T - J O I N T S IN RECTANGULAR H O L L O W SECTIONS

SUBJECT TO COMBINED A C T I O N S
Downloaded from ascelibrary.org by UFES - Universidade Federal do Espirito Santo on 10/14/19. Copyright ASCE. For personal use only; all rights reserved.

By Xiao-Ling Zhao 1 and Gregory J. Hancock 2

ABSTRACT: An extensive test program on a range of cold-formed rectangular and

square hollow sections subjected to combined bending and concentrated force is
described. The concentrated force was applied by welding a square hollow section
(called the branch member) to the center of the member under test (called the chord
member) and perpendicular to it. The chord member was simply supported with
the concentrated force at the center of the span. The span was varied in order to
determine the interaction relationship between bending moment and concentrated
force. Tests under pure moment and pure concentrated force were performed for
control purposes. The ratio of the width of the branch member to the chord member
(P) used varied from 0.5 to 1.0 so that failure modes ranging from local bending
and buckling of the face of the chord member to web crippling were observed. In
addition, the thickness of the chord-member sections was varied so as to produce
a range of section slenderness and failure modes from buckling to yielding. The
results are compared with the design procedures in the 1986 AISI Specification,
Australian Standard AS1538 of 1988, CIDECT Monograph No. 6 of 1986, and
other proposed procedures. In addition, a comparison with plastic mechanism models
is provided.

INTRODUCTION

The design of welded connections in tubular structures has been well re-
searched (CIDECT 1986) and is still being heavily investigated (Proc. 1989).
However, a large amount of this research applies mainly to joints in trian-
gulated structures (trusses). Joints can occur in tubular structures where the
main member is subjected to a bending moment rather than axial force, par-
ticularly for rigid-jointed frames. This paper describes tests of joints in this
latter category, including comparisons with design procedures.
In recent years, the design of tubular structures has been included in the
steel design standards for hot-rolled members [Canadian Standards Associ-
ation (Steel 1984), British Standards Institution (Structural 1985), American
Institute of Steel Construction (Load 1986), Standards Association of Aus-
tralia (SAA 1984), Eurocode 3 (Common 1983)] and cold-formed members
[British Standards Institution (Structural 1987), Standards Association of
Australia (Cold-Formed 1988), American Iron and Steel Institute (Specifi-
cation 1986), ECCS (European 1987)]. In these standards, although design
rules exist for tubular members for pure axial force, pure bending, and com-
bined axial force with bending, very little design information is available
for tubular members subject to combined bending and concentrated force.
Rules for this problem in the existing standards mainly apply to I-sections
in the hot-rolled standards, and channel, Z, and hat sections in the cold-
formed standards. It is the purpose of the research program described in this
'Postgrad. Student, School of Civ. and Mining Engrg., Univ. of Sydney, NSW
2006 Australia.
2
BHP Steel Prof, of Steel Struct., School of Civ. and Mining Engrg., Univ. of
Sydney, NSW 2006 Australia.
Note. Discussion open until January 1, 1992. To extend the closing date one month,
a written request must be filed with the ASCE Manager of Journals. The manuscript
for this paper was submitted for review and possible publication on May 3, 1990.
This paper is part of the Journal of Structural Engineering, Vol. 117, No. 8, Au-
gust, 1991. ©ASCE, ISSN 0733-9445/91/0008-2258/$1.00 + $.15 per page. Paper
No. 26048.

2258

J. Struct. Eng., 1991, 117(8): 2258-2277

 paper to attempt to provide design guidance in this area for tubular members.
In this paper, an extensive test program on a range of cold-formed rect-
angular and square hollow sections subjected to pure moment, pure concen-
Downloaded from ascelibrary.org by UFES - Universidade Federal do Espirito Santo on 10/14/19. Copyright ASCE. For personal use only; all rights reserved.

trated force, and combined bending moment and concentrated force is de-
scribed. The parameters varied in the tests were: (1) The ratio of the width
of the branch member to the chord member ((3); (2) the slenderness of the
chord member sections; (3) the span of the simply supported chord members;
and (4) the shape of the sections (i.e. square or rectangular).
The results are compared with design procedure in the 1986 AISI speci-
fication (Specification 1986), the Australian standard AS1538 (Cold-Formed
1988), CIDECT Monograph No. 6, (1986), the British Standard BS5950-
Part 5 (Structural 1987), the AWS (Structural 1990), and a design proposal
by Packer (1984). A comparison with plastic mechanism models is also pro-
vided.

TEST SECTIONS AND PROPERTIES

Selection of Sections and Fabrication

Cold-formed rectangular hollow sections (RHS) and square hollow sec-
tions (SHS) were used in the tests. These sections were produced by cold-
forming and electric-resistance welding and were the same type as described
in earlier column tests by Key et al. (1988) and earlier beam tests by Hasan
and Hancock (1989) for which typical stress-strain curves and residual stress
patterns were given. Six different section sizes (three RHS and three SHS)
were chosen for the chord members and two sizes of SHS were chosen for
the branch members, as shown in Table 1. The section nomenclature is shown
in Fig. 1. Further details of these sections can be found in the manufacturer's
catalog (Tubeline 1984). The nominal yield stress of the material was 350
MPa. The sections were manufactured to the requirements of AS 1163 (Struc-
tural 1981).
The branch members were welded to the chord members by fillet welds
transversely and butt welds longitudinally when p = 1.0; and fillet welds
both longitudinally and transversely when p = 0.5. The measured size (hor-
izontal leg length) of the fillet welds varied from 6.5 mm to 7.0 mm.

