Casafont - 1
Casafont - 1
Casafont - 1
Abstract
An experimental testing campaign on tensile bolted joints between straps is reported. Two dominant failure modes are identified:
(1) tilting, bearing and tearing of the sheets (TS) and (2) tilting, bearing and net-section failure (NSF). The analysis in terms of ductility
and strength shows that bolted connections are less adequate than screwed connections (reported in Part 1 of this paper) for the seismic
design of X-braced shear walls in lightweight structures. NSF joints are more ductile than TS joints in the sense that they undergo larger
displacements before failure. However, if washers are not used, both types of connections fail before energy dissipation through yielding
of the diagonal straps can occur. Some design recommendations to improve the seismic performance of bolted joints, including the use of
washers, are given. The accuracy of Eurocode 3 formulas to predict the ultimate load is also analyzed.
r 2006 Elsevier Ltd. All rights reserved.
Keywords: Light gauge steel; Lightweight steel; Seismic design; Joints; Bolts; Experimental research; Ductility; Failure modes; Net-section failure; Bearing
0263-8231/$ - see front matter r 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.tws.2006.04.006
ARTICLE IN PRESS
678 M. Casafont et al. / Thin-Walled Structures 44 (2006) 677–691
The investigation is focused on connections. The main The objectives and the scheme of the paper are similar to
objectives are, on the one side, to gain knowledge about what was done for screwed connections. The goals of the
their ductility and their behavior under cyclic loads; and, experiments are:
on the other side, to identify which type of joints are most
suitable for seismic actions, i.e., to know which are the 1. Obtain parameters such as the initial stiffness, yielding
joints that have enough strength to allow the dissipative load, ultimate load and maximum displacement.
yielding of diagonal straps. 2. Obtain complete force–displacement (F–d) curves,
The initial steps of the investigation were presented in a needed for the finite element modeling of X-braced
previous paper [2], devoted to experimental testing of frames [3].
screwed connections subjected to shear loading. The main 3. Identify the failure modes.
conclusion of the analysis of the experimental data is that
the mode of failure is a key issue in seismic design of joints. After that, we analyze the experimental results in order
It is verified that screwed connections should be designed to:
to fail in the net-section failure (NSF) mode, because it is
the most ductile type of failure and because it takes place 1. Classify the various failure modes in terms of their
after the yielding of the straps. seismic suitability (strength and ductility).
In view of this result, it seems that reliable equations to 2. Determine the relation between parameters in joint
predict the mode of failure of a joint are needed to tackle design (steel grade, strap thicknesses, number and
the design of X-braced frames. That is the reason why part diameter of bolts, etc.) and failure mode.
of the first paper is also devoted to verify de accuracy of the 3. Compare experimental ultimate loads of the joints to the
current Eurocode 3 Part 1-3 proposals for the calculation strengths calculated by means of the Eurocode 3 Part 1.3
of joint resistance. It should be pointed out that the design formulas.
Eurocode 3 formulas for the net-section mode of failure
showed to work satisfactory, while the bearing formulas It will also be very interesting to compare the behavior of
gave rather conservative predictions for some of the screwed and bolted connections.
screwed joints. An outline of paper follows. The laboratory experiments
The second part of the investigation, presented in this are described in Section 2 (test specimens) and 3 (test
paper, is focused on bolted connections subjected to shear procedure), and the results are summarized in Section 4
loading. The analysis of the joint behavior is based on the (monotonic tensile tests) and 5 (load–unload tensile tests).
results of a testing campaign performed in the framework Three main features are studied: the modes of failure, the
of the RFCS research project ‘‘Seismic design of Light force–displacement curves, and the ductility and stiffness of
Gauge Steel Framed Buildings’’. Lap joints between two the connections. The remainder of the paper is devoted to
straps connected by means of two rows of bolts are tested the analysis of the results. The seismic suitability of the
under monotonic and cyclic load, see Fig. 1. joints is discussed in Section 6. Then, in Section 7, the
ultimate loads are compared to the values predicted by the
Eurocode. Recommendations for design and the conclud-
ing remarks of Sections 8 and 9 close the paper.
2. Test specimens
The bolt joints tested are similar to the screw joints of the
investigation reported in the first part of the paper (Fig. 1)
[2]. For example, the steel grade of the straps is the same,
either S 350 GD+Z or S 250 GD+Z [5]. The nominal and
mechanical properties of these steels are shown in Table 1.
It should be noticed that the experimental fyt and fut are
rather higher than the nominal fy and fu.
Bolts of two different diameters are used to connect the
straps: 8 and 10 mm. The heads of these bolts are
hexagonal and the shafts are threaded all along their
length. All the straps of steel grade S 350 GD+Z are
connected with washers, while washers are only used in 4
out of the 38 S 250 GD+Z straps. When used, washers of
20 mm F are placed under the bolt head and nut.
