Download as pdf or txt
Download as pdf or txt
You are on page 1of 16

Experiments on Long-Span Cold-Formed Steel

Single C-Section Portal Frames


Rinchen Rinchen 1 and Kim J. R. Rasmussen, M.ASCE 2

Abstract: This paper presents a series of full-scale tests on long-span cold-formed steel single C-section portal frames subjected to gravity
Downloaded from ascelibrary.org by University of Central Florida on 11/21/20. Copyright ASCE. For personal use only; all rights reserved.

load and combined gravity and lateral load. The tests were carried out with the objectives of observing the structural performance of an
existing portal frame configuration, determining the ultimate strength of the existing and modified portal frame configurations, and validating
the structural behavior simulated by numerical models. The tests on cold-formed steel single C-section portal frames demonstrated that the
flexural-torsional buckling of columns was the dominant structural behavior. The ultimate failure deformations were concentrated at the eaves
and the apex and were governed by the type of fasteners used for the connections of column lips to the connection brackets. The strengths
of the frames with modified joint configuration were slightly higher than those of the frames with the existing proprietary joint configuration.
The ultimate loads obtained from the tests are compared with ultimate loads predicted by the Direct Strength Method. A method for including
the bimoment in the design of CFS portal frames is highlighted. DOI: 10.1061/(ASCE)ST.1943-541X.0002487. © 2019 American Society of
Civil Engineers.

Introduction bolted back-to-back at the eaves and apex joints. Several tests were
also conducted on CFS connections and frames to evaluate their
Portal frames are structural frameworks that are commonly used seismic performance (Sabbagh et al. 2010, 2012; Mojtabaei et al.
in the construction of single-story buildings such as residential 2018; Ye et al. 2019).
houses, barns, sheds, workshops, garages, and warehouses. They In the present test series, CFS portal frames comprising single
are also often used where there is a requirement for a large free C-sections as columns and rafters with semirigid joints subjected to
space within the building as in aircraft hangars and in large indus- gravity load and horizontal load are examined. The test configura-
trial buildings. Among the various forms of portal frames, the tion consisted of two-bay CFS portal frames with center-to-center
single span portal frame tends to be the most popular one owing span of 13.6 m and apex height of 6.8 m. The frames were spaced at
to its simplicity in form and construction. 3.65 m with rafters braced by purlins while the columns were un-
Although several tests on portal frames and connections made braced between their ends. Such types of portal frames are ideal for
from cold-formed steel (CFS) sections have been carried out in the garages and storage sheds where the requirement for vehicular ac-
past, the focus has been largely on CFS portal frames composed cess to the inside of the building precludes the provision of girts in
of doubly-symmetric sections fabricated by placing two individual the facades. However, of the six frames, one was tested with col-
C-sections back-to-back (Kirk 1986; Chung and Lau 1999; Lim umns restrained by girts at a spacing of 1.25 m along its height for
and Nethercot 2002, 2004; Dubina et al. 2004; Yu et al. 2005; comparing the behavior of this frame with the behavior of frames
Stratan et al. 2006; Rhodes and Burns 2006; Kwon et al. 2006; without column restraints. In all the full-scale frame tests, cladding
Elkersh 2010; Jackson et al. 2012; Zhang and Rasmussen 2014; was not provided and therefore stressed-skin action was not con-
Wrzesien et al. 2015, 2016; Blum and Rasmussen 2019). The sidered. As presented in the following pages, the structural behav-
works of Wrzesien et al. (2015, 2016), however, were mainly ior of the single C-section CFS portal frame can be quite distinct
focused on interaction of joint effects and stressed skin action that from the behavior of doubly symmetric section CFS portal frames.
has the beneficial effect of enhancing frame strength and stiffness. One of the salient features that distinguishes the behavior of single
Full-scale tests on single C-section portal frames and connections C-section CFS portal frames is their propensity to undergo large
have been relatively rare apart from the tests by Baigent and flexural-torsional deformations right from the onset of loading,
Hancock (1982), Mills and LaBoube (2004), Dundu and Kemp resulting primarily from the noncoincident location of the shear
(2006), and Dunai (2007). Dundu (2011) also proposed a design center and the centroid of the cross-sections, compounded by the
approach for CFS portal frames in which the rafter and column nonsymmetric arrangement of bolts at the joints.
members are formed from single CFS channel sections which are In the following sections, a series of coupons extracted from the
column sections and brackets to ascertain the material properties
are described first. This is followed by the description of the full-
1
Research Associate, School of Civil Engineering, Univ. of Sydney, scale frame tests including the test preparation, testing process,
NSW 2006, Australia (corresponding author). ORCID: https://orcid.org and results. The frame capacities predicted by the Direct Strength
/0000-0003-1679-9323. Email: rinchen.rinchen@sydney.edu.au; rrin8775@
Method (DSM) are subsequently compared with the ultimate
uni.sydney.edu.au
2
Challis Professor, School of Civil Engineering, Univ. of Sydney,
capacities obtained from the tests.
NSW 2006, Australia. Email: kim.rasmussen@sydney.edu.au
Note. This manuscript was submitted on October 7, 2018; approved
on June 5, 2019; published online on November 15, 2019. Discussion Material Tests
period open until April 15, 2020; separate discussions must be submitted
for individual papers. This paper is part of the Journal of Structural The mechanical properties of the CFS sections are required for the
Engineering, © ASCE, ISSN 0733-9445. accurate prediction of structural behavior using numerical analysis.

© ASCE 04019187-1 J. Struct. Eng.

