Thin-Walled Structures 44 (2006) 197–210

www.elsevier.com/locate/tws

Experimental testing of joints for seismic design

of lightweight structures. Part 1. Screwed joints in straps*
Miquel Casafont a, Alfredo Arnedo b, Francesc Roure a, Antonio Rodrı́guez-Ferran c,*
a
Departament de Resistència de Materials i Estructures a l’Enginyeria, E.T.S. d’Enginyeria Industrial de Barcelona,
Universitat Politècnica de Catalunya, Av. Diagonal 647, 08028 Barcelona, Spain
b
Departament d’Enginyeria de la Construcció, E.T.S. d’Enginyers de Camins, Canals i Ports de Barcelona,
Universitat Politècnica de Catalunya, Jordi Girona 1, 08034 Barcelona, Spain
c
Laboratori de Càlcul Numèric (LaCàN), Departament de Matemàtica Aplicada III, E.T.S. d’Enginyers de Camins, Canals i Ports de Barcelona,
Universitat Politècnica de Catalunya, Edifici C2, Campus Nord UPC, Jordi Girona 1, 08034 Barcelona, Spain
Received 8 July 2005; received in revised form 9 January 2006; accepted 24 January 2006

Abstract

Connections in x-braced shear walls play a crucial role in the seismic performance of lightweight structures: they should be strong enough to
allow energy dissipation via plastification of the diagonal straps. An extensive experimental research on tensile screwed joints between straps is
reported here. Two dominant failure modes are identified: (1) tilting and net section failure (TCNSF) and (2) tilting, bearing and pull out (TCBC
PO). The analysis in terms of strength and ductility shows that TCNSF joints are suitable for seismic design, because the straps yield before the
connections fail. TCBCPO joints, on the contrary, are not suited because the connection fails before the straps can yield. The influence of various
design parameters (strap thickness, number and diameter of screws, steel grade) in the failure mode is studied, and design criteria to induce a TC
NSF response are given.
q 2006 Elsevier Ltd. All rights reserved.

Keywords: Light gauge steel; Lightweight steel; Seismic design; Joints; Experimental research; Ductility; Failure modes; Net section failure; Bearing

1. Introduction open new markets. This guidance must take into account that
mass and stiffness are quite different from conventional steel
Lightweight steel structures offer an interesting alternative structures.
to more traditional construction technologies, especially for For any type of structure, two basic seismic requirements
one or two-storey residential and office buildings. are (1) enough lateral stiffness to resist seismic loads without
The design of lightweight structures is yet not covered in significant structural damage and (2) energy dissipation
detail by national or international (e.g. European) regulations. mechanisms via plastic deformation. For lightweight struc-
While elastic design of such structures is nowadays standard, tures, both requirements can be met by means of x-braced shear
there is less knowledge about their performance under walls. Although cladding may have some structural effect [1,2],
exceptional loads (e.g. seismic response, fire resistance) or it is common to neglect its contribution [3] and rely only on the
regarding comfort (e.g. acoustic performance, floor vibrations). steel structure. This leads to simple, safe and easy-to-build
This paper deals with seismic response of lightweight steel
framed buildings, consisting of standard frames and a few
structures. Simple design guides for engineers (including
x-braced frames for seismic reinforcement.
methods of analysis and rules about detailing) are needed to
With such an approach, the structural response of the whole
building is controlled by its x-braced frames, and the joints
*
Research supported by the Research Fund for Coal and Steel (grant 7210- play a crucial role. The connections of the x-brace to the frame
PR-377) and the “Ministerio de Educación y Ciencia” (grant DPI2002-12388-
E).
must provide enough strength to allow dissipative effect (i.e.
* Corresponding author. Tel.: C34 93 401 56 97; fax: C34 93 401 18 25. plastification) of diagonal straps, resulting in an overall ductile
E-mail address: [email protected] (A. Rodrı́guez-Ferran). behaviour. A common design practice is to use straps of
URL: http://www-lacan.upc.edu (A. Rodrı́guez-Ferran). constant width for x-bracing. This poses some constraints in the
0263-8231/$ - see front matter q 2006 Elsevier Ltd. All rights reserved. design of the joint, because the screws cause a reduction in the
doi:10.1016/j.tws.2006.01.002 net section of the strap.
198 M. Casafont et al. / Thin-Walled Structures 44 (2006) 197–210

Fasteners working as shear connections are the most

effective mechanism of force transfer due to their good
behaviour in terms of stiffness, strength and deformation
capacity [4]. Several different types of fasteners can be used.
On the one hand, traditional connecting systems are available,
such as rivets, bolts, screws and pins. On the other hand, a
number of new connecting techniques are being developed:
new types of rivets [5,6] and clinching systems [5–10]. Self-
drilling screws exhibit a good behaviour, both in terms of
strength and ductility [6,8].
In the framework of the RFCS research project ‘Seismic
Fig. 2. Specimen ready to be tested.
Design of Light Gauge Steel Framed Buildings’, we have
carried out a testing campaign of joints in x-braced frames. The An outline of paper follows. The laboratory experiments are
first phase of this campaign consists of lap joints between two described in Section 2 (Test specimens) and Section 3 (test
steel straps joined by means of two rows of screws, see Fig. 1. procedure), and the results are summarized in Section 4
Both monotonic and cyclic tests are carried out. (monotonic tensile tests) and Section 5 (load–unload tensile
The goals of the experiments are to: tests). Three main features are studied: the modes of failure, the
force–displacement curves, and the ductility and stiffness of the
1. obtain parameters such as the initial stiffness, yielding load, connections. The remainder of the paper is devoted to the
ultimate load and maximum displacement, analysis of the results. Some rules for joint design are discussed
2. obtain complete force–displacement (F–d) curves, needed in Section 6. Then, in Section 7, the ultimate loads are
for the finite element modelling of x-braced frames [11], compared to the values predicted by the Eurocode. The
3. identify the various failure modes. concluding remarks of Section 8 close the paper.

After that, we analyze the experimental results in order to:

2. Test specimens
1. classify the various failure modes in terms of their seismic
suitability (strength and ductility), Tests were performed on screwed connections between two
2. determine the relation between parameters in joint design straps (Fig. 1), whose steel grade was either S350 GDCZ or
(steel grade, strap thicknesses, number and diameter of S250 GDCZ [12]. Table 1 shows the nominal and measured
screws,.) and failure mode, mechanical properties of the steels. It should be noticed that the
3. compare experimental ultimate loads of the joints to the experimental fyt and fut are rather higher than the nominal fy and
strengths calculated by means of the Eurocode 3 Part 1.3 fu. These experimentally measured values are used in the last
design formulas. section of the paper, where the strength of the joints is
calculated by means of analytical formulas.
Screws of two different diameters were used to connect the
straps: 4.8 and 6.3 mm. The shaft length of the 4.8 mm
diameter screws was always 10 mm (threaded part), and they
could have either flat and square heads or hexagonal heads. The
shaft of the 6.3 diameter screws, whose head was always
hexagonal, could be 10 or 30 mm long (threaded part).
The length of the steel straps ranged between 350 and
475 mm, depending on the number of screws of the joint. Their
Table 1
Steel mechanical properties