The torque applied to the bolts is not measured. Bolts
are tightened by hand using standard tools, so the torque
Fig. 1. Bolt joint. should be small.
ARTICLE IN PRESS
M. Casafont et al. / Thin-Walled Structures 44 (2006) 677–691 679
Table 1 Table 2
Steel mechanical properties Main dimensions of specimens and results of tests
Steel fy fu t fyt fut Connection t1t t2t a1t a2t Failure mode Put (N)
(N/mm2) (N/mm2) (mm) (N/mm2) (N/mm2)
0.85-0.85-10-2-S250 0.88 0.89 100.0 100.0 T+B+NSF 22731
S-350 GD+Z 350 420 1 392 520 0.85-1-10-2-S250 0.88 1.03 100.0 100.0 T+B+NSF 23842
1.5 387 519 0.85-1.5-10-2-S250 0.88 1.54 100.0 100.0 T+B+NSF 24450
3 385 512 1-1-10-2-S250-A 1.02 1.04 100.0 100.3 T+B+NSF 27884
1-1-10-2-S250-B 1.04 1.04 100.0 100.0 T+B+NSF 28160
S-250 GD+Z 250 330 0.85 285 345 1-1-10-2-S250-C 1.02 1.04 100.0 100.0 T+B+NSF 28095
1 303 393 1-1-10-2-S250-D 1.04 1.02 100.0 100.0 T+B+NSF 27828
1.5 317 391 1-1-10-2-S250-E(W) 1.01 1.04 100.0 100.1 T+B+NSF 30039
1-1-10-2-S250-F(W) 1.03 1.02 100.0 100.0 T+B+NSF 30746
fy is the nominal yield stress, fu the nominal ultimate stress, t the nominal
1-1-10-2-S250-G 1.03 1.03 99.9 100.0 T+B+NSF 28332
thickness, fyt the measured yield stress, fut the measured ultimate stress.
1-1.5-10-2-S250 1.03 1.59 100.0 100.0 T+B+NSF 30112
1.5-1.5-10-2-S250 1.57 1.59 100.0 100.0 T+B+NSF 44734
0.85-0.85-8-2-S250 0.88 0.88 100.0 100.0 T+B+TS 20529
The nominal length of the straps is either 350 mm, when 0.85-1-8-2-S250 0.87 1.04 100.0 100.0 T+B+NSF 20433
0.85-1.5-8-2-S250 0.88 1.59 100.0 100.0 T+B+NSF 23131
connected by means of one column of bolts, or 375 mm,
1-1-8-2-S250-A 1.03 1.04 100.0 100.0 T+B+TS 22829
when connected by means of two columns. Their thickness 1-1-8-2-S250-B 1.04 1.58 100.0 100.0 T+B+TS 22441
ranges from 0.85 to 1.5 mm, and their width is always the 1-1.5-8-2-S250 1.04 1.58 100.1 100.0 T+B+NSF 28419
same, 100 mm. The tolerance of the bolt holes drilled in the 1.5-1.5-8-2-S250 1.59 1.59 100.0 100.0 T+B+TS 37572
straps is 1 mm. 0.85-0.85-10-1-S250 0.89 0.89 100.0 100.0 T+B+TS 14335
0.85-1-10-1-S250 0.88 1.04 100.0 100.0 T+B+TS 12933
Fig. 3 shows the position of the bolts: the spacing and
0.85-1.5-10-1-S250 0.88 1.57 100.0 100.0 T+B+TS 12508
the longitudinal and transverse edge distances. The joint 1-1-10-1-S250-A 1.05 1.05 100.0 100.0 T+B+TS 15391
lay out is identical for all the specimens. 1-1-10-1-S250-B 1.05 1.04 100.0 100.0 T+B+TS 15279
The bolt connections are listed in Table 2 together with 1-1-10-1-S250-C 1.04 1.04 100.0 100.0 T+B+TS 14395
their main dimensions and test results. The first column of 1-1-10-1-S250-D 1.04 1.04 100.0 100.0 T+B+TS 15326
1-1-10-1-S250-E(W) 1.04 1.03 100.0 100.0 T+B+TS 17537
the table shows the joint notation, whose meaning is
1-1-10-1-S250-F(W) 1.02 1.02 100.0 100.0 T+B+TS 17865
explained in the following example: 1-1-10-1-S250-G 1.03 1.03 100.0 100.0 T+B+TS 15132
1-1.5-10-1-S250 1.05 1.59 100.0 100.0 T+B+TS 14158
t1 t2 F nc sg lðWÞ; 1.5-1.5-10-1-S250 1.60 1.59 100.0 100.0 T+B+TS 25872
1 1 10 2 S250 EðWÞ; 0.85-0.85-8-1-S250 0.88 0.88 100.0 100.0 T+B+TS 12374
0.85-1-8-1-S250 0.88 1.03 100.2 100.0 T+B+TS 10082
0.85-1.5-8-1-S250 0.89 1.58 100.1 100.2 T+B+TS 10344
where t1 is the thickness of the first strap (t1 ¼ 1 mm), t2 the
1-1-8-1-S250-A 1.05 1.04 100.1 100.0 T+B+TS 12189
thickness of the second strap (t2 ¼ 1 mm), F the diameter 1-1-8-1-S250-B 1.04 1.04 99.9 100.0 T+B+TS 12180
of the bolt (F ¼ 10 mm), nc the number of bolt columns 1-1.5-8-1-S250 1.04 1.57 99.9 100.1 T+B+TS 13137
(nc ¼ 2), sg the steel grade (S250 GD+Z), l: letter used 1.5-1.5-8-1-S250 1.58 1.58 100.0 100.0 T+B+TS 21855
when there are two or more identical joints (E), (W) 1-1-10-1-S350(W) 0.99 0.98 100.0 100.1 T+B+TS 19575
1.5-1.5-10-1-S350(W) 1.48 1.48 100.2 100.2 T+B+TS 45471
denotes that washers are used.