J. Struct. Eng., 2020, 146(1): 04019187


Hence, toward establishing the material properties and determining To make the stress-strain relations invariant to the rate of load-
the postyield material behavior, uniaxial tensile tests were per- ing, the static curves were obtained by reducing the original dy-
formed on four sets of coupons with each set consisting of three namic curve by the stress relaxations that occurred during the
specimens. The first two sets, each with three specimens, were ex- pauses while testing. The static stress-strain curves obtained from
tracted from the center portion of the web of the column sections the coupon tests and the corresponding initial elastoplastic portion
while the other two sets, each with three specimens, were cut from of the curves are presented in Figs. 1–3 for the column section and
the flat portion of the bottom flanges of the eaves and apex brack- eaves and apex brackets, respectively. The mechanical properties
ets, respectively. Of the two sets extracted from the column, one set of the mean of the three coupon tests in each group are presented in
belonged to the longitudinal direction and the other set to the trans-
Table 1, including the elastic Young’s modulus (E), the proportion-
verse direction.
ality stress taken as the 0.01% proof stress (σ0.01% ), the equivalent
The specimens were cut to dimensions in accordance with the
yield stress taken as the 0.2% proof stress (σ0.2% ), the ultimate ten-
Australian Standard AS 1391 (AS 2007). The coupon tests were
performed in a 50 kN capacity MTS tensile testing machine. To al- sile strength (σu ), the uniform strain corresponding to the ultimate
Downloaded from ascelibrary.org by University of Central Florida on 11/21/20. Copyright ASCE. For personal use only; all rights reserved.

low stress relaxation and to obtain static material properties, the tests tensile strength (εu ), the total elongation after fracture (εf ), and the
were paused for at least 100 s (Huang and Young 2014) for three Ramberg-Osgood parameter (n) describing the roundedness of the
different stages of loading during the tests: near the yield stress, near stress-strain curve in the transition from the elastic to the inelastic
the ultimate stress, and one in between. Both an extensometer and range, calculated as n ¼ lnð20Þ=lnðσ0.2% =σ0.01% Þ. The values of n
strain gauges were used for measuring the extension of the coupons. given in Table 1 agree well with the test results for G450 grade
The use of strain gauges at the midheight of coupons allowed ac- CFS (Huang and Young 2014) where n-values ranged from 17.6
curate determination of the initial elastic modulus. to 33.9.

Fig. 1. Stress-strain curves of column coupons (longitudinal direction).

Fig. 2. Stress-strain curves of eaves bracket coupons.

© ASCE 04019187-2 J. Struct. Eng.

J. Struct. Eng., 2020, 146(1): 04019187


Downloaded from ascelibrary.org by University of Central Florida on 11/21/20. Copyright ASCE. For personal use only; all rights reserved.

Fig. 3. Stress-strain curves of apex bracket coupons.

Table 1. Mechanical properties of C30024 section and connection brackets (static values)
Coupon source E (GPa) σ0.01% (MPa) σ0.2% (MPa) σu (MPa) σu =σ0.2% N εu (%) εf (%)
Column (long) 207 415 495 524 1.06 17.1 6.2 14.0
Column (trans) 219 500 548 566 1.03 33.4 3.2 10.3
Eaves bracket 206 474 513 534 1.04 38.0 5.7 14.0
Apex bracket 208 435 504 528 1.05 21.1 6.5 15.0

Single C-Section Portal Frame Tests

Frame Geometry
Six full-scale frame tests were carried out on single C-section CFS
portal frames having a center-to-center span of 13.6 m, eaves height
of 5.7 m, and apex height of 6.8 m (Fig. 4). The tests were carried
out with the aim of observing the structural performance of the
existing portal frame configuration, enhancing the strength of the
system, and calibrating numerical models.
The schematic layout of the frame is shown in Fig. 5 and the test
frame elevation is shown in Fig. 6. The test configuration consisted
of two-bay CFS portal frames with rafters braced by a series of
purlins. The columns and rafters of the central frame are single
C-section members (C30024) connected at the base, eaves, and
apex by using connection brackets that are specific to the single
C-section portal frames. It is this central frame (labeled as test
frame in Fig. 5) that was instrumented and loaded to failure. The
two end frames are doubly-symmetric sections formed by back-
to-back connection of C-sections with knee bracing at the eaves Fig. 4. Single C-section CFS portal frame (Frame test 1).
regions. These frames were reused from a previous test series on
double C-section portal frames (Blum and Rasmussen 2019),
which, by its reuse, not only minimized the cost and reduced was-
tage of material but also facilitated the test setup. These end frames, (less than 1%), the nominal dimensions of the sections are pre-
apart from providing elastic support to the far end of the purlins, sented in Table 2 (BlueScope Lysaght 2014a, b) while the typical
did not participate in any significant way in frame load sharing, cross-section of C-section is shown as inset in Fig. 6.
which were achieved using a unique bolting method employed to The six tests included three tests on frames with connection
connect purlins to the cleats as highlighted in the loading section brackets that are currently used in the BlueScope Lysaght CFS
below. Of the six frames tested, the columns of the first five frames portal frame system (identified as existing system in Table 3) and
were unbraced while in the sixth frame, the columns were re- another three tests on the same nominal frame configuration but
strained by girts at the spacing of 1.25 m along their height. with modifications to the eaves and apex brackets and inclusion of
As the difference between the measured and nominal dimen- column restraints (identified as modified system in Table 3). In the
sions of the C30024 test sections was found to be negligibly small modified joint configurations, the lip of the eaves brackets was

© ASCE 04019187-3 J. Struct. Eng.

J. Struct. Eng., 2020, 146(1): 04019187


Downloaded from ascelibrary.org by University of Central Florida on 11/21/20. Copyright ASCE. For personal use only; all rights reserved.

Fig. 5. Test frame layout.

Fig. 6. Test frame elevation.

extended, the screws were replaced by M8 bolts, and the top flange category of the system, two load types were considered: gravity
of the apex bracket was extended to include a lip. In Frame test 6, load and gravity load in combination with horizontal load.
partial restraints to the columns were provided by girts spanning
between the columns of central frame and the end frames. For each
Frame Connections

Table 2. Nominal dimensions (in millimeters) of the C-sections The columns and rafters of the central frames were connected at the
eaves, apex, and base using brackets specific to the single C-section
Structural members Designation t D B L portal frames. Both the eaves and apex brackets had a nominal
Test frame columns and rafters C30024 2.4 300 96 27.5 thickness of 3.0 mm while the base strap was 6.0 mm thick. As
End frame columns C20019 1.9 203 76 19.0 a result of the relatively light gauge of the brackets and relatively
End frame rafters C20015 1.5 203 76 15.5 small number of fasteners, the connections were semirigid. While
End frame haunch C15015 1.5 152 64 15.5 the M12 bolts with integrated washers used for the connection of
Purlins SC15012 1.2 152 60.5 18.0
purlins to the cleats were tightened using a handheld spanner to

© ASCE 04019187-4 J. Struct. Eng.