Steel fy fu t (mm) fyt fut

(N/mm2) (N/mm2) (N/mm2) (N/mm2)
S350 350 420 1 392 520
GDCZ
1.5 387 519
3 385 512
S250 250 330 0.85 285 345
GDCZ
1 303 393
1.5 317 391

fy, Nominal yield stress; fu, nominal ultimate stress; t, nominal thickness; fyt,
Fig. 1. Screwed joint. measured yield stress; fut, measured ultimate stress.
 M. Casafont et al. / Thin-Walled Structures 44 (2006) 197–210 199

shows the joint notation, whose meaning is explained in the

Column
following example
t1 Kt2 KFKncKsgKI
25

Row

1K1:5K6:3K6KS350KE

100
Row
t1, thickness of the first strap (t1Z1 mm); t2, thickness of the
second strap (t2Z1.5 mm); F, diameter of the screw (FZ
25

25 25 25 25 25 25
6.3 mm), nc, number of screw columns (ncZ6), sg, steel grade
(S350 GDCZ), I: letter used when there are two or more
Fig. 3. Joint layout. identical joints (E).

thickness was also variable from 0.85 to 3 mm, but their width
was always the same, 100 mm. 3. Test procedure
Fig. 3 shows the position of the screws: the spacing and the
longitudinal and transverse edge distances. The joint layout The first operation was to measure the actual dimensions of
was identical for all the specimens. the joint components (see some of the measured values in
The screwed connections are listed in Tables 2a and 2b Tables 2a and 2b and the full collection of measurements in
together with the test results. The first column of the table [13]). Afterwards, the specimens were labelled and a line was
Table 2a
Results of monotonic tests on joints made of steel grade S350 GDCZ

Connection t1t t2t a1t a2t Failure mode Put (N)

1-1-6.3-6-S350-A 0.99 0.99 100.2 100.1 TCNSF 40,970
1-1-6.3-6-S350-B 0.98 0.98 100.1 100.1 TCNSF 41,246
1-1-6.3-6-S350-C 0.98 0.99 100.2 100.2 TCNSF 40,843
1-1-6.3-6-S350-D 0.99 0.99 100.1 100.2 TCNSF 40,494
1-1-6.3-6-S350-E 0.99 0.98 100.2 100.1 TCNSF 41,157
1-1.5-6.3-6-S350 0.99 1.49 100.2 100.2 NSF 42,735
1.5-1.5-6.3-6-S350 1.50 1.49 100.1 100.1 (T)CNSF 65,295
1.5-3-6.3-6-S350 1.50 3.00 100.7 100.1 NSF 66,592
1-1-4.8-6-S350-A 0.98 0.98 100.1 100.1 TCNSF 40,179
1-1-4.8-6-S350-B 0.98 0.99 100.1 100.1 TCNSF 39,890
1-3-4.8-6-S350-A 0.98 3.01 100.2 100.0 (B)CNSF 42,696
1-3-4.8-6-S350-B 0.99 2.99 100.1 100.4 (B)CNSF 43,265
1.5-1.5-4.8-6-S350 1.49 1.49 100.1 100.1 TCNSF 64,472
1-1-6.3-4-S350-A 0.98 0.98 100.1 100.1 TCPO 32,793
1-1-6.3-4-S350-B 0.99 0.98 100.2 100.1 TCPO 32,306
1.5-1.5-6.3-4-S350 1.49 1.49 100.1 100.1 TCPO 57,745
1.5-3-6.3-4-S350-A 1.49 3.00 100.0 100.1 (B)CNSF 66,067
1.5-3-6.3-4-S350-B 1.49 2.99 100.4 100.1 (B)CNSF 66,528
1-1-4.8-4-S350-A 0.99 0.99 100.1 100.1 TC(B)CPO 29,418
1-1-4.8-4-S350-B 0.99 0.99 100.1 100.1 TCPO 29,802
1-3-4.8-4-S350 0.99 2.99 100.4 100.1 (T)CBCNSF 42,614
1.5-1.5-4.8-4-S350 1.49 1.49 100.2 100.1 TCPO 52,125
1-1-6.3-3-S350-A 0.98 0.98 100.1 100.6 TCPO 25,070
1-1-6.3-3-S350-B 0.98 0.98 100.2 100.1 TCPO 25,165
1.5-3-6.3-3-S350 1.49 3.00 100.0 100.1 TC(B)C NSF 64,360
1-1-4.8-3-S350-A 0.99 0.99 100.1 100.1 TCPO 24,131
1-1-4.8-3-S350-B 0.99 0.98 100.1 100.1 TCPO 22,090
1-1.5-4.8-3-S350 0.99 1.49 100.1 100.1 TCBCPO 36,417
1-3-4.8-3-S350 0.98 2.99 99.8 100.2 TC(B)CPT 36,792
1.5-1.5-4.8-3-S350 1.49 1.49 101.8 100.1 TCPO 38,438
1-1-6.3-1-S350-A 0.99 0.99 100.2 100.2 TCPO 8025
1-1-6.3-1-S350-B 0.99 0.99 100.0 100.0 TCPO 8212
1-1.5-6.3-1-S350 0.99 1.50 100.1 100.1 TCPO 14,880
1.5-1.5-6.3-1-S350 1.50 1.50 100.0 100.1 TCPO 15,089
1.5-3-6.3-1-S350-A 1.49 2.99 100.1 100.5 TC(B)CPO 31,172
1.5-3-6.3-1-S350-B 1.49 2.99 100.1 100.1 TC(B)CPO 30,405
1-1-4.8-1-S350-A 1.00 0.99 100.1 100.1 TCPO 7647
1-1-4.8-1-S350-B 0.98 0.99 100.1 100.1 TC(B)CPO 7580
1-1.5-4.8-1-S350 0.99 1.50 100.0 100.1 TC(B)CPO 13,720
1.5-1.5-4.8-1-S350 1.50 1.50 100.1 100.1 TCPO 13,495
200 M. Casafont et al. / Thin-Walled Structures 44 (2006) 197–210

Table 2b
Results of monotonic tests on joints made of steel grade S250 GDCZ

Connection t1t t2t a1t a2t Failure mode Put (N)