1-1-10-2-S350-A(W) 0.98 0.98 100.2 100.1 T+B+NSF 39120
1-1-10-2-S350-B(W) 0.99 0.98 100.1 100.1 T+B+NSF 38710
1.5-1.5-10-2-S350(W) 1.50 1.50 100.0 100.1 T+B+NSF 59690
3. Test procedure
t1t is the measured t1 thickness (Fig. 4), t2t the measured t2 thickness, a1t
the measured a1 width, a2t the measured a2 width, Put the measured
Tests are performed applying the same procedure as the ultimate load, T the tilting, B the bearing, NSF the net-section failure, TS
one previously followed in the experimental campaign of the tearing of the sheet.
screw connections [2].
The first operation is to measure the actual dimensions
of the joint components (see some of the measured values
in Table 2 and the full collection of measurements in [6,7]). Every 0.04 mm, the applied force (F) and the length
Afterwards, the specimens are labelled and a line is drawn increment of the joint (d) are measured and stored in a
along the axis of the straps. This line is used to center the computer. On the basis of these data, F–d curves, such as
specimens in the testing machine. Fig. 2 shows a joint ready the ones shown in Figs. 9–14, are drawn. It can be seen that
to be tested. the specimens are loaded until they fail and the measured
A 250 kN universal testing machine is used to load the load is almost zero.
joints. Tests are displacement-controlled and the load is In the course of an earthquake, displacements change
applied at a rate of 0.01 mm/s when the elongation of the their sign and, as a consequence, joints are subject to
joint is lower than 2 mm; and at a rate of 0.02 mm/s when reversing movements. For this reason, apart from mono-
the elongation is higher. tonic tensile tests, load–unload tests are also performed.
ARTICLE IN PRESS
680 M. Casafont et al. / Thin-Walled Structures 44 (2006) 677–691
Column
25
Row
100
Row
25
25 25
50000 50000
40000 40000
Agt·fyt
30000 30000
Agt·fyt
F (N)
F (N)
20000 20000
10000 10000
0 0
0 5 10 15 20 25 0 5 10 15 20 25
d (mm) d (mm)
Fig. 9. F–d curve of a T+B+NSF joint. Specimen 0.85-1.5-8-2-S250. Fig. 12. F–d curve of a T+B+TS joint. Specimen 1-1-10-1-S250-C.
50000 50000
Agt·fyt
40000 40000
Agt·fyt
30000 30000
F (N)
F (N)
20000 20000
10000 10000
0 0
0 5 10 15 20 25 0 5 10 15 20 25
d (mm) d (mm)
Fig. 10. F–d curve of a T+B+NSF joint. Specimen 1.5-1.5-10-2-S250. Fig. 13. F–d curve of a T+B+TS joint. Specimen 1-1-8-2-S250-A.
50000 50000
40000 40000
Agt·fyt
30000 30000
Agt·fyt
F (N)
F (N)
20000 20000
10000 10000
0 0
0 5 10 15 20 25 0 5 10 15 20 25
d (mm) d (mm)
Fig. 11. F–d curve of a T+B+TS joint. Specimen 0.85-1.5-8-1-S250. Fig. 14. F–d curve of a T+B+NSF joint. Specimen 1-1-10-2-S350-A (W).
All the calculated rd ratios are above 2, as shown in tively, it can also be concluded that T+B+NSF joints are
Table 3, so the joints can be considered ductile. From more ductile than T+B+TS joints.
