations that can reflect the effect of both screws and
mechanical fasteners. Table III presents the predictions from structural adhesives on the performance of such flooring
equations in [5], [14], compared to the stiffness obtained from systems. On-going research involves additional experiments
the experimental data for specimens with screw spacing of 300 and numerical modeling to establish the influence of other key
mm, with or without the presence of an adhesive. The slip parameters, leading to the development of more
moduli used in these equations are based on the average of the comprehensive design equations.
data presented in Table I. To determine the modulus/mm for

International Scholarly and Scientific Research & Innovation 13(9) 2019 513 ISNI:0000000091950263
 World Academy of Science, Engineering and Technology
International Journal of Structural and Construction Engineering
Vol:13, No:9, 2019

ACKNOWLEDGMENT
The authors are thankful for Fusion Building Systems and
Oxford Brookes University, for their valuable contributions to
this research. The authors are also thankful for Ray Salter, lab
manager at Oxford Brookes University, for his continuous
support in the lab. This partnership received financial support
from the Knowledge Transfer Partnerships (KTP) programme.
This successful Knowledge Transfer Partnership project,
funded by the UK Research and Innovation through Innovate
UK, is part of the government’s Industrial Strategy.

REFERENCES
[1] R. M. Lawson, R. G. Ogden, R. Pedreschi, and S. O. Popo-Ola,
“Developments of cold-formed steel sections in composite applications
for residential buildings,” Advances in Structural Engineering, vol. 11
Open Science Index, Structural and Construction Engineering Vol:13, No:9, 2019 publications.waset.org/10010706/pdf

(6), 2008, pp. 651-660.

[2] Committee of European Normalisation (CEN), EN 1994-1-1-2004,
Eurocode 4: Design of composite steel and concrete structures, Part 1-1
General Rules and Rules for buildings, Brussels, the Netherlands.
[3] L. Xu, and F.M. Tangorra, “Experimental investigation of lightweight
residential floors supported by cold-formed steel C-shape joists,”
Journal of Constructional Steel Research, vol. 63 (3), 2007, pp. 422-
435.
[4] P. J. Grubb, R. Thepaut, and C. Mettem, “Lightweight steel/timber
composite solutions: Information and guidance for new product
development,” SCI Publication, SCI ED002, Silwood Park, Ascot, UK,
2003.
[5] C. Loss, and B. Davison, “Innovative composite steel-timber floors with
prefabricated modular components,” Engineering Structures, vol. 132,
pp. 695-713.
[6] A. Hassanieh, H. R. Valipour, M. A. Bradford, “Experimental and
numerical study of steel-timber composite (STC) beams,” Journal of
Constructional Steel Research, vol. 122, 2016, pp. 367-378.
[7] P. Kyvelou, L. Gardner, and D. A. Nethercot, “Testing and analysis of
composite cold-formed steel and wood-based flooring systems,” Journal
of Structural Engineering, vol. 143 (11), 2017, 04017146.
[8] P. Kyvelou, L. Gardner, D. A. Nethercot, “Composite action between
cold-formed steel beams and wood-based floorboards,” in the
International Journal of Structural Stability and Dynamics, vol.15 (08),
2015, 1540029.
[9] R. M. Lawson, D. Lam, E. S. Aggelopoulos, and S. Nellinger,
“Serviceability performance of steel-concrete composite beams,” in
Proceedings of the Institution of Civil Engineers on Structures and
Buildings, vol. 170 Issue SB2.
[10] A. Hassanieh, H. R. Valipour, and M. A. Bradford, “Experimental and
analytical behavior of steel-timber composite connections,”
Construction and Building Materials, vol. 118, 2016, pp. 63-75.
[11] B. S. Lakkavalli, and Y. Liu, “Experimental study of composite cold-
formed steel C-section floor joists,” Journal of Constructional Steel
Research, vol. 62, 2006, pp. 995-1006.
[12] Committee of European Normalisation (CEN), EN 26891, Timber
structures – Joints made with mechanical fasteners – General principles
for the determination of strength and deformation characteristics.
Brussels, the Netherlands, 1991.
[13] European Organisation for Technical Approvals (EOTA), TR 002, Test
methods for light composite wood-based beams and columns,
Charlottenlund, Denmark, October, 2000.
[14] P. Kyvelou, L. Gardner, D. A. Nethercot, “Design of cold-formed steel
flooring systems,” Structures, 2017, pp. 242-252.

International Scholarly and Scientific Research & Innovation 13(9) 2019 514 ISNI:0000000091950263