Mechanical Properties
Tensile and compression coupons were taken from the flanges, webs, and

TABLE 1. Section Dimensions

Section Section geometry
Type number (D x B x t) Shape (B - It) It
(1) (2) (3) (4) (5)
Chord Cll 102 x 51 X 4.9 RHS 8.37
Chord C12 102 x 51 x 3.2 RHS 13.9
Chord C13 102 x 51 x 2.0 RHS 23.4
Chord C21 102 x 102 X 9.5 SHS 8.69
Chord C22 102 x 102 X 6.3 SHS 14.1
Chord C23 102 x 102 x 4.0 SHS 23.4
Branch Bl 51 x 51 x 4.9 SHS 8.37
Branch B2 102 x 102 x 8.0 SHS 10.7

2259

J. Struct. Eng., 1991, 117(8): 2258-2277

Downloaded from ascelibrary.org by UFES - Universidade Federal do Espirito Santo on 10/14/19. Copyright ASCE. For personal use only; all rights reserved.

: x:
FIG. 1. Section Nomenclature

corners of the sections. The flat tensile coupons were prepared and tested
according to Standards Association of Australia {Methods 1974) (ASTM:
Standard 1989) to determine the yield stress (Fyt), the tensile strength (F,„)
and the percentage of elongation (e„) at the ultimate strength. All the corner
tensile coupons except those from sections C21 and C22, had their ends
flattened in a press and were gripped in the same manner as the flat tensile
coupons. The corner coupons from sections C21 and C22 had holes drilled
in their ends and were tested by loading through pins in the holes. The
compression coupons were rectangular (95 mm X 25 mm), with all edges
being milled, and tested in a greased steel jig of a similar type to that de-
scribed by Karren (1967).
The tensile and compression coupons were tested in a 250 kN capacity
Instron Universal Testing Machine using a calibrated extensometer on a 10
mm gauge length. The average measured values of the tensile yield stress
(Fyt), the tensile strength (F,„) and the percentage of elongation at ultimate
(e„) are presented in Table 2 where they are compared with nominal yield
stress (Fy„) and tensile strength (Fm). The 0.2% proof stress was used as the
yield stress for all coupons since they all exhibited gradual yielding. The
mean tensile strengths of the corner coupons, which was found to be 32.6%
higher than those of the flat coupons, was slightly higher than that normally
measured for specimens of this type (Key and Hancock 1988). The mea-
sured yield stress (Fyc) of the compression specimens taken from the flats is

2260

J. Struct. Eng., 1991, 117(8): 2258-2277

 TABLE 2. Results of Tensile Coupon Tests

Section Fr, F,„ £i,

Downloaded from ascelibrary.org by UFES - Universidade Federal do Espirito Santo on 10/14/19. Copyright ASCE. For personal use only; all rights reserved.

number Position (MPa) (MPa) FJF,, Fvl/F,„ FJF„ (%)

(D (2) (3) (4) (5) (6) (7) (8)
Cll Corner 558 575 1.031 1.594 1.338 8.13
Cll Flat 379 418 1.104 1.083 0.973 13.38
C12 Corner 468 510 1.089 1.336 1.185 10.0
C12 Flat 330 431 1.307 0.942 1.002 19.96
C13 Corner 506 547 1.083 1.444 1.273 10.63
C13 Flat 400 467 1.166 1.143 1.085 14.77
C21 Corner 513 558 1.088 1.466 1.298 4.72
C21 Flat 421 454 1.079 1.202 1.055 8.11
C22 Corner 539 573 1.064 1.539 1.334 4.72
C22 Flat 412 455 1.104 1.176 1.057 15.83
C23 Corner 543 560 1.031 1.548 1.302 8.75
C23 Flat 417 471 1.129 1.192 1.095 17.12
Note: F,„ = 350 MPa; F„ = 430 MPa.

TABLE 3. Results of Compression Coupon Tests

Section number F,c (MPa) Fyc/Fy„ FJFyl

(D (2) (3) (4)
Cll 322 0.919 0.850
C12 325 0.929 0.985
C13 373 1.065 0.933
C21 435 1.242 1.033
C22 363 1.038 0.881
C23 378 1.080 0.906
Note: F,„ = 350 MPa.

given in Table 3. The ratio of the mean measured yield stress in compression
in the flats was found to be 7.5% lower than the mean measured yield stress
in tension in the flats.

PURE CONCENTRATED FORCE TESTS

Specimen Geometry and Test Procedure

The pure concentrated force tests are labeled series 1 (SI). Eight specimen
types were tested under concentrated force alone as set out in Table 4, in
which the specimen number column one indicates the series (SI), the branch
member (Bl or B2), and the chord member (Cll —> C23). The test results
are also given in Table 4, in which Pn and Pf2 are the defined failure loads
of test numbers 1 and 2, respectively. All the tests were performed in a 2,000
kN capacity Dartec servo-controlled testing machine. Loading was applied
at the top end of the branch members as shown in Fig. 2(a). The test spec-
imens were seated on the solid steel base plate of the Dartec testing machine,
which provided support to the specimen along its entire length.
Six displacement transducers were used to measure the deformation of the
2261

J. Struct. Eng., 1991, 117(8): 2258-2277

 TABLE 4. Results of Concentrated Force Tests

Specimen P„, P,a

Downloaded from ascelibrary.org by UFES - Universidade Federal do Espirito Santo on 10/14/19. Copyright ASCE. For personal use only; all rights reserved.