Figs. 15 and 16, which depict the values of the ductility In relation to ductility, it was also investigated whether
ratios and the values of displacement at failure, respec- the two types of T+B+TS failure mentioned in Section 4.1
ARTICLE IN PRESS
M. Casafont et al. / Thin-Walled Structures 44 (2006) 677–691 683
Table 3 25
Ductility ratios and stiffness
T+B+NSF
Connection rd rf k1 k2 du T+B+TS
20
(mm)
rd
1-1-10-2-S250-A 7.0 0.94 75260 7538 14.3
1-1-10-2-S250-B 9.9 0.94 80692 8612 19.4
10
1-1-10-2-S250-C 7.2 0.96 78000 9174 15.4
1-1-10-2-S250-D 5.6 0.95 75965 8087 13.4
1-1-10-2-S250-E(W) 16.5 1.02 69000 9450 26.8
1-1-10-2-S250-F(W) 13.4 1.05 61031 8919 24.9 5
1-1-10-2-S250-G 6.5 0.94 74353 7549 15.7
1-1.5-10-2-S250 6.6 1.00 85337 11271 14.9
1.5-1.5-10-2-S250 6.0 0.92 121200 14890 15.5 0
0.85-0.85-8-2-S250 8.9 0.86 67083 — 8.9 0 10000 20000 30000 40000 50000
0.85-1-8-2-S250 13.1 0.86 68744 3470 10.6
Put (N)
0.85-1.5-8-2-S250 12.9 0.97 81365 5096 16.2
1-1-8-2-S250-A 3.9 0.76 83900 5883 7.7 Fig. 15. Ratio rd ¼ d u =d y vs. ultimate load Put.
1-1-8-2-S250-B 3.6 0.75 84000 6842 10.1
1-1.5-8-2-S250 4.8 0.94 66700 9306 10.3
1.5-1.5-8-2-S250 3.1 0.77 78520 10761 10.4 25
0.85-0.85-10-1-S250 8.0 0.59 97228 — 11.1
0.85-1-10-1-S250 4.6 0.54 64200 3592 9.0 T+B+NSF
0.85-1.5-10-1-S250 4.7 0.52 54208 5206 8.4
20 T+B+TS
1-1-10-1-S250-A 4.0 0.51 76850 4427 10.2
1-1-10-1-S250-B 4.5 0.50 57454 4585 9.5
1-1-10-1-S250-C 6.0 0.48 60836 4431 13.5
1-1-10-1-S250-D 4.0 0.51 60622 4147 9.5 15
1-1-10-1-S250-E(W) 6.1 0.58 76350 6600 9.7
du
These formulas have been applied here to predict the of the connection only depends on k2, because slippage
stiffness of the connections tested. The values obtained always occurs before the service life of the structure. For
with both equations are similar, and many of them above this reason, k1 is not considered relevant. An F–d model is
the experimental results (Figs. 17 and 18): even proposed in [12], where k2 is used for the stiffness
before and after slipping. However, it should be noticed
Eq. (1): Mean value of k2t =k2 : gk2 mean ¼ 0:90,
that k1 is considerably higher than k2 (Table 3), and that, if
Standard deviation of k2t =k2 : sg ¼ 0:24. a connection is designed without slipping [12], k1 becomes
Eq. (2): Mean value of k2t =k2 : gk2 mean ¼ 0:86, relevant.
Standard deviation of k2t =k2 : sg ¼ 0:21,
4.4. The effect of washers on the strength and behavior of
where k2t is the experimental stiffness, k2 the calculated joints
stiffness.
It should be noted that Eq. (1) was defined to Joints 1-1-10-2-S250 and 1-1-10-1-S250 were tested
overestimate the value of k2 [12], and that the thickness with and without washers. Comparing the results of the
of the straps tested are out of the range of validity of tests, it can be seen that washers increase the strength of
Eq. (2) [13]. connections. The resistance of joint 1-1-10-2-S250, which
No formula has been found in the literature to predict fails T-B-TS, is 8.6% higher when washers are used. The
the value of k1. In fact, in [12] it is stated that the stiffness increase in strength of joint 1-1-10-S250, which fails
T+B+NSF, is even higher, close to 17%.
Washers also increase the ductility of the T+B+NSF
1.6 joint. This can be clearly seen in Fig. 19, where the F–d
t1=0.85 curves of the joint with and without washers are compared.
1.4 t2=1.00 The corresponding values of displacement ratio also allow
t3=1.50 to detect this gain in ductility: rd ¼ 7:40 and rdw ¼ 14:97
1.2
(mean values).