0.85-1-6.3-6-S250 0.89 1.04 100.0 100.0 (T)CNSF 27,040
0.85-1.5-6.3-6-S250 0.89 1.60 100.0 100.0 NSF 26,311
1-1-6.3-6-S250-A 1.04 1.04 100.0 100.0 (T)CNSF 34,258
1-1-6.3-6-S250-B 1.04 1.04 100.0 100.0 TCNSF 34,064
0.85-0.85-4.8-6-S250 0.85 0.85 100.0 100.0 TCNSF 25,126
0.85-1-4.8-6-S250 0.85 1.00 100.0 100.0 (T)CNSF 27,587
0.85-1.5-4.8-6-S250 0.85 1.50 100.0 100.0 (T)C(B)CNSF 27,268
1-1-4.8-6-S250-A 1.00 1.00 100.0 100.0 TC(B)CNSF 32,642
1-1-4.8-6-S250-B 1.00 1.00 100.0 100.0 TCNSF 34,775
0.85-0.85-6.3-4-S250 0.88 0.88 100.0 100.0 TCPO 21,092
0.85-1-6.3-4-S250 0.89 1.04 100.0 100.0 TCNSF 25,660
0.85-1.5-6.3-4-S250 0.88 1.61 100.0 100.0 NSF 25,648
1-1-6.3-4-S350-A 1.04 1.05 100.0 100.0 TCPO 27,617
1-1-6.3-4-S350-B 1.05 1.05 100.0 100.0 TCPO 26,005
0.85-0.85-4.8-4-S250 0.88 0.88 99.9 99.9 TCPO 21,795
0.85-1-4.8-4-S250 0.88 1.04 100.0 100.0 TCPO 25,105
0.85-1.5-4.8-4-S250 0.88 1.59 100.0 100.0 TCBCNSF 28,173
1-1-4.8-4-S250-A 1.03 1.04 100.0 100.0 TCPO 26,540
1-1-4.8-4-S250-B 1.04 1.04 100.0 100.0 TCPO 25,941
0.85-1-6.3-3-S250 0.87 1.04 100.0 100.0 TCPO 20,006
0.85-1.5-6.3-3-S250 0.88 1.62 100.0 100.0 TCBCNSF 26,734
1-1-6.3-3-S250-A 1.03 1.03 100.0 100.0 TCPO 19,804
1-1-6.3-3-S250-B 1.04 1.04 100.0 100.0 TCPO 19,950
0.85-1-4.8-3-S250 0.87 1.04 100.0 100.0 TCPO 20,579
0.85-1.5-4.8-3-S250 0.88 1.60 100.0 100.0 TCBCNSF 24,936
1-1-4.8-3-S250-A 1.03 1.03 100.0 100.0 TCPO 20,868
1-1-4.8-3-S250-B 1.04 1.03 100.0 100.0 TCPO 21,230
0.85-0.85-6.3-1-S250 0.88 0.87 100.0 99.9 TCPO 5203
0.85-1-6.3-1-S250 0.87 1.03 100.0 99.9 TCPO 6686
0.85-1.5-6.3-1-S250 0.88 1.59 100.0 99.9 TC(B)CPO 12,232
1-1-6.3-1-S250-A 1.03 1.03 99.9 99.9 TCPO 6923
1-1-6.3-1-S250-B 1.04 1.03 99.9 99.9 TCPO 7013
1-1-6.3-1-S250-C 1.04 1.04 99.9 100.0 TCPO 6881
1-1-6.3-1-S250-D 1.03 1.04 99.9 99.9 TCPO 7332
1-1-6.3-1-S250-E 1.05 1.03 99.9 99.9 TCPO 7354
1-1-6.3-1-S250-F 1.04 1.04 100.0 99.9 TCPO 7052
1-1-6.3-1-S250-G 1.03 1.03 100.0 100.0 TCPO 7138
0.85-1-4.8-1-S250 0.88 1.03 100.0 100.0 TC(B)CPO 6800
0.85-1.5-4.8-1-S250 0.88 1.58 100.0 100.0 TCPO 9349
1-1-4.8-1-S250-A 1.04 1.03 100.0 100.0 TCPO 6183
1-1-4.8-1-S250-B 1.04 1.04 100.0 100.0 TCPO 6625

t1t, Measured t1 thickness; t2t, measured t2 thickness; a1t, measured a1 width; a2t, measured a2 width; Put, measured ultimate load; T, tilting; NSF, net section failure;
B, bearing; PO, pull out; PT, pull through.

drawn along the axis of the straps. This line was used to centre Throughout the course of an earthquake, displacements
the specimens in the testing machine. Fig. 2 shows a joint ready change their sign and, as a consequence, joints are subject to
to be tested. reversing movements. For this reason, apart from monotonic
A 250 kN universal testing machine was used to load the tensile tests, load–unload tests were also performed. In these
joints. Tests were displacement-controlled and the load was tests, it was particularly important to capture the unloading
applied at a rate of 0.01 mm/s when the elongation of the joint branch of the cyclic axial load response.
was lower than 2 mm; and at a rate of 0.02 mm/s when the The cyclic tests were carried out unloading four times to
elongation was higher. a near zero load (see Figs. 19 and 20). Only tension forces
Every 0.04 mm, the applied force (F) and the length were applied to the diagonal straps, because they do not
increment of the joint (d) were measured and stored in a have compression strength. The experimental procedure
computer. On the basis of these data, F–d curves, such as the followed was similar to the one explained above. The only
ones shown in Figs. 9–14, could be drawn. It can be seen that difference was that the unloading process was load-
the specimens were loaded until they failed and the measured controlled to ensure that the specimens were not
load was almost zero. compressed.
 M. Casafont et al. / Thin-Walled Structures 44 (2006) 197–210 201

t2 t1

t2 t1

Fig. 4. Thickness of the straps.

4. Results of monotonic tensile tests

4.1. Modes of failure

The failure modes observed in the tests involved a Fig. 6. TiltingCnet section failure. Specimen: 1-1-6.3-6-S350.
combination of two or more of the following phenomena:
with six columns), the second one was obtained. When the
bearing (B), tilting (T), pull-out (PO), pull-through (PT) and
pulling out process began, local plastification around the screw
net section failure (NSF) [14–16].
could also be seen in some joints.
The thickness of the straps and the number of screw
When the straps had different thickness (t1!t2, Fig. 4), there
columns were the parameters that determined the mode of
were some specimens for which bearing of the thinner sheet
collapse of the joints. When both straps had the same thickness
was as significant as tilting during the first steps of the loading
(t1Zt2, Fig. 4), tilting always occurred from the beginning of
process (Fig. 7). The joints more prone to bearing were those
the tests. The final mode of failure, however, was a different with small diameter screws and great differences between t1
phenomenon, which took place after a period of joint and t2. The final mode of failure also depended on the number
elongation. This final mode could be either pull-out (Fig. 5), of screw columns. Pull-out and pull-through were observed in
or net section failure (Fig. 6), depending on the number of joints that had only one column of screws, while the final mode
screw columns: if the number of screws was small (four of those joints with six columns was always net section failure.
columns or less), the first mode occurred; if it was large (joints Joints with an intermediate number of screws could fail in any
of the three mentioned ultimate modes of collapse: pull-out,
pull-through (Fig. 8) or net section failure.

4.2. Force–displacement curves

All modes of failure observed in the tests are listed in

column 6 of Tables 2a and 2b. It can be concluded from this list
that there were basically two modes of failure:

Fig. 5. TiltingCpull out. Specimen: 1-1-6.3-4-S350. Fig. 7. TiltingCbearingCnet section failure. Specimen: 0.85-1.5-4.8-3-S250.
202 M. Casafont et al. / Thin-Walled Structures 44 (2006) 197–210

60000

50000

40000

F (N)
30000

20000

10000

0
0 10 20 30 40
d (mm)

Fig. 11. F–d curve of a TCBCPO joint. Specimen 1.5-1.5-6.3-4-S350.