J. Struct. Eng., 2020, 146(1): 04019187


Table 3. Portal frame test series The detailed dimensions including the fastener locations can be
Test series Loading Joint configuration found in Rinchen (2018).
Frame test 1 Vertical load Existing system Apex Connection
Frame test 2 Vertical load Existing system The apex connection was formed by bolting the two ends of the
Frame test 3 Vertical load þ5 kN Existing system rafters to the apex bracket using 12 M16 8.8/TF bolts (Fig. 8).
horizontal load
The lower lips of the rafters and the lip of the bracket were fastened
Frame test 4 Vertical load Modified system
Frame test 5 Vertical load þ5 kN Modified system together using two 5.5 mm diameter self-drilling Tek screws. The
horizontal load position of each bolt is identified on the joint with numbers from 1
Frame test 6 Vertical load Modified system with to 12 in Fig. 8. Similarly, the two screws are identified by S1
columns restrained by girts and S2.
For the modified system, the two Tek screws of the apex joint
configuration were replaced by M8 bolts. Besides, the upper flange
Downloaded from ascelibrary.org by University of Central Florida on 11/21/20. Copyright ASCE. For personal use only; all rights reserved.

of the apex bracket was extended to form a 35 mm wide downward


lip to minimize upward bending of the apex bracket. The detailed
achieve the snug-tight condition, the M16 bolts were pretensioned
dimensions of the original and the modified apex bracket along
using the turn-of-nut method in which the bolts were initially tight-
with the apex connection can be found in Rinchen (2018).
ened to the snug-tight condition by using the standard Podger span-
ner with the full effort of a person or by a few impacts of an impact Base Connection
wrench. Upon completing the snug-tight phase, both the nut and Fig. 9 shows the base connection of the single C-section portal
bolts were marked, and while holding the bolt in place, the nut frames. The base connections were formed by bolting the column
was further rotated by 180° with the aid of a high-range impact flanges to the 6.0 mm thick base strap using six M16 8.8/TF bolts.
hammer. This resulted in significantly stiffer connections compared The base strap was clamped to the upper base plate with the aid
to the snug-tight bolting method adopted by many researchers in of a 10 mm thick clamping plate and three M20 8.8/TF bolts. The
the past. upper base plate was connected to four load cells that were cali-
brated to provide reactions (vertical load and major and minor axis
Eaves Connection bending moments) based on the strain measurement recorded
The eave connection was formed by bolting the column and rafter during the tests. Both the base plate and base plate clamp were cut
to the eaves bracket using twelve M16 8.8/TF bolts (Fig. 7). The lip from 300 grade steel. The further details of the base connections are
of the eaves bracket was fastened to the lips of the column and provided in Rinchen (2018).
rafter using four 5.5 mm diameter self-drilling Tek screws. Addi-
tionally, the lips of column and rafter were fastened to the lips of a
zed stiffener using another two Tek screws each. The relative posi- Loading
tion of each bolt of the joint is identified by a number from 1 to 12, The load spreading system consisted of different lengths of metal
as shown in Fig. 7. Similarly, the positions of the Tek screws are straps as vertical hangers and horizontal square hollow steel beams
shown with a number from S1 to S8 in the same figure. The detailed arranged in three layers with the center of the lowest beam con-
dimensions, including the fastener locations, can be found in nected to the loading jack (Fig. 10). The ends of the vertical metal
Rinchen (2018). straps and horizontal beams were bolted together as pin connec-
For the modified system, all eight Tek screws of the eaves joint tions using threaded bolts.
configuration were replaced by M8 bolts in addition to increasing To simulate vertical gravity loading on the central frame, point
the width of the eaves bracket lips by 20 mm to provide room for loads were applied to the purlins on either side of the rafter
the M8 bolts. The bolts are referenced with the same numbering (Fig. 11) such that the resultant vertical load passed through the
convention (S1–S8) as used for screws in the discussions to follow. shear center of the rafter (Fig. 12). The purlin, in turn, transferred

Fig. 7. Eaves connection.

© ASCE 04019187-5 J. Struct. Eng.

J. Struct. Eng., 2020, 146(1): 04019187


Downloaded from ascelibrary.org by University of Central Florida on 11/21/20. Copyright ASCE. For personal use only; all rights reserved.

Fig. 8. Apex connection.

Fig. 9. Base connection.

Fig. 10. Vertical load spreading system.

the load to the web of the rafters of the central frame through the of the eccentric location of the shear center with respect to the
purlin cleats bolted rigidly to its web. Note that the direct appli- centroid.
cation of load onto the rafter would not only cause localized failure To minimize moments developing at the purlin-to-cleat plate
at the load application point but also would twist the rafter because connections, only one of the two bolts was snug-tightened. At the

© ASCE 04019187-6 J. Struct. Eng.

J. Struct. Eng., 2020, 146(1): 04019187


Downloaded from ascelibrary.org by University of Central Florida on 11/21/20. Copyright ASCE. For personal use only; all rights reserved.

Fig. 11. Loading bracket connection: (a) rear side of rafter; and (b) front side of rafter.

Fig. 12. Details of point load application on purlins.

cleats on the end frames, the bottom bolt was snug-tightened with
spanners while the top bolt was finger-tightened. Conversely, at the
central frame, the top bolt was snug-tightened while the bottom bolt
was finger-tightened.
Frame tests 3 and 5 were subjected to a constant horizontal load
of 5 kN in addition to the vertical loads. The 5 kN horizontal load
was kept constant throughout the duration of the tests and was ap-
plied prior to applying vertical loading. The horizontal load was
achieved by hanging a concrete block from the north eaves bracket
using a steel wire (Fig. 13). The steel wire was connected to the
eaves bracket by means of a 130 × 110 mm 2.4 mm thick steel Fig. 13. Arrangement for horizontal load application.
plate attached to the web of the eaves bracket.
To maintain verticality of the loads applied during the tests,
the position of the vertical loading jack was constantly adjusted
by engaging a second horizontal hydraulic jack whose piston was installed at the apex, eaves, mid-height of columns, and at the base
connected to the base of the vertical loading jack (Fig. 14). This of the central frame. Additionally, a transducer was also installed
was made possible as the support system for the vertical jack was at the apex of one of the end frames. While the global displace-
designed to move horizontally on rollers in the plane of the frame ments and rotations were monitored at the column base, column
while it was effectively prevented from moving in the vertical and midheight, at the eaves and at the apex, local deformations were
out-of-plane directions. The horizontal actuator was operated in monitored below the eaves at the locations where stresses were
displacement control mode using the reading from the potentiom- likely to be significant.
eter installed at the apex as input signal. Thereby, the adjustment of
the vertical jack was automatic and ensured the jack stayed truly
Measured Initial Imperfections
vertical throughout the test.
As imperfections may affect the buckling strength of frames,
imperfections for the six columns used for the first three tests were
Instrumentation measured to ascertain the level of deviation of the cross-sections
To monitor the deformation of the frame during the tests, a total of and the overall geometry of the members before assembling into
47 transducers having stroke lengths of 50, 100, and 200 mm were frames. The thickness of the C-sections, the brackets (eaves, apex,

© ASCE 04019187-7 J. Struct. Eng.