1 On the contrary, washers almost do not change the
ductility of joint 1-1-10-1-S250, that fails T+B+TS. Its
k2t/k2
0.8
20000
F (N)
0.6
15000
0.4
10000
0.2 5000
Without washers With washers
0 0
0 4000 8000 12000 16000 0 5 10 15 20 25 30 35
k2t (N/mm) d (mm)
Fig. 18. k2t/k2 values. k2t calculated with (2). Fig. 19. F–d curve of a T+B+NSF joint. Specimen 1-1-10-2-S250.
ARTICLE IN PRESS
M. Casafont et al. / Thin-Walled Structures 44 (2006) 677–691 685
35000 F–d curves, the modes of failure and the ultimate loads are
Agt·fyt
similar to those of the monotonic tensile tests. For the sake
30000
of comparison, Figs. 23 and 24 show, plotted in the same
25000 graph, curves of monotonic and load–unload tests.
It can also be seen that the unloading paths are similar to
20000 the reloading paths, and that no stiffness degradation
F (N)
With washers
15000
occurs.
Without washers
10000
6. Joint design for X-braced dissipative frames
5000
A force ductility ratio is defined as rf ¼ Put =ðAgt f yt Þ,
0 where Put is the experimental ultimate load of the joint, and
0 5 10 15 20 25 30 35
d (mm)
(Agtfyt) is the yielding load of the strap, calculated from the
measured gross cross section area, Agt ¼ at tt (Table 2), and
Fig. 20. F–d curve of a T+B+TS joint. Specimen 1-1-10-1-S250. the measured yield stress of the steel, fyt (Table 1). This
ratio allows to know if a joint is suitable for seismic design
[2]. Joints with ratios higher than 1 have good seismic
behavior, because they fail once the strap has already
yielded and, as a consequence, the dissipative action can
develop.
Fig. 25 shows the rf values calculated for all the
specimens connected without washers. The ratios of the
30000
25000
20000
F (N)
15000
5000
0
0 5 10 15 20 25
d (mm)
30000
25000
20000
F (N)
15000
10000
Fig. 22. T+B+TS failure of specimen 1-1-10-1-S250-F, with washers.
5000
Fig. 25. Strength ductility ratio vs. ultimate load. 7. Strength of bolted connections
0.9
Bearing resistance:
Pb;Rd ¼ 2:5ab kt f u dt=gM2 , (3)
where ab is the smallest of 1.0 or e1/(3d); e1the end
0.8
0 10000 20000 30000 40000 50000 60000 70000 distance from the center of the bolt to the adjacent end
Put (N) of the connected part, in the direction of load transfer; d
the nominal diameter of the bolt;
Fig. 26. Strength ductility ratio vs. ultimate load for screw and bolt NSF
joints. kt ¼ ð0:8t þ 1:5Þ=2:5
for 0:75 mmptp1:25 mm, kt ¼ 1:0 for t41:25 mm;
where t is the thickness of the thinner connected
T+B+TS joints are rather low, ranging from 0.4 to about strap; fu the ultimate tensile strength of the strap
0.8. Ratios of T+B+NSF joints are better, but most of (fut in Table 1); gM2 the material partial factor.
them are also below 1. Therefore, it can be concluded from Net-section resistance:
these results that, although bolted connections are ductile, Pn;Rd ¼ ð1 þ 3rðd 0 =u 0:3ÞÞAn f u =gM2
as discussed in Section 4.3, they are not suitable for seismic
but Pn;Rd pAn f u =gM2 , ð4Þ
design. They do not allow the dissipative action of diagonal
straps in X-braced frames. where r is the (number of bolts at the cross-section)/
When washers are used, the strength of T+B+TS (total number of bolts in the connection); d0 the nominal
connections increases, but their ductility does not change diameter of the hole; u ¼ 2e2 but upp2 ; e2 the edge
considerably (Section 4.4). On the contrary, the effect of distance from the center of the bolt to the adjacent edge
washers on T+B+NSF joints is more significant (Fig. 26). of the connected part, in the direction perpendicular to
They exhibit longer hardening branches and failure loads the direction of load transfer; p2 the spacing center-to-
higher than the yielding loads of the straps. This results in center of bolts in the direction perpendicular to the
good rf values (rf 41), as shown in Table 3. direction of load transfer; An the net cross-sectional area
The F–d curves included in this paper are plotted of the strap; fu the ultimate tensile strength of the strap
together with the calculated yielding load of the straps, (fut in Table 1); gM2 the material partial factor.
ARTICLE IN PRESS
M. Casafont et al. / Thin-Walled Structures 44 (2006) 677–691 687
Table 4
Calculated strengths and strength ratios
Connection Pu3 mm (N) Pn,Rd (N) Pb,Rd (N) Put Put Pu3 mm
Pn;Rd Pb;Rd Pb;Rd
Pu3 mmt is the load at 3 mm displacement, Pn,Rd the calculated net-section resistance, Pb,Rd the calculated bearing resistance, Put the measured ultimate
load.