– TCBCPO: tiltingCbearingCpull out (or pull through),

– TCNSF: tilting (and bearing)Cnet section failure.

The behaviour of the joints, which failed TCNSF was

Fig. 8. TiltingCpull through. Specimen: 1-3-4.8-3-S250.
different from the behaviour of those, which failed TCBCPO.
Figs. 9 and 10 show the two types of F–d curves obtained for
the joints that underwent the TCNSF mode. The first one
(Fig. 9) has three branches: elastic behaviour, hardening and
60000
failure; while the second one has an additional yielding branch.
Four types of F–d curves were observed for the joints that
50000 failed TCBCPO (Figs. 11–14). Actually, most of the tests on
these joints gave the curves in Figs. 11 and 12. It might be said
40000 that the first one (Fig. 11) has three branches and the second
one four, as in Figs. 9 and 10. However, these branches are not
F (N)

30000
as clearly defined as for TCNSF joints, because both yielding
and failure occur gradually.
20000
The F–d curves in Figs. 13 and 14 were observed in a few
10000 specimens that had a small number of screws (one or three
columns). Curves in Fig. 13 have only two branches: elastic
0 behaviour and failure; while curve in Fig. 14 can be seen as a
0 10 20 30 40
d (mm) special case of curve in Fig. 12.
Fig. 13 also shows that the length of the failure branches
Fig. 9. F–d curve of a TCNSF joint. Specimen 1-1-4.8-6-S350. increases with the length of the screw shaft. This behaviour is
not observed in case of net section failure.

60000
60000

50000
50000

40000 40000
F (N)
F (N)

30000 30000

20000 20000

10000 10000

0 0
0 10 20 30 40 0 10 20 30 40
d (mm) d (mm)

Fig. 10. F–d curve of a TCNSF joint. Specimen 0.85-1.5-6.3-6-S250. Fig. 12. F–d curve of a TCBCPO joint. Specimen 1-1-6.3-4-S350.
 M. Casafont et al. / Thin-Walled Structures 44 (2006) 197–210 203

60000 45

40 T+NSF
50000 T+B+PO
35
40000 30

du (mm)
F (N)

25
30000
20
20000
15

10000 10

5
0
0 10 20 30 40 0
d (mm) 0 20000 40000 60000 80000
Put (N)
Fig. 13. F–d curve of a TCBCPO joint. Specimen 1-1-6.3-1-S250. Shaft
lengths: 10 and 30 mm. Fig. 16. Displacement du vs ultimate load Put.

60000 be seen in Fig. 15, which also reveals that the TCBCPO
failure (circles) is less ductile than the TCNSF failure
50000
(crosses). The same trend is observed in Fig. 16, where du is
40000 plotted for all the joints tested.
Fig. 17 demonstrates more clearly the loss of ductility
F (N)

30000 associated to the TCBCPO failure. The F–d curves plotted

correspond to joints that were similar and had also similar
20000
ultimate loads, but that failed in a different mode.
The effect of screw number, screw diameter, sheet thickness
10000
and steel grade on ductility was also investigated. It was
0 observed that ductility increased when the steel grade changed
0 10 20 30 40 from S350 GDCZ to S250 GDCZ, and that, for those
d (mm)
connections that failed TCBCPO, ductility was also slightly
Fig. 14. F–d curve of a TCBCPO joint. Specimen 0.85-1-4.8-1-S250. higher when the thickness of the straps was different (t1!t2).
A modification of the other mentioned parameters affected
4.3. Ductility and stiffness of the connections
ductility only when it provoked a change in the mode of failure
The ductility of the connections was studied by means of the from TCBCPO to TCNSF.
displacement ductility ratio rdZdu/dy, where du is the Finally, the stiffness of the connections was calculated.
displacement corresponding to the maximum load, and dy is Column 4 in Tables 3–5 shows the slope of the linear loading
the displacement at yielding. branch of the joints tested. It was observed that the stiffness of
The connections showed medium to high degree of the connection depends mainly on the number of screws: the
ductility, most of the ratios calculated are above 2. This can larger is the number of screws, the higher is the stiffness
(Fig. 18).
35

30 T+NSF
T+B+PO 50000
25
40000
20
T+NSF
F (N)
rd

30000
15
20000
10 T+B+PO
10000
5
0
0 0 10 20 30 40 50
0 20000 40000 60000 80000 d (mm)
Put (N)
Fig. 17. F–d curve: 0.85-1.5-4.8-4-S250 (TCNSF) 1-1-4.8-4-S250 (TCBC
Fig. 15. Ratio rdZdu/dy vs ultimate load Put. PO).
204 M. Casafont et al. / Thin-Walled Structures 44 (2006) 197–210

Table 3
Parameters and calculated values of TCNSF joints

Connection rd rf kj (N/mm) Py/Antfyt Pud (N) Put/Antfut du (mm)

1-1-6.3-6-S350-A 15.4 1.11 47,573 1.03 43,127 0.95 19.52
1-1-6.3-6-S350-B 12.6 1.12 20,676 1.02 42,965 0.96 23.05
1-1-6.3-6-S350-C 10.6 1.11 18,788 0.99 42,545 0.96 24.00
1-1-6.3-6-S350-D 9.10 1.09 21,468 0.89 43,542 0.93 21.00
1-1-6.3-S350-E 14.5 1.12 20,702 1.02 42,873 0.96 22.40
1-1.5-6.3-6-S350 10.4 1.16 33,469 1.07 43,167 0.99 14.01
1.5-1.5-6.3-6-S350 11.9 1.16 39,941 0.96 66,628 0.98 23.30
1.5-3-6.3-6-S350 10.3 1.18 62,902 1.04 67,265 0.99 13.85
1-1-4.8-6-S350-A* 10.8 1.09 23,508 0.85 44,644 0.90 21.60
1-1-4.8-6-S350-B* 14.8 1.08 21,316 0.98 44,323 0.90 20.04
1-3-4.8-6-S350-A 14.4 1.16 48,720 0.97 44,475 0.96 13.46
1-3-4.8-6-S350-B 12.4 1.16 38,760 1.00 45,068 0.96 14.30
1.5-1.5-4.8-6-S350 11.3 1.16 36,270 1.00 67,866 0.95 22.85
1.5-3-6.3-4-S350-A 8.9 1.18 50,653 1.06 66,067 1.00 13.21
1.5-3-6.3-4-S350-B 8.6 1.18 44,270 1.06 65,869 1.01 12.80
1-3-4.8-4-S350* 10.1 1.14 39,832 1.02 44,857 0.95 14.90
1.5-3-6.3-3-S350* 6.3 1.14 34,255 1.08 66,351 0.97 14.56
0.85-1-6.3-6-S250 28.6 1.12 27,589 1.03 25,752 1.05 28.30
0.85-1.5-6.3-6-S250 27.3 1.10 19,947 1.01 25,300 1.04 22.50
1-1-6.3-6-S250-A 21.5 1.15 24,195 1.05 33,919 1.01 32.50
1-1-6.3-6-S250-B 23.7 1.13 20,568 1.05 34,409 0.99 31.76
0.85-0.85-4.8-6-S250* 18.3 1.05 19,535 0.93 26,174 0.96 29.65
0.85-1-4.8-6-S250* 25.9 1.15 24,192 1.02 26,274 1.05 32.70
0.85-1.5-4.8-6-S250 29.1 1.18 25,465 0.95 25,249 1.08 29.05
1-1-4.8-6-S250-A 14.4 1.12 16,667 0.89 34,002 0.96 25.10
1-1-4.8-6-S250-B 23.4 1.19 19,300 0.93 34,094 1.02 35.83
0.85-1-6.3-4-S250* 12.4 1.06 15,515 0.94 25,661 1.00 22.95
0.85-1.5-6.3-4-S250* 19.3 1.07 25,706 0.98 25,394 1.01 18.60
0.85-1.5-4.8-4-S250 25.3 1.05 22,205 1.01 29,347 0.96 21.25
0.85-1.5-6.3-3-S250* 16.1 1.13 13,765 1.00 25,221 1.06 30.55
0.85-1.5-4.8-3-S250* 4.0 1.04 24,973 0.92 26,249 0.95 8.4
*
joints that were predicted to fail TCBCPO. rd, ductility ratio du/dy; rf, ductility ratio Put/(Atfyt); kj, measured joint stiffness; Py, yielding load; Pud, calculated
ultimate load; Put, measured ultimate load; du, displacement at maximum load.