J. Struct. Eng., 2020, 146(1): 04019187


stage of loading) and resetting transducers. The frames ultimately
failed by fracture of screw connections at the eaves or local buck-
ling near the compression flange-web junction of the columns and
rafters.

Test Results
As the vertical load was progressively increased, the apex deflected
downward and the two eaves displaced horizontally in the outward
directions. At the same time, the two columns began to twist in
opposite directions. The twisting of the columns commenced at
the onset of loading and continued until the attainment of the
Downloaded from ascelibrary.org by University of Central Florida on 11/21/20. Copyright ASCE. For personal use only; all rights reserved.

ultimate load with major torsional deformations occurring above


midheight. As the loading continued to increase in the nonlinear
range, the top flange of the apex bracket bent upwards and local
buckling deformations began to appear on the web of the columns
below the eaves. Similar local buckling deformation started to ap-
pear within the web of the rafters at the apex region.
In the first three frame tests that had screws connecting the lips
Fig. 14. Load actuator system. of the columns and rafters to the eaves bracket, failure was trig-
gered by the fracture of screws, resulting in the bending of the eaves
bracket. However, in the last three tests, which featured M8 bolts in
lieu of screws connecting the lip of the eaves bracket to the lips of
base straps) and the zed stiffeners were also measured to confirm
the column and rafter, the failure occurred predominantly by local
the base metal thickness of the sections required for numerical
buckling of the columns and rafters near their flange-web junctions.
modelling. Furthermore, global imperfection of frame including
The (typical) failure modes of Frame test 1 are shown in Fig. 15.
out-of-plumbness, height to underside of apex, diagonal distance
and failure by local buckling of members are shown in Fig. 16. The
from eaves to base at the other end, and column twists were mea-
sured for each frame tests before installing the load spreading sys- ultimate loads on the frames are provided in Table 4.
tems. The detailed information of the recorded imperfections can be Although the configuration of Frame test 1 and Frame test 2 was
found in Rinchen (2018). nominally the same, the ultimate capacity of the Frame test 1 was
lower than that of Frame test 2 by 12.8%, which would have been
affected by the different method employed for the installation of
Frame Tests Frame test 1, which was a trial run. Test frame 1 was initially as-
Frames were tested by incrementally applying vertical load with the sembled on the floor with bolts snug-tightened at the eaves and
jack displacement controlled at the rate of 2.0 mm=min using a apex connections. The completed frame was then lifted in position
250 kN capacity hydraulic jack with a stroke length of 250 mm. using two forklifts, one at either end while the overhead crane lifted
Such a slow rate was crucial to ensure the loading was essentially the frame near the apex using a sling strap attached to a sling beam.
static. Apart from Frame test 3, the frames were unloaded when the Once positioned in place, the bolts at the apex, eaves, and base
vertical jack ran out of travel, the vertical load arrangement was connections were fully tensioned using turn-of-nut method while
reset, and the frame subsequently reloaded to obtain the full pre- the frame was still held in position by the overhead crane. However,
and postultimate load-displacement response. One of the key ob- it was realized that this erection procedure was not only impractical
servations made in the single C-section portal frame tests was the as the frame tended to bend and twist when lifted but also required
propensity of the members of the frames to undergo large defor- considerable effort, having to engage two forklifts and substantial
mations prior to reaching the ultimate load. Hence, displacement manpower to aid the installation. The procedure led to relatively
transducers were adjusted wherever required by momentarily paus- large initial twist rotations [Rinchen (2018) for detailed values] that
ing the test at various stages of loading (especially at the advanced may have significantly affected the ultimate load capacity of the

Fig. 15. Failure modes of Frame test 1: (a) bending of eaves bracket; and (b) shearing of screws.

© ASCE 04019187-8 J. Struct. Eng.

J. Struct. Eng., 2020, 146(1): 04019187


Downloaded from ascelibrary.org by University of Central Florida on 11/21/20. Copyright ASCE. For personal use only; all rights reserved.

Fig. 16. Local buckling of members: (a) Column C3 in Frame test 5; (b) Column C1 in Frame test 6; and (c) rafter R3 in Frame test 6.

Table 4. Comparison of ultimate capacities of CFS portal frames


Tests, DSM, DSM, N test N test
Frame N test (kN) N f (kN) N fB (kN) N f N fB
Frame test 1 27.2 32.6 27.4 0.83 0.99
Frame test 2 31.2 32.6 27.4 0.96 1.14
Frame test 3 27.1 29.0 25.6 0.93 1.06
Frame test 4 31.4 32.6 27.4 0.96 1.15
Frame test 5 30.5 29.0 25.6 1.05 1.19
Frame test 6 34.9 38.2 34.3 0.91 1.02
Mean 0.94 1.09
Coefficient of variation 0.08 0.07

frame. This reflects that the method used for the frame installation
plays a crucial role for the structural performance of the frame.
Based on the results of Frame tests 1, 2, and 3, improvements in
the ultimate load capacity of Test frames 4, 5, and 6 were achieved
by the replacement of screws by M8 bolts in the eaves and apex
connections. In these last three tests, failure occurred in the mem-
bers rather than in the connections, although failure occurred near
the connections (Fig. 16) and may have been influenced by stress
concentrations induced by fasteners and/or connection brackets.
However, significant enhancement of strength and reduction of col- Fig. 17. Load versus vertical displacement of apex.
umn twist deformation were observed in Frame test 6 with bracing
of columns installed using girts along their height (Fig. 20). Hence,
to optimize the structural performance, screws may have insuffi-
cient strength and ductility and may be best replaced by bolts, and,
where practically possible, column bracing should be used to re- and 5, the slight offset in the load-displacement graph at a load of
duce twist rotations and provide lateral restraint. 7.13 kN accounts for the marginal increase in vertical displacement
of the apex at the time of application of the 5 kN horizontal load
Apex Deformation at the north eaves. These displacements could be captured as the
Graphs of the total gravity load versus vertical displacement of the transducers were installed on the frame and the data recorders were
apex are shown in Fig. 17. The positive directions of the deforma- activated prior to the installation of load spreading system and
tions are shown in the inset diagrams. The dashed lines in the initial subsequent application of horizontal load.
parts of the plots represent the linear load-displacement relationship The plots are marked by numerous momentary drops in loads
resulting from the weight of the load spreading system and the self- (kinks). The first kink for Frame tests 1, 2, and 3 on the load-
weight of the frame at roof level. In obtaining the initial linear re- displacement curve correspond to the load drops caused by the
sponse, it was assumed that the frame remained elastic under its shearing of apex screws, while the similar kinks in Frame tests 4
self-weight and the weight of the load spreading system, whereby and 6 are caused by the slip of M8 bolts at the apex region. The
the initial part of the load-displacement response measured under subsequent large kinks on the graphs mark the load drop due to
vertical loading was extrapolated to a zero load. For Frame tests 3 pauses in loading executed for adjusting the transducers as several