Pfl Pf2 Pf Pu
number (kN) (kN) (Mean) (kN) (kN) (Mean)
(1) (2) (3) (4) (5) (6) (7)
S1B1C11 307 326 316 — — —
S1B1C12 162 164 163 — — —
S1B1C13 77.2 75.4 76.3 — — —
S1B1C21 424 420 422 442 434 438
S1B1C22 176 174 175 244 246 245
S1B1C23 66.7 68.2 67.4 127 142 135
S1B2C21 1,207 — 1,207 — — —
S1B2C22 652 — 652 — — •
—

web and top flange, as shown in Fig. 2(b). Transducers 3, A,. 5, and 6 were
located on the specimen centerline. A Spectra automatic data-acquisition sys-
tem was used to record all the transducer readings.

Results
For (3 = 1.0, Pf is the maximum load in a typical load-deflection (8m)
(connection deformation of the top flange at the specimen centerline deter-
mined using the mean value of the readings from gage 3 and gage 4) curve,
as shown in Fig. 3. For p = 0.5, Pf is the yield load (Py) determined from
the intersection of two tangents in a typical load-deflection (8,„) curve, as
shown in Fig. 4. The values of Pu, the ultimate load, are also given in Table

<") Loading Ram

L
Branch Member Chord Member

Steel Base Plate

of Testing Machine

L=800mm

(b)
Small Plate Stuck on
the Branch Member

FIG. 2. Pure-Concentrated-Force Tests: (a) Schematic View of Test Arrange-

ment; (b-c) Transducer Arrangement

2262

J. Struct. Eng., 1991, 117(8): 2258-2277

 400
Downloaded from ascelibrary.org by UFES - Universidade Federal do Espirito Santo on 10/14/19. Copyright ASCE. For personal use only; all rights reserved.

Pf=307 kN
300

f
I 200

I • S1B1C11

100

0.0 2.0 4.0 6.0 8.0 10.0 12.0

Centreline Deflection on Top Flange 8m (mm)

FIG. 3. Typical Load-Deflection Curve (0 = 1.0)

4 for the (3 = 0.5 specimens. The failure modes are shown in Figs. 5. As
(3 varies from 0.5 to 1.0 the failure modes vary from local bending of the
face of the chord member, as shown in Fig. 5(a), to web buckling, as shown
in Fig. 5(b).

200

150

100
Pf=66.7kN _.
. S1B1C23

50

0.0 10.0 . 20.0 30.0 40.0

Centreline Deflection on Top Flange ' 8m (mm)

FIG. 4. Typical Load-Deflection Curve (0 = 0.5)

2263

J. Struct. Eng., 1991, 117(8): 2258-2277

 PURE BENDING TESTS
Downloaded from ascelibrary.org by UFES - Universidade Federal do Espirito Santo on 10/14/19. Copyright ASCE. For personal use only; all rights reserved.

Specimen Geometry and Test Procedure

The pure bending tests are labeled series 2 (S2). Two beam tests were
performed for each of the six sections (Cll —» C23) in Table 1. The tests
were carried out in a 2,000 kN capacity Dartec servo-controlled testing ma-
chine, and the schematic view of the general test setup is shown in Fig. 6.
The test specimens were supported on half rounds placed upon Teflon pads
to simulate the simply supported end conditions in a manner similar to that
described by Hasan and Hancock (1989).
Four strain gages were mounted at midspan, two on the top flange, and
two on the bottom flange of each specimen to measure the longitudinal strains
adjacent to the edges of the RHS faces, as shown in Fig. 6. Three displace-
ment transducers were used to measure the vertical deflection at the midspan
and under the loading points, as shown in Fig. 6. A Spectra automatic data-

idi •

[ * = l)o j

.-; ^ l.()

i
i

FIG. 5. Failure Modes: (a) p = 0.5; (b) (3 = 1.0

2264

J. Struct. Eng., 1991, 117(8): 2258-2277

 Loading Ram
Downloaded from ascelibrary.org by UFES - Universidade Federal do Espirito Santo on 10/14/19. Copyright ASCE. For personal use only; all rights reserved.

Strain Gauge Pairs

Spreader Beam
Test specimen(RHS)
Strain Loading Plate
Gauges
at Midspan ^•Fillet Weld
y ^ -Support Plate
^

Support
Transducers

FIG. 6. Schematic View of Pure-Bending Test Arrangement

acquisition system was used to record all the strain-gage and transducer read-
ings.

Results
The maximum bending moment (Mmax) obtained from each of the tests is
listed in Table 5 and is compared with both its nominal plastic moment
(Mpn), which is based on the nominal yield stress of (Fyn) 350 MPa, the
nominal section dimensions, and thickness; and its theoretical plastic mo-
ment (Mpl), which is bsaed on the measured tensile yield stress of the flats
(Fj,) and the actual measured overall dimensions and thickness. The second
test for each section size was performed several months after the first and
the approximately constant increase in capacity is believed to be attributable
to strain aging. The tensile and compression coupon tests described in the
section headed "Mechanical Properties" were performed approximately the
same time as the second test for each section size.
Graphs of dimensionless bending moment versus dimensionless curvature
are shown in Fig. 7 for the first test performed for each section size. The

TABLE 5. Results of Pure Bending Tests

Specimen M mral M m „x2 Mraax

number (kNm) (kNm) (Mean) MmJMpn Mm„/Mpl
(D (2) (3) (4) (5) (6)
S2C11 17.6 19.9 18.8 1.32 1.21
S2C12 11.5 12.4 11.9 1.19 1.25
S2C13 7.57 8.25 7.91 1.18 1.04
S3C21 53.1 56.1 54.6 1.47 1.22
S2C22 35.3 39.9 37.6 1.36 1.15
S2C23 23.7 25.5 24.6 1.29 1.08

2265

J. Struct. Eng., 1991, 117(8): 2258-2277

 1.6
•
Downloaded from ascelibrary.org by UFES - Universidade Federal do Espirito Santo on 10/14/19. Copyright ASCE. For personal use only; all rights reserved.