5. Results of load–unload tensile tests overstrength to allow the development of the cyclic yielding in
the bracings.
The unloading of the specimens did not change the results of A new strength ductility ratio is defined to know in which
the tests. The F–d curves, the modes of failure and the ultimate degree the connections tested allow the dissipative action: rfZ
loads were similar to those of the monotonic tensile tests. For Put/(Atfyt), where Put is the experimental ultimate load of the
the sake of comparison, Figs. 19 and 20 show, plotted in the connection, At is the measured gross cross section area of the
same graph, curves of monotonic and load–unload tests. In thinner strap (from Tables 2a and 2b), and fyt is the measured
Fig. 19, both curves are almost coincident, apart from the yield stress of the steel (Table 1). Tables 3–5 and Fig. 21 show
unloading branches. the ratios calculated for all the connections.
It could also be seen that the unloading paths were similar to In Section 4.3, it has been shown that medium values of rd
the loading paths, and that no stiffness degradation occurred. were obtained for TCBCPO joints. It could also be seen that
these joints can keep their strength for displacements higher
6. Joint design for x-braced dissipative frames than 3 mm (Figs. 12 and 14), which is the limit value of
displacement chosen by the ECCS recommendations [18].
As discussed in Section 1, diagonal straps may act as There is still strength after what is known as ‘the first drop at
dissipative members when used as bracings of low rise cold- 3 mm’. Therefore, TCBCPO joints may be considered
formed steel buildings. According to Eurocode 8 [17], some ductile. However, from the seismic point of view, and
rules must be followed to ensure the dissipative action of the according to the approach presented in this paper, these joints
bracings. The main one is that they should be designed so that are not suitable for x-braced dissipative frames. This is because
yielding of the straps takes place before failure of the their rf ratios are lower than 1 (Fig. 21), which means that they
connection and before collapse of the other members of the do not allow the dissipative action of the straps. Joints collapse
structure. The non-dissipative parts should have sufficient before yielding of the bracings.
 M. Casafont et al. / Thin-Walled Structures 44 (2006) 197–210 205

Table 4
Parameters and calculated values of TCBCPO joints connecting two identical straps

Connection rd rf kj (N/mm) P3/Pud Pud (N) Put/Pud du (mm)

1-1-6.3-4-S350-A 3.3 0.89 18,116 0.92 30,086 1.09 6.50
1-1-6.3-4-S350-B 2.8 0.86 14,494 0.94 29,913 1.08 5.8
1.5-1.5-6.3-4-S350 5.7 1.03 36,674 0.91 57,745 1 9.05
1-1-4.8-4-S350-A 3.6 0.79 21,231 0.86 28,287 1.04 8.40
1-1-4.8-4-S350-B 4.0 0.80 18,431 0.86 28,383 1.05 8.44
1.5-1.5-4.8-4-S350 3.9 0.93 35,254 0.85 49,644 1.05 7.45
1-1-6.3-3-S350-A 2.8 0.67 11,526 0.91 22,586 1.11 5.48
1-1-6.3-3-S350-B 3.5 0.68 16,197 1.01 22,671 1.11 5.60
1-1-4.8-3-S350-A 2.5 0.65 17,606 0.81 20,278 1.19 12.86
1-1-4.8-3-S350-B 4.3 0.59 17,201 0.97 19,549 1.13 8.28
1.5-1.5-4.8-3-S350 3.7 0.67 22,529 0.91 37,319 1.03 7.75
1-1-6.3-1-S350-A 4.3 0.21 6318 0.90 7571 1.06 5.05
1-1-6.3-1-S350-B 3.9 0.22 7007 1.03 7611 1.08 4.08
1.5-1.5-6.3-1-S350 2.9 0.27 12,934 1.10 13,718 1.1 3.00
1-1-4.8-1-S350-A 7.5 0.20 5229 0.79 6536 1.12 7.52
1-1-4.8-1-S350-B 5.4 0.20 5848 0.91 6768 1.17 7.68
1.5-1.5-4.8-1-S350 4.0 0.24 15,573 0.97 12,612 1.07 5.00
0.85-0.85-6.3-4-S250 5.6 0.88 11,5431 0.94 17,149 1.23 6.10
1-1-6.3-4-S250-A 5.1 0.91 16,351 1.02 25,107 1.1 7.40
1-1-6.3-4-S250-B 5.2 0.85 15,498 0.98 25,005 1.04 6.2
0.85-0.85-4.8-4-S250 7.4 0.91 52,243 1.38 14,827 1.47 6.06
1-1-4.8-4-S250-A 4.0 0.88 12,493 0.94 21,754 1.22 6.53
1-1-4.8-4-S250-B 3.1 0.86 15,498 0.92 21,984 1.18 6.29
1-1-6.3-3-S250-A 1.0 0.66 13,837 1.08 18,169 1.09 3.00
1-1-6.3-3-S250-B 4.7 0.65 16,093 0.97 18,646 1.07 5.40
1-1-4.8-3-S250-A 3.3 0.69 17,742 1.06 16,053 1.3 6.95
1-1-4.8-3-S250-B 3.4 0.70 16,143 1.03 15,963 1.33 7.36
0.85-0.85-6.3-1-S250 1.6 0.22 4300 1.22 4130 1.26 3.00
1-1-6.3-1-S250-A 1.5 0.23 4271 1.12 6127 1.13 3.15
1-1-6.3-1-S250-B 2.8 0.23 2856 1.03 6152 1.14 5.60
1-1-6.3-1-S250-C 3.5 0.23 6479 1.00 6255 1.1 4.20
1-1-6.3-1-S250-D 3.9 0.24 5467 1.03 6110 1.2 5.30
1-1-6.3-1-S250-E 3.6 0.24 5516 1.09 6128 1.2 5.75
1-1-6.3-1-S250-F 3.0 0.23 5868 1.05 6133 1.15 4.96
1-1-6.3-1-S250-G 4.6 0.24 6775 1.082 6153 1.16 4.60
1-1-4.8-1-S250-A 4.9 0.20 3715 1.015 5377 1.15 5.05
1-1-4.8-1-S250-B 3.9 0.22 2282 0.82 5387 1.23 7.28

rd, Ductility ratio du/dy; rf, ductility ratio Put/(Atfyt); kj, measured joint stiffness; P3,; load at 3 mm displacement, Pud, calculated ultimate load; Put, measured ultimate
load; du, displacement at maximum load.