© ASCE 04019187-9 J. Struct. Eng.

J. Struct. Eng., 2020, 146(1): 04019187


Downloaded from ascelibrary.org by University of Central Florida on 11/21/20. Copyright ASCE. For personal use only; all rights reserved.

Fig. 18. Flexural-torsional deformation of columns: (a) Column C1, and (b) Column C3 from Frame test 1; (c) Column C1, and (d) Column C3 from
Frame test 4.

of them had to be adjusted at different times to record the defor- EB3, the twist deformation of the north column (C3) was larger
mation beyond their stroke length. The large drops in load corre- than that of the south column (C1) due to frame sway.
spond to the complete unloading of the test frames as required Fig. 19 shows the typical displacements of the north and south
when the vertical jack ran out of travel and had to be reset. columns at their midspan. The arrows indicate the positive direc-
The slopes of the load-displacement curves were determined in tions of the displacements and twist rotations. The twist rotations at
between 7.13 and 11.13 kN both for initial loading and reloading the midheight of the two columns are shown in Fig. 20. As shown
portions. For the frames with unbraced columns (Frame tests 1–5), in the figure, twisting of the columns occurred from the onset of
the average slope for initial loading was −0.202 kN=mm while loading and is attributed to bimoments induced by the concentrated
the slope for the reloading portion was −0.203 kN=mm. However, loads transferred by the fasteners at points in the cross-section
for the frame with braced columns (Frame test 6), the initial where the sectorial coordinate is nonzero.
loading slope was −0.238 kN=mm while the reloading slope was Apart from Frame test 6, all frame tests show a similar trend in
−0.256 kN=mm indicating that the frame stiffness had slightly the measured twist rotations. In the case of Frame test 6, due to the
changed due to the restraint imposed by the girts on the columns.

Flexural-Torsional Deformation of Columns


The unique behavior exhibited by all frames except Frame test 6 is
the gradual twisting of both columns in the direction opposite to
each other besides the horizontal displacements (bending about
the minor axis) induced as the load was progressively increased.
Fig. 18 shows the flexural-torsional deformation profile of columns
from Frame test 1 and Frame test 4, representing a frame with
existing joint configuration and modified joint configuration, re-
Fig. 19. Typical displacement of columns at midheight under gravity
spectively. In the case of Frame test 3 and Frame test 5, which
load.
had a constant horizontal load of 5 kN applied at the eaves bracket

© ASCE 04019187-10 J. Struct. Eng.

J. Struct. Eng., 2020, 146(1): 04019187


Downloaded from ascelibrary.org by University of Central Florida on 11/21/20. Copyright ASCE. For personal use only; all rights reserved.

Fig. 20. Applied load versus twist rotation of columns at 2.5 m above the base plate.

presence of girts along the height of the columns, twist rotations are
restrained especially in the initial stage. However, as the loading
progressed, upon slipping of bolts at the ends of the girts, the twist
deformation follows a similar trend to the other frames. The sudden
jump in the plots at 7.13 kN for Frame test 3 and Frame test 5
represent the instantaneous change of twist rotations induced by
the application of the 5 kN horizontal load at the north eaves
bracket.
The columns, besides twisting, also deflected both in the in-
plane and out-of-plane directions. The in-plane flexural deforma-
tion was induced by the compressive force in the inner flange of the
columns arising from the moment in the eaves region. Conversely,
the out-of-plane displacements are due to the axial load acting
through the web instead of acting through the centroid at the eaves
bracket.

Local Buckling Deformation of Columns


As the loading progressed into the nonlinear range, local buckling
deformations started to appear in the web of the columns below the
eaves. With increasing load, local buckling deformations extended
downward toward the midheight of the column (Fig. 21). Local
buckles also appeared in the rafters near the apex upon further
loading.
The magnitude of local buckling deformation was higher near
the inner flange-web junction in the columns and at the top flange-
web junction in the rafters as a result of major axis bending. A grid
marked on the south column (C1) of Frame test 2 below the
eaves indicated that the local buckle half-wavelength was approx-
imately 300 mm.
The local displacements and flange rotations were measured Fig. 21. Local buckling deformation of Columns C1 and C3 in Frame
250 mm below the eaves bracket on each column, using 10 trans- test 2.
ducers. The typical local deformation profiles of the columns are
shown in Fig. 22 where the positive directions of local deforma-
tion of flanges and webs (for the advanced stage of loading) are
indicated by the arrows. Figs. 23–25 show the local deformation Frame test 4 where the rotation is between 1.5° and 2.5°. However,
of the flanges and web of columns measured 250 mm below the the rotations of the inside flanges are generally larger than the
eaves. rotations of the outside flanges due to the internal major axis bend-
The rotation of the outside flange of columns is observed to be ing moment and the restraint of inside lips of columns to the eaves
generally less than 1.5° with the exception of Frame test 1 and brackets by screws or M8 bolts.

© ASCE 04019187-11 J. Struct. Eng.

J. Struct. Eng., 2020, 146(1): 04019187


Fig. 22. Typical local and global deformation profiles of columns below the eaves.
Downloaded from ascelibrary.org by University of Central Florida on 11/21/20. Copyright ASCE. For personal use only; all rights reserved.

Fig. 23. Applied load versus rotation of outside flange (OF) of columns at 250 mm below the eaves bracket.