The strength calculations are carried out taking gM2 Table 2, it can be concluded that good predictions of the
equal to 1 and using the core thickness of the strap: failure mode are obtained by means of Eqs. (3) and (4). The
tcor ¼ tttcoating ¼ tt–0.04 mm, where tt is the measured Eurocode 3 equations fail in only four specimens, most of
thickness of the thinner steel sheet (Table 2). which show calculated bearing strengths similar to
The results obtained applying the above formulas can be calculated net-section strengths (see values of specimens
observed in Table 4, where the third and fourth columns 0.85-0.85-8-2-S250, 1-1-8-2-S250-A and 1-1-8-2-S250-B).
include the net-section and the bearing resistance, respec-
tively. The mode of failure of the joints is predicted on the 7.1. Connections failing T+B+NSF
basis of these calculated strengths. The critical mode is the
one that gives the lowest ultimate load. In view of the The NSF Eq. (4) gives acceptable predictions of the
values of the mentioned columns and the sixth column of T+B+NSF failure load, although they are slightly
ARTICLE IN PRESS
688 M. Casafont et al. / Thin-Walled Structures 44 (2006) 677–691
1.1
7.2. Connections failing T+B+TS
1.0
The Eurocode 3 strengths obtained for the bearing mode
of failure are not as acceptable as the strengths obtained
0.9 for the net-section mode (Table 4):
gu mean ¼ 1:05; sg ¼ 0:16.
0.8
0 10000 20000 30000 40000 50000 60000 70000 The mean value of the Put/Pb,Rd ratio is good, but the
Put (N) dispersion is too high (see also Fig. 29). There can be seen
values of strength clearly underestimated for joints with
Fig. 27. Ratio Put/Pn,Rd for T+B+NSF joints. Pn,Rd with reduction washers and joints with 10 mm F bolts. The worst results,
factor.
however, are those obtained for some 8 mm F joints, whose
bearing strength is overestimated:
1.4
T+B+NSF joints with washers : gu mean ¼ 1:23; sg ¼ 0:16,
T+B+NSF (W) 10 mm F joints : gu mean ¼ 1:12; sg ¼ 0:15,
1.3
8 mm F joints : gu mean ¼ 0:97; sg ¼ 0:14.
Fig. 28. Ratio Put/Pb,Rd for T+B+TS joints. Pn,Rd without reduction 1.6
factor. T+B+TS (10)
T+B+TS (8)
conservative. The Put/Pn,Rd ratios of Table 4 and Fig. 27 1.4
T+B+TS (W)
show the accuracy of this formula:
Mean value of Put =Pn;Rd : gu mean ¼ 1:14, 1.2
Put/Pb,Rd
1.6
T+B+TS (10)
T+B+TS (8)
1.4
1.2
Put/Pb,Rd
1
0.8
0.6
2 4 6 8 10 12 14
d/t
Fig. 30. Bearing failure in specimen 0.85-1-10-1-S250.
Fig. 33. Ratio Put/Pn,Rd for T+B+NSF joints.
1.60
T+B+TS (10)
T+B+TS (8)
1.40
T+B+TS (W)
1.20
P3 /Pb,Rd
1.00
0.80
0.60
0 10000 20000 30000 40000 50000
P3 (N)
Fig. 31. Failure of specimen 1.5-1.5-8-1-S250.
Fig. 34. Ratio Pu3 mm/Pb,Rd for T+B+TS joints.
of 6–11 mm. The same happened in the tests of screw bolts is high, and it is not easy to keep the net-section area
connections. large enough to allow the dissipative yielding of the straps.
However, the behavior of the T+B+NSF connections
8. Recommendations improves when washers are used, their rf ratios become
higher than 1. This is observed for all the joints tested with
Recommendations are given in this section for the washers. Therefore, the main recommendation is that bolt
design of bolted joints between straps of cold formed joints should be used with washers and designed to fail
steel structures. These recommendations, which are mainly NSF.
related to the geometric layout of the connection, can be Other recommendations may be given in order to
classified in two groups: increase the rf values:
1. Choose the steel with the highest fu/fy ratio, which will
(a) Recommendations for increasing the ductility of the
directly increase the Fn,Rd/Agfy ratio.
joint.
2. Use only a row of bolts, which increases the net-section
(b) Recommendations for improving the seismic perfor-
area.
mance of the joint.
3. Drill the minimum feasible bolt diameter, so that the
maximum net-section area is available.