Table 5
Parameters and calculated values of TCBCPO joints connecting two different straps

Connection rd rf kj (N/mm) P3/Pud Pud (N) Put/Pud du (mm)

1-1.5-4.8-3-S350 5.2 0.98 20,490 1.06 22,761 1.60 11.05
1-3-4.8-3-S350 4.9 1.00 18,295 1.19 28,086 1.31 8.40
1-1.5-6.3-1-S350 6.3 0.40 10,570 1.16 9073 1.64 6.50
1.5-3-6.3-1-S350-A 9.2 0.55 51,171 1.36 17,415 1.79 7.55
1.5-3-6.3-1-S350-B 9.9 0.54 17,071 0. 77 18,207 1.67 9.94
1-1.5-4.8-1-S350 8.0 0.37 8524 1.14 7580 1.81 8.25
0.85-1-4.8-4-S250 12.9 1.05 17,496 0.89 14,943 1.68 20.69
0.85-1-6.3-3-S250 4.9 0.84 10,110 1.23 13,703 1.46 8.80
0.85-1-4.8-3-S250 4.1 0.86 13,956 1.20 11,561 1.78 8.70
0.85-1-6.3-1-S250 6.4 0.28 4410 1.07 4518 1.48 6.50
0.85-1.5-6.3-1-S250 13.3 0.51 5923 1.36 5797 2.11 19.45
0.85-1-4.8-1-S250 7.0 0.28 5824 1.11 3886 1.75 7.11
0.85-1.5-4.8-1-S250 4.7 0.39 19,531 1.62 4795 1.95 4.3

rd, Ductility ratio du/dy; rf, ductility ratio Put/(Atfyt); kj, measured joint stiffness; P3, load at 3 mm displacement; Pud, calculated ultimate load; Put, measured ultimate
load; du, failure load displacement.
206 M. Casafont et al. / Thin-Walled Structures 44 (2006) 197–210

70000 1.4

60000 1.2

50000 1
kj (N/mm)

40000 0.8 T+NSF

T+NSF

rf
T+B+PO T+B+PO
30000 0.6

20000 0.4

10000 0.2

0 0
0 4 8 12 16 0 20000 40000 60000 80000
n Put (N)
Fig. 18. Stiffness vs number of screws (kj vs n).
Fig. 21. Strength ductility ratio vs ultimate load.

50000 when a modification of these parameters led to a change in the

45000
mode of failure from TCBCPO to TCNSF.
When the design of an x-braced frame with TCNSF joints
40000
is tackled by means of a beam finite element model, the
35000
behaviour of the set composed of the joint and the strap can be
30000 described with a trilinear curve (Fig. 22). In the first part of this
F (N)

25000 curve, both the joint and the member subject to the axial force
20000 keep their original linear behaviour Therefore, the stiffness of
15000 the set is
10000
1
5000 kZ 1
(1)
kj C k1m
0
0 5 10 15 20 25 30 35 40
where kj is the stiffness of the joint and km is the stiffness of the
d (mm)
member.
Fig. 19. F–d curve of a load–unload test. Specimen 1-1-6.3-6-S250. For the joints presented in this paper, these values can be
calculated from their geometry and the Young Modulus of the
steel as
On the contrary, all the joints, which failed TCNSF show rf
EAn
ductility factors greater than 1. Therefore, these are the joints to kj Z (2)
[j
be used in x-braced frames because, with them, yielding of the
straps can occur.
The effect of sheet thickness, number of screws and steel EA
km Z (3)
grade on rf was also investigated. Ductility was achieved only [m
where, A is the gross cross-section area, An is the net cross-
section area, [j is the length of the joint, and [m is the length of
the member.
50000
45000
40000 F
35000
30000
F (N)

A n·fu
nt member
25000 A·f y joi 3
20000 A n·fy 2
15000
joint

er
mb

10000
me

1
5000 k=
F kj=
F k m=
F
d ∆j ∆m
0
0 5 10 15 20 25 30 35 40
d (mm) ∆j ∆m d

Fig. 20. F–d curve of a load–unload test. Specimen 1-1-6.3-4-S350. Fig. 22. Model of the joint-member behaviour.
 M. Casafont et al. / Thin-Walled Structures 44 (2006) 197–210 207

As the member is much longer than the joint, the stiffness of 2.2
the set, k, is very close to the stiffness of the member, km. 2
In the second part of the curve, when the applied force
1.8
exceeds Anfy, there is a change in stiffness. The member

Put / Pud
remains with the same km of Eq. (3), but the joint stiffness is 1.6
T+NSF
governed by the tangent modulus 1.4 T+B+PO

Et An 1.2
kj Z (4)
[j 1

Finally, once the applied force reaches Afy, the member is 0.8
0.8 1 1.2 1.4 1.6 1.8
also governed by Et:
t1 (mm)
Et A
km Z (5) Fig. 23. Ratio Put/Pud vs thickness t1 (mm).
[m
The end of this third branch corresponds to the ultimate
resistance of the joint Anfu.
It should be noticed that ductility is high when the quotient
fu/fy is high and An/A is close to 1. – Net-section resistance
An fu
Pn;Rd Z (10)
7. Strength of screwed connections gM2
where, An is the net cross-sectional area of the strap.
The connection maximum load carrying capacities obtained
The strength calculations were carried out taking gM2 equal
in the experimental campaign were compared to those that
to 1 and using the core thickness of the strap: t1corZt1tK
result from design calculations.
tcoatingZt1tK0.04 mm, where t1t is the measured thickness of
The Eurocode 3 Part 1.3 was applied to predict the strength
the thinner steel sheet (Tables 2a and 2b).
of the specimens. According to this code, there are three
The results obtained applying the above formulas can be
possible modes of failure when loading a joint in shear: bearing
observed in Tables 3–5 and in Figs. 23 and 24, which show the
(which also includes tilting), net-section failure and shear
predicted strength values Pud(PudZPb,Rd or Pn,Rd), the
failure of the screws. This last mode is not considered in this
experimental ultimate loads Put, and also the ratio between
paper, because it was not observed in the tests. In fact, joints
Put and Pud(guZPut/Pud). These results are analysed in the
had been designed so that the shear resistance was 1.2 times
following sections.
higher than the bearing and net section failure resistance, as
prescribed in Eurocode 3.
The strengths were calculated as follows 7.1. Connections failing TCNSF