Fig. 24. Applied load versus rotation of inside flange (IF) of columns at 250 mm below the eaves bracket.

© ASCE 04019187-12 J. Struct. Eng.

J. Struct. Eng., 2020, 146(1): 04019187


Downloaded from ascelibrary.org by University of Central Florida on 11/21/20. Copyright ASCE. For personal use only; all rights reserved.

Fig. 25. Applied load versus local displacement of web of columns at 250 mm below the eaves bracket.

Design Provisions where ϕc and ϕb are capacity reduction factors for compression and
bending; N c is the member compression capacity; M b is member
CFS portal frames are usually designed based on conventional de- moment capacity about the major axis; and N  , M  are second-
sign methods using either the DSM or the Effective Width Method order design axial compression force and design bending moment
(EWM) with the former mostly preferred to the latter owing to about the major axis; respectively. As both the section geometric
its convenience. The DSM uses software such as THIN-WALL properties and material properties were based on the actual tests,
(Papangelis and Hancock 1995) or CUFSM (Schafer and Ádány ϕc and ϕb were taken as unity. Note that Eq. (1) does not include
2006) to determine the local buckling axial load and bending mo- the minor axis bending moment obtained from the second order
ment, the distortional buckling axial load and bending moment, the analysis as this was negligible.
member buckling axial load and bending moment, and the shear Table 4 shows the ultimate capacities of the single C-section
local buckling load. portal frames predicted by the conventional design method using
Member capacities of columns in compression and bending the DSM (Nf ) and the ultimate loads obtained from the tests (N test ).
were obtained based on the DSM capacity check in accordance As observed in Table 4, the capacities predicted by the DSM (N f )
with AS (2018). For this, the measured values of material properties in conjunction with a linear interaction equation [Eq. (1)] are gen-
and thickness were used. Frame models were created in MASTAN2 erally slightly higher than the test values, implying that the frame
(Ziemian and McGuire 2018) and second order elastic analyses capacity may be overestimated by the DSM. The DSM was devel-
were performed with gravity loads and gravity plus lateral (wind) oped for the design of members subjected to linear stress distribu-
loads to obtain the design forces and moments in the columns and tions in the section resulting from axial load or bending moment as
rafters. Frame imperfections (out-of-plumbness and twist rotation) per classical theory of structural mechanics. It requires the deter-
were directly included in the structural model by updating the mination of member capacities in compression and bending inde-
nominal frame geometry to the measured imperfect geometry. The pendently of each other, and accounts for interaction through a
semirigidity of the joints was accounted for by including the initial linear relation [Eq. (1)]. On average, for CFS portal frames, this
flexural stiffness of the respective connections derived from the procedure leads to 6% higher strength predictions compared to the
separate connection tests (Rinchen 2018). As the replacement of actual strength.
5.5 mm diameter Tek screws by 8 mm diameter bolts had negligible This difference is attributed to the presence of a bimoment in
influence on the initial flexural stiffness of the joint, the frame mod- the column at the eaves connection, which is not accounted for in
els were grouped into three: Frame B1 representing the frames with the DSM (Lim et al. 2016). This bimoment arises from the concen-
gravity loads (corresponding to Frame tests 1, 2, and 4), Frame B2 trated loads transferred by the fasteners at points in the cross-
representing the frames with combination of lateral load and grav- section where the sectorial coordinate is nonzero. For this case, the
ity load (corresponding to Frame tests 3 and 5), and Frame B3 rep- bimoment is given by Vlasov (1961)
resenting the frame with gravity load and columns braced in the
X
n
out-of-plane direction (corresponding to Frame test 6). B¼ Fli ωi ð2Þ
The capacities of the frames were determined by satisfying the i¼1
following interaction equation:
where Fli is the vertical (longitudinal) force component of bolt
shear Fi, transferred at the ith fastener and ωi is the sectorial co-
N M
þ ¼ 1.0 ð1Þ ordinate at the corresponding point in the cross-section. Assuming
ϕc N c ϕb M b the bolt group behaves elastically, the bolt shear Fi is given by

© ASCE 04019187-13 J. Struct. Eng.

J. Struct. Eng., 2020, 146(1): 04019187


Downloaded from ascelibrary.org by University of Central Florida on 11/21/20. Copyright ASCE. For personal use only; all rights reserved.

Fig. 26. Signature curves of buckling stress and moments for C-sections: (a) buckling stress; and (b) buckling moment.