A 1 1-
1.4
,
1.2
J £ J

J\
1.0

0.8
w .ocal Buckling

B/D=1.0
+ S2C21 "
n S2C22
0.6 T S2C23
¥ B/D=0.5
0.4 « S2C11
A S2C12
1 ' S2C13
0.2

0.0
0.0 2.0 4.0 6.0 10.0
Dimensionless Curvature K/Kp

FIG. 7. Dimensionless Curvature versus Dimensionless Moment

values of the curvature (K) were determined from both the strain measure-
ments and deflection measurements in the same manner as described by Hasan
and Hancock (1989). There was very good correlation between the values
of curvature determined by these two methods except in the case of specimen
S2C13, which buckled locally after which point the average curvature based
on the deflection was used. The curvatures used in Fig. 7 were determined
from deflection measurements. The increase in moment over the nominal
capacity is partly attributable to the strain hardening at low values of strain
and the strength-enhancing effects of cold-forming and partly attributable to
the increase in the yield stress (Table 2) above the nominal value of 350
MPa.

COMBINED COMPRESSION AND BENDING TESTS

Specimen Geometry and Test Procedure

The combined compression and bending (interaction) tests are labeled se-
ries 3 (S3). A total of 38 interaction tests were performed as set out in Table
6, in which the specimen number (column two) indicates the series (S3), the
branch member (Bl or B2), the chord member (Cll —> C23), and the span
(A2 —> A0.5), as described in the following.
The specimens listed in Table 6, excluding those marked with asterisks,
were tested in a 2,000 kN capacity Dartec servo-controlled testing machine.
The specimens marked with a superscript a in Table 6 was tested in a 1,200
kN capacity Avery testing machine and those with a superscript b in Table
6 were tested in a 400 kN capacity Mohr & Federhaff testing machine. These
tests were all short-span tests. The same technique was used to simulate the
simply supported-end conditions as used for the pure-bending tests. How-
ever, as shown in Fig. 8(a), loading was applied by the single branch mem-
2266

J. Struct. Eng., 1991, 117(8): 2258-2277

 TABLE 6. Dimension and Failure Loads of Specimens (Interaction Tests)
Specimen a L h Pfi Pfi Pf
Downloaded from ascelibrary.org by UFES - Universidade Federal do Espirito Santo on 10/14/19. Copyright ASCE. For personal use only; all rights reserved.

p number (mm) (mm) (mm) (kN) (kN) (Mean)

(1) (2) (3) (4) (5) (6) (7) (8)
1.0 S3B1C11A2 230 1,200 200 190 202 196
1.0 S3BlCllAlb 115 530 100 312 312 312
1.0 S3BlCllA0.5 b 57 414 100 320 320 320
1.0 S3B1C12A2 286 1,200 200 99.8 101 100
1.0 S3B1C12A1" 143 586 100 157 156 156
1.0 S3BlC12A0.5 b 72 444 100 161 161 161
1.0 S3B1C13A2 392 1,200 200 45.5 44.5 45.0
1.0 S3B1C13A1 196 860 200 73.4 71.3 72.3
1.0 S3B1C13A0.5" 98 496 100 73.2 73.4 73.3
1.0 S3B2C21A2" 176 652 200 910 — 910
1.0 S3B2C22A2 218 1,200 200 437 — 437
0.5 S3B1C21A2 500 1,470 200 238 238 238
0.5 S3B1C21A1 250 1,200 200 381 382 381
0.5 S3B1C21A0.5" 125 550 200 399 398 399
0.5 S3B1C22A2 836 2,400 200 89.9 89.5 89.7
0.5 S3B1C22A1 418 1,300 200 143 143 143
0.5 S3B1C22A0.5 209 1,200 200 166 163 165
0.5 S3B1C23A1.5 1,080 2,400 200 40.4 41.0 40.7
0.5 S3B1C23A0.75 540 1,570 200 58.9 60.2 59.5
0.5 S3B1C23A0.5 360 1,500 200 62.2 64.5 63.3
"Tested in a 1,200 kN capacity Avery testing machine.
b
Tested in a 400 kN capacity Mohr & Federhaff testing machine.

ber rather than two-point loading. Two strain gages were mounted on the
bottom flange at the center of each specimen to measure the longitudinal
strains adjacent to the edges of the RHS faces. Seven displacement trans-
ducers were used to measure the deformation on the web and flange and the
vertical deflection at the midspan, as shown in Fig. 8(b). Transducers 3, 4,
5, 6, and 7 were located on the specimen centerline. Transducers 5 and 6
were fixed on the web by using a magnetic base. A Spectra automatic data-
acquisition system was used to record all the strain-gage and transducer read-
ings. A photograph of the testing arrangement in the Dartec rig is shown in
Fig. 9.
The span of the simply supported chord members was varied in order to
determine the interaction relationship between moment and concentrated force.
The span used in the tests was calculated according to

a = (1)
—
in which, a = distance from the center of the chord member to the sup-
porting point as shown in Fig. 8(a), Mmax = maximum bending moment
determined from pure-bending tests; and Pf = failure load determined from
the pure-concentrated force tests.
The factor k was chosen in the tests to allow determination of the inter-
action of bending moment and concentrated force over a range of their ratio.
2267

J. Struct. Eng., 1991, 117(8): 2258-2277

 (a) Loading Ram
Downloaded from ascelibrary.org by UFES - Universidade Federal do Espirito Santo on 10/14/19. Copyright ASCE. For personal use only; all rights reserved.