– Bearing resistance In Figs. 23 and 24, it can be observed that the strength
predictions are slightly higher than the experimental ultimate
afu dt1 loads for some of the TCNSF connections (guZPut/Pud!1).
Pb;Rd Z (6)
gM2 This is shown more clearly in Fig. 25 and in Table 3. In spite of
this, the results of the calculations can be considered
where
rffiffiffiffi satisfactory:
t
a Z 3:2 1 % 2:1 if t1 Z t2 (7)
d
rffiffiffiffi 2.2
t
a Z 3:2 1 % 2:1 if t2 R 2:5t1 and t1 ! 1:0 mm (8) 2
d
1.8

a Z 2:1 if t2 R 2:5t1 and t1 R 1:0 mm (9)

Put / Pud

1.6 T+NSF
1.4 T+B+PO
a is determined by linear interpolation if t1!t2!
2.5t1, 1.2
d nominal diameter of the screw, 1
t1 thickness of the thinner connected strap,
0.8
t2 thickness of the thicker connected strap, 0.8 1.3 1.8 2.3 2.8 3.3
fu ultimate tensile strength of the strap (fut in
t2 / t1
Table 1),
gM2 material partial factor. Fig. 24. Ratio Put/Pud vs ratio t2/t1.
208 M. Casafont et al. / Thin-Walled Structures 44 (2006) 197–210

1.1 1.6
1.5
1.05 1.4

Put / Pud
1.3
Put / Pud

1
1 column
1.2
> 1 column
0.95 1.1
1
0.9 0.9
0.8
0.85 0.8 1 1.2 1.4 1.6 1.8
0.8 1 1.2 1.4 1.6 1.8
t1 (mm)
t1 (mm)
Fig. 27. Ratio Put/Pud for TCBCPO joints (t1Zt2).
Fig. 25. Ratio Put/Pud for TCNSF joints.

Mean value of Put =Pud : gu mean Z 0:98 connecting two straps of the same thickness (Figs. 23 and 27
Standard deviation of Put =Pud : Sg Z 0:043: and Table 4):

The yielding loads were also similar to the yielding values gu mean Z 1:14; sg Z 0:095:
obtained in the laboratory. Fig. 26, which shows the ratio
Nevertheless, these values may be considered acceptable.
between the experimental and the theoretical yielding loads
Actually, the bearing resistance Eq. (6) included in the
(Pyt and PydZAnfy, respectively), demonstrates that Pyt may
Eurocode was developed considering a limit value of failure
also be correctly predicted by means of Pyd:
load equal to the applied force at 3 mm displacement [5,18]. In
Mean value of Pyt =Pyd : gy mean Z 0:99 the tests performed at the laboratory, it was observed that the
Standard deviation of Pyt =Pyd : Sg Z 0:057: 3 mm displacement loads (P3 in Table 4) are close to the values
predicted by Eqs. (6)–(9) when applied to connections with
Finally, it should be pointed out that about the 30% of the only one column of screws (Fig. 28):
connections that failed TCNSF were predicted to fail TCBC
g3 mean Z 1:00; sg Z 0:11:
PO when applying the above equations. For these joints, if Eqs.
(6)–(9) had been applied instead of Eq. (10), the calculated It should be pointed out, however, that the ultimate load of
ultimate loads would have been lower than those shown in these specimens is obtained for a higher displacement, about
Table 3. 5 mm.
One of the reasons why the mode of failure is not correctly When there is more than one column of screws, the 3 mm
predicted may be because sometimes Eqs. (6)–(9) under- loads are not so close to the values given by the formulas:
estimate the bearing strength of the joint, as it will be seen in
the following sections. g3 mean Z 0:96; sg Z 0:12:
The ultimate loads obtained for these connections also occur
7.2. Connections with two identical straps (t1Zt2) failing TC for higher displacements, about 7 mm.
BCPO
7.3. Connections with two different straps (t1!t2) failing TC
Bearing formulas (6)–(9) give strength values slightly lower BCPO
than the experimental ultimate loads for TCBCPO joints
1.1 Conservative results were obtained when the equations
defined for the bearing mode of failure were applied to predict
1.05 the strength of TCBCPO joints connecting two straps of
different thickness (Fig. 29 and Table 5):
Pyt / Pyd

1
gu mean Z 1:96; sg Z 0:21:
0.95 If the calculated strengths are compared to the 3 mm
displacement loads, there are still substantial differences:
0.9
g3 mean Z 1:20; sg Z 0:18:
0.85
0.8 1 1.2 1.4 1.6 1.8
From these results, it may be concluded that Eqs. (6)–(9) do
not lead to good predictions of the failure loads. These
t1 (mm)
equations excessively underestimate the strength of the t1!t2
Fig. 26. Ratio Pyt/Pyd for TCNSF joints. type of joints.
 M. Casafont et al. / Thin-Walled Structures 44 (2006) 197–210 209

1.6 4.5
1.5

Put / Put2 screws

4
1.4
1.3
P3 / Pud

3.5
1 colum
1.2
> 1 column
1.1 3

1
2.5
0.9
0.8 2
0.8 1 1.2 1.4 1.6 1.8 5 6 7 8 9
t1 (mm) n

Fig. 28. Ratio P3/Pud for TCBCPO joints (t1Zt2). Fig. 31. Ratio Put/Put2screws vs number of screws.

Eq. (9), defined for specimens with thickness greater than

1 mm, instead of Eq. (8), defined for specimens with thickness
2.2
lower than 1 mm [19].
2 In the end, the bearing strengths were determined using Eqs.
(6)–(7) and
1.8
rffiffiffiffi
Put / Pud

t
1.6
1 column
> 1 column
if t2 R 2:5t1 and t1 ! 0:85 mm a Z 3:2 1 % 2:1 (11)
d
1.4
if t2 R 2:5t1 and t1 R 0:85 mm a Z 2:1 (12)
1.2
if t1!t2!2.5t1 are determined by linear interpolation, but
1
0.8 1 1.2 1.4 1.6
using teqZ(t1Ct2)/2 instead of t1 in (6) and (7).
Fig. 30 shows that the above equations are still conservative,
t1 (mm)
but in a lower degree than the previous ones:
Fig. 29. Ratio Put/Pud for TCBCPO joints (t1!t2).
gu mean Z 1:26; sg Z 0:14:
It was believed convenient to repeat the calculations
Further work should be done to improve the predictions
introducing a small change in the Eurocode 3 formulas. As it
given by the bearing formulas.
will be seen subsequently, the strength of those joints with t1!
Two additional questions are pointed out with reference to
t2!2.5t1 was determined using an equivalent thickness, which
the experimental ultimate loads of all the connections tested.
is the mean value of t1 and t2. According to [15], a similar
The first one is that, from the results obtained at the laboratory,
approach is used in the Canadian design standard, which
it can be concluded that parameters such as the type of screw
provides a formula based on a combination of t1 and t2. head (hexagonal or square) and shaft length did not affect the
Another change was introduced. It can be seen in Fig. 29 strength of the specimens. The second one is that the joints
that the experimental strength of two of the connections with tested in this campaign did not clearly show group effect. The
t1!1 mm is clearly underestimated. For this reason, it was experimental bearing strengths of the specimens with six and
decided to investigate whether these analytical results could be eight screws were compared to the strength of the specimens
improved by calculating the strength of these joints applying with two screws. Fig. 31 allows seeing that an increase in the
number of screws did not provoke a decrease in the strength per
2.2 screw. For instance, the strength of eight-screw joints was
similar to four times the strength of two-screw joints.
2