Mr is 6% higher than the average test strengths. However, with the


Fi ¼ Pn i 2
ð3Þ incorporation of bimoment, the DSM predictions result in more
i¼1 ri
conservative design with average test strengths 9% higher than the
where M is the major axis moment and ri is the radial distance of design strength predictions.
ith bolt from the instantaneous center of rotation (ICR). The con- As is evident from the test results, an increase of 11.1% in
tribution from the axial force (N) can be ignored without introduc- ultimate load was observed in Frame test 6 (frame with braced
ing significant error as the resulting distribution of bolt forces leads columns) when compared to Frame test 4 (frame with unbraced
to a substantially smaller contribution to the bimoment than the column but of similar setup to Frame test 6) mainly due to the re-
major axis moment when calculated according to Eq. (2). duction in flexural-torsional deformation of columns owing to the
Alternatively, the bimoment can be simplified to (Hancock presence of girts along their heights. In general, the flexural-
1985) torsional deformation of columns can be minimized, and hence the
strength and stiffness of the CFS portal frames can be enhanced,
B ¼ eM ð4Þ if the columns are joined back-to-back or by providing cladding
where stressed-skin action can be taken advantage of. In fact, it was
where e is the distance from the shear center to the web centerline. reported that (Wrzesien et al. 2015) in the CFS portal frames
The use of this equation implicitly assumes that (1) the bolts on the comprising channel sections joined back-to-back, the frame with
flange are located close to zero sectorial coordinates and (2) screws stressed-skin action not only carried approximately three times
on the lips shear off before attaining the ultimate load thereby re- more horizontal load than the bare frame but also drastically re-
lieving stresses at their locations, so that only the bolts located in duced the deflection (by 90%) relative to the bare frame. Such dras-
the web contribute to the bimoment. As a result of the bimoment, tic effect, however, may not be achievable for single C-section CFS
both the compression and tension flanges of the C-section experi- frames owing to their inherent propensity to undergo flexural-
ence nonuniform stress distributions. torsional deformation as observed in the tests. Nonetheless, stressed
The effect of the bimoment may be accounted for by including skin action is an important structural effect that can be accounted
both the stress distributions induced by the bimoment and the major for when designing CFS portal buildings with claddings.
axis moment in determining the major axis local buckling moment
(Mol ) and distortional buckling moment (M od ), and substituting
these values of moment into the DSM equation in calculating the Conclusions
major axis bending strength (Lim et al. 2016). Fig. 26 shows the
influence of bimoment on the local and distortional buckling Six full-scale tests on 13.6 m span by 6.8 m high CFS single
stresses and buckling moments. Note that the stress labelled C-section portal frames and accompanying coupon tests have been
“Moment + Bimoment” in Fig. 26(a) is the total stress produced described.
by the combination of major axis moment and bimoment, and that The tests demonstrated that torsion of columns, followed by
under this combined stress distribution, the buckling curve only flexural-torsional buckling, is the dominant structural behavior
exhibits a minimum for the local buckling mode. While the mini- mainly arising from the eccentric location of shear center and cent-
mum local buckling stress is marginally higher than that for major roid of the cross-section. Flexural-torsional buckling deformations
axis moment only, the corresponding major axis local buckling mo- of the columns were exhibited by all frames especially at higher
ment is significantly smaller than when only a major axis moment loads regardless of the presence of lateral restraints, although a
is applied as shown in Fig. 26(b). This reduction in local buckling stiffness enhancement and much smaller twist rotations were ob-
moment affects the design capacity. served when restraints (girts) were provided to the columns. The
The design capacity calculation using DSM is repeated using ultimate failure deformations were concentrated at the eaves and
the values of “Moment + Bimoment” local buckling moment the apex and were governed by the type of fasteners used for the
shown in Fig. 26. The ultimate capacities (N fB ) are provided in connection of the column lips to the connection brackets. In the first
Table 4. As observed in Table 4, the ultimate capacities of frames three frame tests, the shearing of screws at the eaves triggered the
predicted by the DSM without considering the effect of bimoment failure of the frames whereas in the last three frame tests, the

© ASCE 04019187-14 J. Struct. Eng.

J. Struct. Eng., 2020, 146(1): 04019187


failures were associated with local buckling and the formation of Dundu, M. 2011. “Design approach of cold-formed steel portal frames.”
spatial plastic mechanisms below the eaves or near the apex. The Int. J. Steel Struct. 11 (3): 259–273. https://doi.org/10.1007/s13296
strengths of the frames with modified joint configuration were -011-3002-2.
slightly higher than those of the frames with the existing propri- Dundu, M., and A. R. Kemp. 2006. “Strength requirements of single cold-
etary joint configuration, whereas a significant increase in strength formed channels connected back-to-back.” J. Constr. Steel Res. 62 (3):
250–261. https://doi.org/10.1016/j.jcsr.2005.07.006.
was observed in the last frame test where columns were braced in
Elkersh, I. 2010. “Experimental investigation of bolted cold formed steel
the out-of-plane direction with girts. To optimize the structural
frame apex connections under pure moment.” Ain Shams Eng. J. 1 (1):
performance, bolts should be used in lieu of screws in fastening 11–20. https://doi.org/10.1016/j.asej.2010.09.002.
members to connection brackets, and where practically possible, Hancock, G. J. 1985. “Portal frames composed of cold-formed channel
column bracing should be used to reduce twist rotations and lateral and zed-sections.” In Steel framed structures-stability and strength,
displacements of columns. Furthermore, due to the use of preten- edited by R. Narayanan 241–275. London: Elsevier Applied Science
sioned bolts for the main frame connections, bolt hole elongation Publishers.
was minimal or nonexistent. This indicates that pretensioning of Huang, Y., and B. Young. 2014. “The art of coupon tests.” J. Constr.
Downloaded from ascelibrary.org by University of Central Florida on 11/21/20. Copyright ASCE. For personal use only; all rights reserved.