Test Specimen(RHS)
Branch Member
Fillet Weld

Half Round
•Teflon Pad r~~~i
Support

L _ J

(b) n (c) 3

V X
Small Plate Stuck on
the Branch Member

-Si

FIG. 8. Interaction Tests: (a) Schematic View of Test Arrangement; (b-c) Trans-
ducer Arrangement

Hence tests with k = 1, correspond to maximum combined moment and

concentrated force and are called Al in the nomenclature. Tests with k =
0.5 had a lower ratio of moment to concentrated force and are called A0.5
in the nomenclature. Tests with k — 2 had a greater ratio of moment to
concentrated force and are called A2 in the nomenclature. Other tests, with
k = 0.75 and 1.5, are called A0.75 and A1.5, respectively, in the nomen-
clature.

Results and Comment

Typical load versus connection deformation (8,„) curves are compared with
those of the pure-concentrated-force tests in Figs. 10. The values of 8m were
the difference between the reading of gage 7 and the mean readings of gages
3 and 4 in Fig. 8(b). For P = 1.0 [Fig. 10(a)], the web-crippling capacity
is reduced. For p = 0.5 [Fig. 10(b)], the effect of bending moment reduces
the failure load and the curve becomes flatter after failure.
The interaction diagrams are shown in Fig. 11(a) (P = 1.0) and Fig. 11(b)
(P = 0.5). The normalizing values Afmax and Pf in the diagram are the test
values (A/max2 in Table 5 and Pf2 in Table 4) for the pure moment and the
2268

J. Struct. Eng., 1991, 117(8): 2258-2277

Downloaded from ascelibrary.org by UFES - Universidade Federal do Espirito Santo on 10/14/19. Copyright ASCE. For personal use only; all rights reserved.

FIG. 9. Test Arrangement in Dartec

pure concentrated force cases, respectively. For all tests, the web-crippling
capacity is reduced by the effect of bending moment producing longitudinal
stress in the web of the RHS section. The increase in capacity above M =
Mmax is most likely caused by the finite size of the branch member welded
to the chord member. For (3 = 1.0, increasing the slenderness of the chord
member (Cll —> CI3) has a very small effect on the interaction diagram.
For M/Mmlx less than approximately 1.0, there is very little effect on the
ratio P/Pf. For (3 = 0.5, increasing the section slenderness of the chord
member (C21 —» C23) has a greater effect on the interaction diagram. For
M/Mmax less than 0.5, there is very little interaction effect on the ratio P/
Pf. When A//Mmax exceeds 0.5, the effect of the slenderness of the chord
member on the degree of interaction becomes significant.
The following interaction formula is proposed for design (3 < 1.0:
M P
( V + f —V < 1.0 (2)

in which the value of M'max = theoretical value of the fully plastic bending
moment determined from (S x Fy), where S = plastic section modulus; and
Fy can be the nominal tensile yield stress (Fy„) or measured tensile yield
stress (Fyl). The theoretical interaction diagram based on Fy„ is more con-
servative than that based on Fyt. The value of P'f is the theoretical value of
the failure load under concentrated force alone, which can be determined
from various plastic mechanism models such as the CIDECT (1986) model
or Kato model described in the section headed "Comparison with Plastic
Mechanism Models" or determined from other theoretical analysis (e.g., fi-
nite-element analyses). The more precise the theoretical analysis to deter-
mine Pf, the better the prediction of the interaction diagram.
In Table 10.2 of section 10.5.1 of the AWS Structural Welding Code
(1990), a stress-interaction term (Qf) is specified for square and rectangular
chord sections. The term Qf is a linear function of the utilization ratio U,
2269

J. Struct. Eng., 1991, 117(8): 2258-2277

Downloaded from ascelibrary.org by UFES - Universidade Federal do Espirito Santo on 10/14/19. Copyright ASCE. For personal use only; all rights reserved.

\r
V .
0.0 2.0 4.0
• S1B1C12
» S3B1C12A2

6.0 8.0

(a) Connection Deformation Sm (mm)

o
*"""" • " r T W " ™ ^ T • T I
. S1B1C23
50 A. S3B1C23A0.5
» S3B1C23A0.75
. S3B1C23A1.5

10.0 20.0 30.0 40.0

(b) Connection Deformation 8m (mm)

FIG. 10. Typical Load versus Connection Deformation: (a) p = 1.0; (b) (3 = 0.5

which depends upon the actual to allowable longitudinal compression stress

in the chord member. The resulting interaction is based on the yield moment
(My) rather than the maximum moment (Mmax) in the chord member and
therefore cannot simply be plotted in Figs. 11. However, it produces graphs
that are slightly lower than either the test results or the circular interaction
formula, (2), in the region of the greatest interaction.

COMPARISONS OF CONCENTRATED FORCE TESTS WITH DESIGN

PROCEDURES AND THEORETICAL MODELS

Comparison with Proposed Design Procedures

For p = 1.0, the test values of the web-crippling load are compared with
those determined from the proposed design procedures listed in Appendix-
D of Zhao and Hancock (1990). The formulas in the AISI Specification (1986)
2270

J. Struct. Eng., 1991, 117(8): 2258-2277

 1.6

•
Downloaded from ascelibrary.org by UFES - Universidade Federal do Espirito Santo on 10/14/19. Copyright ASCE. For personal use only; all rights reserved.