(a) Recommendations for increasing the ductility of the 4. Enlarge the width of the straps in the perforated section
joint: The experimental tests prove that the T+B+NSF to avoid the NSF.
mode of failure is rather more ductile than T+B+TS 5. Place the bolts so that e241.66d, which avoids any
mode. Therefore, it is recommended to design joints that reduction of the net-section area (see the net-section
fail NSF, i.e., to design joints whose bearing strength is strength formula (4)).
higher than the net-section strength (Fb,Rd4Fn,Rd).
From the design point of view and according to the When the rf ratios of T+B+NSF bolted connections
Eurocode 3 calculation formulas: with washers are compared to the ratios of T+NSF
screwed connections [2], it is concluded that the seismic
1. The bearing strength does not increase linearly with the behavior of screws is better. Ratios of screw joints are
thickness of the sheet [12]. For this reason, the Eurocode significantly higher than 1, while ratios of bolts with
3 bearing Eq. (3) includes the factor kt, which gives washers are only slightly higher than 1 (Fig. 26). Conse-
values below 1 when the thickness of the sheet is lower quently, it is also recommended to use screws instead of
than 1.25 mm. Therefore, if a high value of Fb,Rd is bolts in dissipative straps of cold-formed structures.
wanted, so that Fb,Rd4Fn,Rd, sheet thickness above Nevertheless, it is believed that more investigations
1.25 mm are recommended. should be devoted to the seismic suitability of T+B+NSF
2. The bearing formula was defined for the bearing failure bolted joints with washers. In the present paper, from the
and also for the end-tearing failure. The effect of tearing results of a few tests, it has been possible to show the
on strength is considered by means of the ab factor in relevance of washers in the seismic behavior of joints. More
Eq. (3) [12]. To avoid values of ab below 1, bolts should tests should be performed to confirm this point and to
be placed far enough from the end edge of the strap improve the design of connections with washers. For
(e143d according to Eurocode 3). This will also increase instance, it can be investigated whether the use of washers
Fb,Rd with respect to Fn,Rd. in joints of only one row of bolts results in acceptable
dissipative intervals.
A ductile failure (Fb,Rd4Fn,Rd) does not guarantee that Finally, it should be pointed out that the problem of the
the joint will be suitable for seismic design. Other net-section area reduction due to the bolt holes may be
requirements should be accomplished, as it is explained in solved following the fourth recommendation mentioned
the following point (see also Section 6). above. For example, straps of non constant width can be
(b) Recommendations for improving the seismic perfor- used. However, this solution is difficult from the manu-
mance of the joint: The objective is to design joints whose facturing point of view.
strength is higher than the yielding load of the gross cross-
section (Fn,Rd4Agtfy), i.e., joints with rf 41, so that the 9. Conclusions
dissipative action of the straps can develop.
In Section 6, it has been shown that bolted connections The present investigation on bolted connections between
without washers do not satisfy this condition. On the one straps has given results that, in some senses, are similar to
hand, bolt joints failing T+B+TS exhibit very low values the ones obtained in the previous investigation on screwed
of rf ratio and, consequently, they will never be suitable for connections. For instance, here two modes of failure are
seismic design. On the other hand, T+B+NSF joints show also observed, the T+B+TS and the T+B+NSF modes,
higher rf ratios, but most of them are lower than 1. The which show different ductility. Both modes are ductile,
main problem with these joints is that the diameter of the their displacement ductility ratios are above 2 (rd42),
ARTICLE IN PRESS
M. Casafont et al. / Thin-Walled Structures 44 (2006) 677–691 691
but the ductility of the joints that undergo the NSF X-braced frame. However, it is also shown that bolted
(T+B+NSF) is higher than the ductility of the joints that joints with washers, designed to fail NSF and to keep the
fail bearing (T+B+TS). Therefore, regarding the ductility maximum available net-section area, allow the dissipative
of the joint itself, the T+B+NSF type of joint is preferred. yielding of the strap. The problem is that in the present
The main difference with screw joints is that none of the investigation only a small group of joints are tested with
mentioned types of bolted connections are suitable for washers, and it is not known whether this type of
seismic design. When the failure mode is T+B+NSF, the connections may be good enough for seismic design.
ultimate load is close to the yielding load of the strap, Further tests should be carried out to give light to this
but most of the times below it. As a consequence, these question.
connections would not allow the development of the
dissipative action of the strap in an X-braced frame. When
the failure is T+B+TS, the strength of the connection is References
even far lower than the yielding load of the strap.
However, the situation improves when washers are used [1] Hancock GJ. Cold-formed steel structures. J Construct Steel Res
2003;59(4):473–87.
and, consequently, the strength of the connection increases. [2] Casafont M, Arnedo A, Roure F, Rodriguez-Ferran A. Experimental
This is very important for T+B+NSF joints, because this testing of joints for seismic design of lightweight structures.
increase in strength is high enough to allow the dissipative Part 1: screwed joints in straps. Thin-Walled Structures 2006;44(2):
yielding of the bracings. Therefore, the use of washers 197–210.