bolts can potentially improve frame performance. Steel Res. 96 (May): 159–175. https://doi.org/10.1016/j.jcsr.2014.01
The ultimate capacities of the frames were determined based on .010.
the conventional design method using DSM and compared with Jackson, C., A. M. Wrzesien, R. P. Johnston, A. Uzzaman, and J. B. P. Lim.
ultimate loads obtained from the tests. The capacities predicted by 2012. “Effect of reduced joint strength and semi-rigid joints on cold-
the DSM were 6% higher than the test strengths on average. The formed steel portal frames.” In Proc., 6th Int. Conf. on Coupled
difference is attributed to the presence of a bimoment induced at the Instabilities in Metal Structures. Glasgow, UK: Strathclyde Univ.
Kirk, P. 1986. “Design of a cold formed section portal frame building
connections in the test frames indicating that the bimoment has
system.” In Proc., 8th Int. Specialty Conf. on Cold-Formed Steel
significant influence on the capacity of the CFS portal frames. The Structures. Rolla, MO: Univ. of Missouri.
bimoment produces a nonuniform stress distribution in the flanges Kwon, Y., H. Chung, and G. Kim. 2006. “Experiments of cold-formed steel
of the C-section columns, while the DSM assumes a uniform stress connections and portal frames.” J. Struct. Eng. 132 (4): 600–607.
distribution in the flanges under the action of a major axis bending https://doi.org/10.1061/(ASCE)0733-9445(2006)132:4(600).
moment. By including the bimoment in the calculation of the Lim, J. B., G. J. Hancock, G. C. Clifton, C. H. Pham, and R. Das.
local and distortional buckling moments, the DSM-based design 2016. “DSM for ultimate strength of bolted moment-connections
procedure results in a conservative design approach with design between cold-formed steel channel members.” J. Constr. Steel Res.
strengths 9% lower than the test strengths on average. 117 (Feb): 196–203. https://doi.org/10.1016/j.jcsr.2015.10.005.
Lim, J. B. P., and D. A. Nethercot. 2002. “Design and development of a
general cold-formed steel portal framing system.” Struct. Eng. 80 (21):
31–40.
Acknowledgments
Lim, J. B. P., and D. A. Nethercot. 2004. “Finite element idealization of a
This research was supported by the Australian Research Council’s cold-formed steel portal frame.” J. Struct. Eng. 130 (1): 78–94. https://
doi.org/10.1061/(ASCE)0733-9445(2004)130:1(78).
Linkage Project funding scheme (Project No. LP120200528). The
Mills, J., and R. LaBoube. 2004. “Self-drilling screw joints for cold-formed
ARC support is gratefully acknowledged. All materials were sup-
channel portal frames.” J. Struct. Eng. 130 (11): 1799–1806. https://doi
plied by BlueScope Lysaght. .org/10.1061/(ASCE)0733-9445(2004)130:11(1799).
Mojtabaei, S. M., M. Z. Kabir, I. Hajirasouliha, and M. Kargar. 2018.
“Analytical and experimental study on the seismic performance of
References cold-formed steel frames.” J. Constr. Steel Res. 143 (Apr): 18–31.
https://doi.org/10.1016/j.jcsr.2017.12.013.
AS (Standards Australia). 2007. Metallic materials-tensile testing at Papangelis, J. P., and G. J. Hancock. 1995. “Computer analysis of thin-
ambient temperature. AS 1391. Sydney, Australia: AS. walled structural members.” Comput. Struct. 56 (1): 157–176. https://
AS (Standards Australia). 2018. Cold-formed steel structures. AS doi.org/10.1016/0045-7949(94)00545-E.
NZS4600. Sydney, Australia: AS. Rhodes, J., and R. Burns. 2006. “Development of a portal frame system on
Baigent, A. H., and G. J. Hancock. 1982. “The strength of cold-formed the basis of component testing.” In Proc., 18th Int. Specialty Conf. on
portal frames.” In Proc., 6th Int. Specialty Conf. on Cold-Formed Steel Cold-Formed Steel Structures. Rolla, MO: Univ. of Missouri.
Structures. Rolla, MO: Univ. of Missouri. Rinchen, R. 2018. “Long span cold-formed steel single C-section portal
BlueScope Lysaght. 2014a. SupaPurlins -Quick selection guide for open frames.” Ph.D. thesis, School of Civil Engineering, Univ. of Sydney.
carports in wind regions N2-N4. Minchinbury, NSW: BlueScope Sabbagh, A. B., R. Mirghaderi, M. Petkovski, and K. Pilakoutas. 2010.
Lysaght. “An integrated thin-walled steel skeleton structure (two full scale tests).”
BlueScope Lysaght. 2014b. Zeds and cees: User guide for design and J. Constr. Steel Res. 66 (3): 470–479. https://doi.org/10.1016/j.jcsr
installation professionals: Part I. Minchinbury, NSW: BlueScope .2009.10.007.
Lysaght. Sabbagh, A. B., M. Petkovski, K. Pilakoutas, and R. Mirghaderi. 2012.
Blum, H. B., and K. J. R. Rasmussen. 2019. “Experimental investigation of “Experimental work on cold-formed steel elements for earthquake
long-span cold-formed steel double-channel portal frames.” J. Constr. resilient moment frame buildings.” Eng. Struct. 42 (Sep): 371–386.
Steel Res. 155 (Apr): 316–330. https://doi.org/10.1016/j.jcsr.2018.11 https://doi.org/10.1016/j.engstruct.2012.04.025.
.020. Schafer, B. W., and S. Ádány. 2006. “Buckling analysis of cold-formed
Chung, K. F., and L. Lau. 1999. “Experimental investigation on bolted steel members using CUFSM: Conventional and constrained finite strip
moment connections among cold formed steel members.” Eng. Struct. methods.” In Proc., 18th Int. Specialty Conf. on Cold-Formed Steel
21 (10): 898–911. https://doi.org/10.1016/S0141-0296(98)00043-1. Structures. Rolla, MO: Univ. of Missouri.
Dubina, D., A. Stratan, A. Ciutina, L. Fulop, and Z. Nagy. 2004. Stratan, A., Z. Nagy, and D. Dubina. 2006. “Cold-formed steel pitched-roof
“Performance of ridge and eaves joints in cold-formed steel portal portal frames of back-to-back plain channel sections and bolted joints.”
frames.” In Proc., 17th Int. Specialty Conf. on Cold-Formed Steel In Proc., 18th Int. Specialty Conf. on Cold-Formed Steel Structures.
Structures. Rolla, MO: Univ. of Missouri. Rolla, MO: Univ. of Missouri.
Dunai, L. 2007. “Innovative steel and composite structures.” Ph.D. thesis, Vlasov, V. Z. 1961. Thin-walled elastic beams. Jerusalem, Israel: Israel
Faculty of Civil Engineering, Hungarian Academy of Sciences. Program for Scientific Translations.

© ASCE 04019187-15 J. Struct. Eng.

J. Struct. Eng., 2020, 146(1): 04019187


Wrzesien, A., J. B. Lim, Y. Xu, I. MacLeod, and R. Lawson. 2015. “Effect resistant frames.” Thin Walled Struct. https://doi.org/10.1016/j.tws
of stressed skin action on the behaviour of cold-formed steel portal .2018.12.015.
frames.” Eng. Struct. 105 (Dec): 123–136. https://doi.org/10.1016/j Yu, W. K., K. F. Chung, and M. F. Wong. 2005. “Analysis of bolted moment
.engstruct.2015.09.026. connections in cold-formed steel beam–column sub-frames.” J. Constr.
Wrzesien, A. M., D. T. Phan, J. B. Lim, H.-H. Lau, I. Hajirasouliha, and Steel Res. 61 (9): 1332–1352. https://doi.org/10.1016/j.jcsr.2005.03
C. S. Tan. 2016. “Effect of stressed-skin action on optimal design .001.
of cold-formed steel square and rectangular-shaped portal frame build- Zhang, X., and K. Rasmussen. 2014. “Tests of cold-formed steel portal
ings.” Int. J. Steel Struct. 16 (2): 299–307. https://doi.org/10.1007 frames with slender sections.” Steel Constr. 7 (3): 199–203. https://doi
/s13296-016-6004-2. .org/10.1002/stco.201410030.
Ye, J., S. M. Mojtabaei, I. Hajirasouliha, and K. Pilakoutas. 2019. “Efficient Ziemian, R., and W. McGuire. 2018. “MASTAN2 v3.5.” Accessed March
design of cold-formed steel bolted-moment connections for earthquake 20, 2018. http://www.mastan2.com.
Downloaded from ascelibrary.org by University of Central Florida on 11/21/20. Copyright ASCE. For personal use only; all rights reserved.

© ASCE 04019187-16 J. Struct. Eng.

J. Struct. Eng., 2020, 146(1): 04019187

You might also like