1.4

• -
1.2

I 1.0
• I
I 0.8

1
'§ 0.6
1
3
3=1.0
0.4 + S3B2C21
T S3B2C22
•
s S3B1C11
i S3B1C12
T S3B1C13
0.2

0.0
0.00 0.20 0.40 0.60 0.80 1.00
(a) Dimensionless Loading P/Pf

FIG. 11(a). Dimensionless Loading versus Dimensionless Moment

and the Australian standard AS1538 (Cold-formed 1988) are based on the
permissible (allowable) stress method. The formula in BS5950 (Structural
1985) is in an ultimate limit-state format derived from the AISI Specification
factored by 1.76. Packer's (1984) formula is in an ultimate limit-state for-
mat. The formulas in CIDECT (1986) and AWS (Structural 1990) are the
ultimate limit state. Comparisons are presented in Table 7. The yield stresses
(/v,FK,F,.„,pv) used in (2) and (3) are the values determined from tensile
coupon tests of the flat material (see Table 2). The value of Pf is the failure
load determined from the pure-concentrated-force tests as given in Table 4.
In CIDECT Monograph No. 6 (1986) and the AWS standard (Structural
1990) better prediction was obtained using an effective length factor for the
web Ke = 0.5 since the reduced effective length better represents the situ-
ation with restraint against web rotation. The formulas from BS5950, the
AISI specification, and AS1538 used in this paper were those considering
restraint against web rotation. BS5950 gives a. better prediction (Pp/Pf =
0.97) than CIDECT Monograph No. 6 or the AWS standard. For the AISI
specification and AS1538, the computed mean factor of safety is 1/0.551

2271

J. Struct. Eng., 1991, 117(8): 2258-2277

 1.4
Downloaded from ascelibrary.org by UFES - Universidade Federal do Espirito Santo on 10/14/19. Copyright ASCE. For personal use only; all rights reserved.

(b) Dimensionless Loading P/Pf

FIG. 11(b). Dimensionless Loading versus Dimensionless Moment

= 1.81, which appears slightly conservative when compared with the usual
factor of safety of 1.76 in permissible stress codes. If the formula for shapes
having single webs (i.e. no restraint against web rotation) in BS5950 is used,
the ratio Pp/Pf would be reduced to 0.765. Hence for T-joints, the restraint

TABLE 7. Comparison with Proposed Design Procedures

Pp/Pf
AISI
(Specification
1986) and
BS5950 AS1538
CIDECT (*y AWS (Ay
Packer (Structural (Cold-formed
number 1.0 0.76 0.72 0.5 1.0 0.76 0.72 0.5 (1984) 1987) 1988)
(D (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)
S1B2C21 0.756 0.770 0.772 0.776 0.912 0.944 0.949 0.970 0.933 1.297 0.737
S1B2C22 0.835 0.888 0.895 0.924 0.861 0.942 0.953 1.008 0.842 0.981 0.557
S1B1C11 0.782 0.880 0.893 0.948 0.630 0.728 0.743 0.814 0.845 0.982 0.558
S1B1C12 0.557 0.789 0.826 0.978 0.410 0.563 0.589 0.719 0.692 0.845 0.480
S1B1C13 0.274 0.447 0.489 0.770 0.192 0.333 0.371 0.695 0.806 0.745 0.423
Mean 0.641 0.755 0.775 0.879 0.601 0.702 0.721 0.841 0.824 0.970 0.551
Coefficient
of variation 0.207 0.161 0.150 0.088 0.272 0.234 0.222 0.128 0.078 0.186 0.106

2272

J. Struct. Eng., 1991, 117(8): 2258-2277

 TABLE 8. Comparison of Tests and Yield-Line Theory
Specimen F„ F„, Pv P,
Downloaded from ascelibrary.org by UFES - Universidade Federal do Espirito Santo on 10/14/19. Copyright ASCE. For personal use only; all rights reserved.

number (MPa) (MPa) (kN) (kN) Pev/Pv PJP» Ph/Py Pku/P« pyp-, PL/Pu
(D (2) (3) (4) (5) (6) (7) (8) 0) (10) (11)
S1B1C21 421 454 422 438 0.690 0.716 1.367 1.418 0.915 0.949
S1B1C22 412 455 175 245 0.716 0.564 1.191 0.939 0.950 0.749
S1B1C23 417 471 67.4 135 0.758 0.428 1.145 0.647 1.005 0.568
Mean 1.234
Coefficient
— — — — 0.721 0.569 1.001 0.957 0.755

of variation — — — — 0.028 0.118 0.096 0.318 0.037 0.156

Note: Py and Pu were determined from T-joint tests (see Table 4). Fy, and Fut were determined from tensile coupon tests
(see Table 2).

against web rotation due to the welding of the branch member to the chord
member can be regarded as adequate for the design rules in BS5950, the
AISI specification, and AS 1538. From Table 7 it can be seen that as the
section web slenderness increases, all the design formulas become more con-
servative, especially for one case in Packer's (1984) formula for the speci-
men S1B1C12, whose measured tensile yield stress was unusually low (330
MPa).