1.8
8. Conclusions
Put / Pud

1colum
1.6
> 1 column
The experimental investigation performed on the behaviour
1.4
of screwed connections has allowed distinguishing two types of
1.2 joints, depending on the mode of failure. On the one hand, there
are those joints that fail in a combination of tilting, bearing and
1
pull out or, sometimes, pull through (the TCBCPO type). On
0.8 1 1.2 1.4 1.6
the other hand, there is a second group of joints whose failure
t1 (mm)
mode is basically a combination of tilting and net section
Fig. 30. Ratio Put/Pud for TCBCPO joints (t1!t2). failure (the TCNSF type).
210 M. Casafont et al. / Thin-Walled Structures 44 (2006) 197–210

From the seismic point of view, the distinction between This paper has presented the results of an investigation on
these two groups is very important because, as it has been screwed joints in straps. These results are part of the outcome
demonstrated in the present paper, only TCNSF joints are of an experimental campaign, which has a broader scope. It
suitable for dissipative x-bracings of lightweight structures. includes tests on bolted joints, stud-track corner joints and
The main problem with the TCBCPO type of joints is that x-braced frames. In future contributions we intend to describe
their failure occurs before the yielding of the diagonal straps the results of these tests.
and, as a consequence, no dissipation can be developed. On the
contrary, TCNSF joints have enough strength to allow the References
cyclic yielding of the diagonals. There can be dissipative
action, which takes place in the load interval between the [1] De Matteis G, Landolfo R. Diaphragm action of sandwich panels in pin-
yielding load of the strap gross cross-section and the ultimate joined steel structures: a seismic study. J Earthquake Eng 2000;4(8):
load of the joint. 251–75.
Moreover, when the ductility of the joint itself was [2] Fülöp LA, Dubina D. Performance of wall-stud cold-formed shear panels
investigated using the displacement ratio (rd), it was observed under monotonic and cyclic loading. Part I: experimental research. Thin
Wall Struct 2004;42(2):228–321.
that the TCNSF failure is rather more ductile than the TCBC [3] Technical instructions. Design of cold-formed loadbearing steel systems
PO failure. This also makes the TCNSF joints more suitable and masonry veneer/steel stud walls. US Army Corps of Engineers, TI
for structures subject to seismic loads. 809-07; 30 November 1998.
Therefore, it seems necessary to design x-braced shear walls [4] Tomà AW. Developments in connections in cold-formed building
with NSF connections in seismic zones. There are different structures and design specifications. Prog Struct Eng Mater 2003;5:
145–52.
ways to do so, for example, by placing enough screw columns
[5] Di Lorenzo G, Landolfo R. Shear experimental response of new
in the joint to avoid bearing, and allow the net section failure of connecting systems for cold-formed structures. J Constr Steel Res 2004;
the strap; or by connecting two steel sheets of different 60:561–79.
thickness (for instance, a thin diagonal strap with a thick [6] Lennon R, Pedreschi R, Sinha BP. Comparative study of some
gusset), which also helps the NSF to occur. mechanical connections in cold formed steel. Constr Build Mater 1999;
Another of the objectives of the experimental investigation 13:109–16.
[7] Pedreschi RF, Shina BP, Davies RJ. End fixity in cold-formed steel
was to study the force–displacement curves that resulted from
sections using press joining. Thin Wall Struct 1997;29(1–4):257–71.
the tests. The identification of the different types of curves and [8] Kolari K. Load-sharing of press-joints in thin-walled steel structures.
their characteristic parameters is required for finite element Fourth international conference on steel and aluminium structures, light-
modelling. One of the main conclusions of this study was that weight steel and aluminium structures (ICSAS 99), Espoo; 1999, p. 577–
to correctly model the behaviour of a NSF connection, both the 84.
elastic stiffness and the tangent stiffness of the joint and the [9] Mäkeläinen P, Kesti J. Advanced method for lightweight steel joining.
J Constr Steel Res 1999;49:107–16.
strap must be taken into account.
[10] Mäkeläinen P, Kesti J, Lu W, Pasternak H, Komann S. Static and cyclic
The last step of the experimental analysis was to compare shear behaviour of the rosette-joint. Fourth international conference on
the test ultimate loads and their corresponding design strengths, steel and aluminium structures, light-weight steel and aluminium
calculated according to the Eurocode 3 Part 1.3. It was verified structures (ICSAS 99), Espoo; 1999, p. 586–92.
that the equation defined for the net section failure gives [11] Pastor N, Rodrı́guez-Ferran A. Hysteretic modelling of x-braced shear
satisfactory strength values. The bearing formulas also result in walls. Thin Wall Struct 2005;43(10):1567–88.
[12] Eurocode 3: Design of steel structures—Part 1-3: General rules—
acceptable predictions of the ultimate loads, but only when
supplementary rules for cold-formed members and sheeting. CEN
they are applied to joints connecting two straps of the same European Committee for Standardisation, Brussels; 2004 (Draft).
thickness. When the thickness of the straps is different, very [13] Morro T, Experimental campaign of shear connections used in light-
conservative design loads are obtained. It is believed that the gauge steel structures, Final degree thesis, Universitat Politècnica de
main drawback of these formulas is that they do not Catalunya; 2004 [in Catalan].
appropriately take into account the thickness of the thicker [14] Yu WW. Cold-formed steel design. 3rd ed. New York: Wiley; 2000.
[15] Rogers CA, Hancock GJ. Screwed connection tests of thin G550 and
strap.
G300 sheet steels. J Struct Eng 1999;125(2):128–36.
Finally, it should be pointed out that the experimental [16] LaBoube RA, Sokol MA. Behaviour of screw connections in residential
ultimate loads of the joints that fail TCBCPO occur for construction. J Struct Eng 2002;128(1):115–8.
displacements higher than 3 mm, which is the reference value [17] prEN 1998-1. Eurocode 8: Design of structures for earthquake resistance.
of the ECCS standards. Actually, the displacements corre- Part 1. General rules, seismic actions and rules for buildings. CEN
sponding to experimental strengths range from about 5 to European Committee for standardization. Brussels; 2002 (Draft).
[18] Technical Committee 7. European recommendations for the testing of
7 mm, depending on the number of screw columns. The
connections in profiled steel sheeting and sections. ECCS publication no.
collapse of the joint takes place later than expected and also for 21; May 1983.
higher loads, which favours the appearance of the NSF mode of [19] Sala M, Screwed connections in light-gauge steel structures, Final degree
failure. thesis, Universitat Politècnica de Catalunya; 2004 [in Spanish].