[3] Pastor N, Rodrı́guez-Ferran A. Hysteretic modelling of x-braced
becomes relevant from both the resistance point of view,
shear walls. Thin-Walled Struct 2005;43(10):1567–88.
and the seismic point of view in the case of T+B+NSF [4] prEN 1998-1. Eurocode 8: Design of structures for earthquake
bolt joints. resistance. Part 1. General rules, seismic acions and rules for
Apart from the ductility of the joints, other questions buildings. CEN European Committee for Standardization, Brussels,
have been investigated. For example, the measurement of 2002 (Draft).
the stiffness of the connections has allowed to conclude [5] Eurocode 3: Design of steel structures—Part 1–3: general rules—
supplementary rules for cold-formed members and sheeting. CEN
that, if an accurate model of the behavior of the joint is European Committee for Standardisation, Brussels, 2004 (Draft).
wanted, it should be taken into account that the value of k [6] Morro T. Experimental campaign of shear connections used in light-
before slipping is considerably higher than its value after gauge steel structures. Final Degree thesis, Universitat Politècnica de
slipping. Catalunya, 2004 [in Catalan].
The effectiveness of the Eurocode 3 Part 1-3 in predicting [7] Fernández C. Bolted joints in straps of lightweight structures, Final
Degree thesis, Universitat Politècnica de Catalunya, 2005 [in
the strength of the joints has also been evaluated. One of Spanish].
the results of this study is that the Eurocode 3 formulas [8] Yu WW. Cold-formed steel design. 3rd ed. New York: Wiley; 2000.
give good predictions of the joint mode of failure. This is [9] Bryan ER. The design of bolted joints in cold-formed steel sections.
very useful in design because, as it has been shown, the Thin-Walled Struct 1993;16(1–4):239–62.
[10] Seleim S, LaBoube R. Behaviour of low ductility steels in cold-
mode of failure of joints is a determining factor in relation
formed steel connections. Thin-Walled Struct 1996;25(2):135–51.
to performance of dissipative frames. When it comes to the [11] Rogers CA, Hancock GJ. Failure modes of bolted-sheet–steel
accuracy of the equations prescribed by this code, it should connections loaded in shear. J Struct Eng 2000;126(3):288–96.
be pointed out that the predictions of the ultimate loads of [12] Zadanfarrokh F, Bryan ER. Testing and design of bolted connections
NSF joints are acceptable, although slightly conservative. in cold formed steel sections. Eleventh International Specialty
The results of the bearing equation are not so good, mainly Conference on Cold-Formed Steel Structures, St. Louis, Missouri,
USA, October 20–21, 1992.
when applied to joints connected by means of 8 mm F [13] Zaharia R, Dubina D. Behaviour of cold formed steel truss bolted
bolts. joints. Fourth International Workshop on Connections in Steel
When designing, one way to know whether the use of Structures, Roanoke, USA, October 22–25, 2000.
washers in a particular joint is effective is by means of a [14] Rogers CA, Hancock GJ. New bolted connection design formulae for
G550 and G300 sheet steels less than 1.0 mm thick. Research report
calculation formula. Nowadays, there are not specific
no. R769. Centre for Advanced Structural Engineering, Department
formulas for bolted connections with washers in Eurocode of Civil Engineering, The University of Sydney, 1998.
3. Due to the relevance of washers in the seismic behavior [15] Rogers CA, Hancock GJ. Bolted connection tests of thin G550 and
of joints, it may be interesting to modify the net-section G300 sheet steels. J Struct Eng 1998;124(7):798–808.
strength equation to take into account their effect. For [16] Rogers CA, Hancock GJ. Bolted connection design for sheet steels
instance, it can be investigated whether good predictions of less than 1.0 mm thick. J Construct Steel Res 1999;51(2):123–46.
[17] Wallace JA, Schuster RM, LaBoube RA. Testing of bolted cold-
the ultimate load of joints with washers are obtained if the formed steel connections in bearing. Final Report, Canadian Cold
reduction factor is removed from the current version of this Formed Steel Research Group, Department of Civil Engineering,
equation. University of Waterloo, 2001.
Finally, the last section of the paper contains a list of [18] Wallace JA, Schuster RM, LaBoube RA. Calibrations of bolted cold-
recommendations for designing bolted joints suitable for formed steel connections in bearing. Final report, American Iron and
Steel Institute, 2001.
seismic construction. In fact, from the results of the tests [19] Technical Committee 7. European Recommendations for the Testing
performed, it can be concluded that it is better to use of Connections in Profiled Steel Sheeting and Sections. ECCS
screws than bolts to connect the straps of a dissipative Publication No. 21, May 1983.