Comparison with Plastic Mechanism Models

For p = 0.5, the concentrated-force tests presented in this paper are com-
pared with the plastic mechanism analysis summarized in CIDECT Mono-
graph No. 6 (1986) and by Kato (1980) and also with a modified version
of the Kato method. The formulas are listed in Appendix-C of Zhao and
Hancock (1990). The Kato model is essentially the same as the CIDECT
model except that the dimensions of the mechanism were altered to allow
for the finite size of the fillet welds and round corners. The loads predicted
by the models are compared with the test results in Table 8. The size of the
fillet welds used in the Kato model is the measured value of 7.0 mm. The
values of Py and Pu in Table 8 are the values of Pf and P„, respectively, in
Table 4. The predicted loads P^, P^, and P'^ are the yield loads predicted
by the CIDECT model, Kato model, and modified Kato model, respectively,
in which the tensile yield stress (F,,) was used in their calculation. The pre-
dicted loads Pcu, Pku, and P'ku are the ultimate loads predicted by the CIDECT
model, Kato model, and modified Kato model, respectively, in which the
tensile strength (Fut) was used in their calculation. The ultimate models do
not account for membrane action.
From Table 8, it can be seen that the CIDECT model is conservative and
the Kato model is unconservative when predicting the yield load. The rea-
sons for the conservatism of the CIDECT model were described in detail in
CIDECT Monograph No. 6 but can be summarized as the effect of mem-
brane forces after large deflections and the effect of local thickening of the
corners on the negative plastic moments along the corners. The reasons for
the unconservativeness of the Kato model is most likely due to consideration
of the hinges on the center of the corners, which did not occur in the tests
as a result of material strain hardening in the corners. The hinges were found
to be located on the web adjacent to the corners. A very simple modification
of the Kato model is to move the hinges from the center of the corners to
the top of the web adjacent to the corners so that bb — h§ as in the CIDECT
model. The yield load (Pj^) and ultimate load (P'k„) based on this modification

2273

J. Struct. Eng., 1991, 117(8): 2258-2277

 are also compared with the test results in Table 8, and produce more accurate
estimates of the failure loads. The prediction of the ultimate loads based on
Fw are less reliable than those based on Fy, probably as a consequence of
Downloaded from ascelibrary.org by UFES - Universidade Federal do Espirito Santo on 10/14/19. Copyright ASCE. For personal use only; all rights reserved.

the effect of membrane stresses on the ultimate load, which were not ac-
counted for in the plastic-mechanism models. Further investigations of plas-
tic mechanisms for both yield and ultimate conditions are being carried out
in the School of Civil and Mining Engineering at the University of Sydney,
Australia.
In section 10.5.1.4 of the AWS Structural Welding Code (1990), yield-
line analysis may be used in lieu of the punching shear method for stepped-
box connections having (3 less than 0.8. Hence the modified Kato model
would be acceptable according to the AWS Structural Welding Code.

CONCLUSIONS

An extensive test program on a range of cold-formed rectangular and square

hollow sections subjected to pure moment, pure concentrated force, and
combined bending moment and concentrated force was described in this pa-
per. The parameters varied in the tests were: (1) The ratio of the width of
the branch member to the chord member ((3); (2) the slenderness of the chord-
member sections; (3) the span of the simply supported chord members; and
(4) the shape of the sections (i.e. square or rectangular).
The following observations and conclusions are made.
In the pure-concentrated-force tests, as (3 varies from 0.5 to 1.0 the failure
mode varies from local bending of the face of the chord to web buckling.
For (3 = 1.0, comparison was made in Table 7, with some proposed design
procedures based on web-crippling formulas. The use of an effective length
factor (Ke) of 0.5 with both the CIDECT Monograph No. 6 formulas and
AWS formulas (Structural 1990) produce conservative results. The formulas
in BS5950 Part 5 (Structural 1985), which are the same as those in the AISI
Specification (1986) and AS1538 (Cold-Formed 1988) factored by 1.76, pro-
duce the best estimates of the failure loads. The formulas used in BS5950
are those that consider restraint against web rotation. For (3 = 0.5, com-
parison was made in Table 8 with the plastic-mechanism analysis summa-
rized in the CIDECT Monograph No. 6 and by Kato. The CIDECT model
is too conservative and the Kato model is unconservative when predicting
the yield load. A simple modification of the Kato model gives better results
in predicting the yield load.
In all the pure bending tests, the maximum bending moment attained by
a specimen exceeded the nominal plastic moment (Mp„). This was partly
attributable to the strain hardening at low values of strain and the strength-
enhancing effects of cold-forming and partly attributable to the increase in
the yield stress (Table 2) above the nominal value of 350 MPa.
In all the interaction tests, the increase in capacity above Mmax is most
likely due to the effect of the finite size of the branch member welded to
the chord member. For the tests with p = 1.0, increasing the slenderness
of the chord member has a very small effect on the degree of interaction
shown in the diagram. For M/Mm:ix less than approximately 1.0, there is
very little effect on the ratio P/Pf. For the tests with p = 0.5, increasing
the section slenderness of the chord member results in greater interaction
between bending and concentrated force. For M/Mmax less than 0.5, there
2274

J. Struct. Eng., 1991, 117(8): 2258-2277

 is very little effect on the ratio P/Pf. However, when M/Mmax exceeds 0.5,
the degree of interaction becomes significant particularly for more slender
Downloaded from ascelibrary.org by UFES - Universidade Federal do Espirito Santo on 10/14/19. Copyright ASCE. For personal use only; all rights reserved.

sections. Until further research in this area is complete, a circular interaction

formula is proposed that is a lower bound to all of the results for the sections
tested.

ACKNOWLEDGMENT

The experiments were carried out in the J. W. Roderick Materials and

Structures Laboratory in the School of Civil and Mining Engineering at the
University of Sydney, Australia. The writers are grateful to Tubemakers of
Australia Limited for supplying the test specimens and additional financial
support. Thanks are given to Australian International Development Assis-
tance Bureau for the AIDAB Award, Sydney University for the UPRA
Scholarship, and the Centre for Advanced Structural Engineering of Sydney
University, Australia, for additional support.

APPENDIX I. REFERENCES

C1DECT monograph no. 6. (1986). "The strength and behavior of statically loaded
welded connections in structural hollow sections." Comite International pour le
Development et I'etude de la Construction Tubulaire, CIDECT, Corby, England.
Cold-formed steel structures code. (1988). "Australian standard AS 1538." Standards
Association of Australia, Sydney, Australia.
Common unified code of practice for steel structures, Draft. (1983). "Eurocode 3."
Commission of the European Communities (EC), Brussels, Belgium.
European recommendation for the design of light gauge steel members, No. 49. (1987).
1st Ed., "Design of cold-formed steel sheeting and sections," European Conven-
tion for Constructional Steelwork (ECCS), Brussels, Belgium.
Hasan, S. W., and Hancock, G. J. (1989). "Plastic bending tests of cold-formed
rectangular hollow sections." J. Australian Inst, of Steel Constr., Sydney, Aus-
tralia, 23(4), 2-19.
Karren, K. W. (1967). "Corner properties of cold-formed steel shapes." / . Struct.
Div., ASCE, 93(1), 401-432.
Kato, B. (1980). "T-joints made of rectangular tubes." Proc, 5th Int. Specialty
Conf. on Cold-formed Steel Struct., St. Louis, Mo.
Key, P. W., Hasan, S. W., and Hancock, G. J. (1988). "Column behavior of cold-
formed hollow sections." J. Struct. Engrg., ASCE, 114(2), 390-407.
Load and resistance factor design specification for structural steel buildings. (1986).
American Institute of Steel Construction (AISC), Chicago, 111.
Methods for tensile testing of metals. (1974). "Australian standard AS1391," Stan-
dards Association of Australia, Sydney, Australia.
Packer, J. A. (1984). "Web crippling of rectangular hollow sections." / . Struct.
Engrg., ASCE, 110(10), 2357-2373.
Proc, Int. Symp. on Tubular Struct. (1989). Lappeenranta Univ., Lappeenranta,
Finland.
SAA steel structures code. (1984). "Amendment No. 2." Standards Association of
Australia, Sydney, Australia.
Specifications for the design of cold-formed steel structural members. (1986). Amer-
ican Iron and Steel Institute (AISI), Washington, D.C.
Steel structures for buildings (limit state design): CAN3-S16.1-M84. (1984), Cana-
dian Standards Association, Rexdale, Ontario, Canada.
Steel structures. (1990). "Australian standard AS4100." Standards Association of
Australia, Sydney, Australia.
Structural Steel Hollow Sections. (1981). "Australian standard AS1163," Standards
Association of Australia, Sydney, Australia.

2275

J. Struct. Eng., 1991, 117(8): 2258-2277

 Standard test methods and definitions for mechanical testing of steel products, ASTM
A370-89. (1989). American Society for Testing and Materials (ASTM), Philadel-
Downloaded from ascelibrary.org by UFES - Universidade Federal do Espirito Santo on 10/14/19. Copyright ASCE. For personal use only; all rights reserved.

phia, Pa.
Structural use of steelwork in building. (1985). "Part 1: Code of practice for design
in simple and continuous construction: hot rolled sections: BS5950," British Stan-
dards Institution, London, England.
Structural use of steelwork in building. (1987). "Part 5: Code of practice for design
of cold-formed sections: BS5950," British Standards Institution, London, England.
Structural welding code—Steel, ANSI/AWS Dl.1-90. (1990). 10th Ed., American
Welding Society (AWS) Inc., Miami, Fla.
Tubeline RHS safe load tables, AS1163 (grade 350). (1984). Tubemakers of Aus-
tralia Limited, Sydney, Australia.
Zhao, X. L., and Hancock, G. J. (1990). "T-joints in rectangular hollow sections
subject to combined bending and concentrated force." Research Report, R616,
University of Sydney, Sydney, Australia.

APPENDIX II. NOTATION

The following symbols are used in this paper:

a = distance from center of chord member to supporting point

[see Fig. 8(a)];
B = width of SHS or RHS;
b0 = width of chord member section;
by = width of branch member section;
D = depth of SHS or RHS;
E = Young's modulus of elasticity (200,000 MPa);
eu = percentage elongation at ultimate stage;
Fm = nominal tensile strength;
F„, = measured tensile strength;
Fy„ = nominal tensile yield stress;
Fyl = measured tensile yield stress;
h = height of branch member;
ho = depth of chord member section;
hi = depth of branch member section;
k = interaction factor (see section headed "Specimen Geometry
and Test Procedure";
L = whole length of chord member [see Fig. 8(a)];
M = bending moment;
MmEX = maximum bending moment;
AC« = theoretical prediction of fully plastic bending moment;
Mpn,Mpl = fully plastic bending moment based on Fylt and Fyl> respec-
tively;
P = load;
Pcy = yield load determined from CIDECT (1986) model;
Pcu = ultimate load determined from CIDECT (1986) model;
Pfi>Pf2>Pf ~ failure load determined from T-joint tests;
P'f = theoretical prediction of failure load;
=
Pky,P'ky yield load determined from Kato (1980) model and modi-
fied Kato model;
Pku,P'ku = ultimate load determined from Kato (1980) model and
modified Kato model;

2276

J. Struct. Eng., 1991, 117(8): 2258-2277

 Pp = load capacity determined from proposed design procedures;
P„ = ultimate load determined from pure-concentrated-force tests;
Downloaded from ascelibrary.org by UFES - Universidade Federal do Espirito Santo on 10/14/19. Copyright ASCE. For personal use only; all rights reserved.

S = plastic section modulus;

s = size of fillet weld;
t = thickness of SHS or R H S ;
t0 = thickness of chord member section;
fi = thickness of branch member section;
(3 = ratio of width of branch member to chord member;
Sm = connection deformation of top flange relative to bottom flange
at specimen centerline or at midspan;
K = curvature of beam;
KP = nominal curvature of beam; and
o-j, = tensile yield stress.

2277

J. Struct. Eng., 1991, 117(8): 2258-2277