Numerical Investigation On Stress Concentration Factors of Square Bird-Beak SHS T-Joints Subject To Axial Forces

You might also like

You are on page 1of 11

Thin-Walled Structures 94 (2015) 435–445

Contents lists available at ScienceDirect

Thin-Walled Structures
journal homepage: www.elsevier.com/locate/tws

Numerical investigation on stress concentration factors


of square bird-beak SHS T-joints subject to axial forces
Bin Cheng a,b, Qin Qian c, Xiao-Ling Zhao d,n
a
Department of Civil Engineering, Shanghai Jiao Tong University, Shanghai 200240, China
b
State Key Laboratory of Ocean Engineering, Shanghai Jiao Tong University, Shanghai 200240, China
c
China Shanghai Architectural Design & Research Institute Co., Ltd., Shanghai 200063, China
d
Department of Civil Engineering, Monash University, Clayton, VIC 3800, Australia

art ic l e i nf o a b s t r a c t

Article history: This paper presents a numerical study on stress concentration square bird-beak square hollow section
Received 3 August 2014 (SHS) welded joints. Both conventional and square bird-beak SHS joints considered have T-shapes and
Accepted 5 May 2015 are simply supported at their chord ends. Refined finite element models are developed to obtain the
Available online 28 May 2015
strain/stress concentration factors (SNCFs/SCFs) of square bird-beak joints with various dimensions. The
Keywords: FE models are validated by comparing with the experimental data. The SNCF differences among
Square hollow section considered hot spots are analyzed for the identification of critical locations. The influences of three major
Bird-beak joint non-dimensional parameters, i.e., brace/chord width ratio β, chord wall slenderness ratio 2γ, and brace/
Axial load chord wall thickness ratio τ, on the stress concentration factors (SCFs) of square bird-beak T-joints are
Hot spot stress
revealed on the basis of numerous parametric studies. Comparisons of joint types are finally made. The
Stress concentration factor
results indicate that, in case of identical non-dimensional parameters, square bird-beak SHS T-joints
Finite element analysis
provide SCFs smaller than conventional SHS T-joints in most occasions, especially when β is small; and
SCFs of square bird-beak T-joints are expected to be lower than CHS T-joints with small β and large
2γ and τ.
& 2015 Elsevier Ltd. All rights reserved.

1. Introduction than those of conventional joints with the same non-dimensional


parameters. The seismic resistance of bird-beak joints was also
Bird-beak joint is an innovative type of tubular constructions studied by Hall and Owen [12], where both conventional and
composed of square hollow sections (SHS). Different from the diamond bird-beak X-joints were tested under inelastic cyclic
conventional SHS-to-SHS welded joints where the chord walls are loading, and better performance were observed for new config-
parallel or perpendicular to the brace walls, the angles between urations. The actual reason for the better mechanical behaviors
chord and brace walls of a bird-beak joint are oblique. In fact, a was supposed to be that the bird-beak orientation would assist in
bird-beak joint is generated by simply rotating the members of a transferring forces between members by natural in-plane action
conventional SHS joint at 451 about theirs longitudinal axes. rather than by bending of the walls. However, limited references
Therefore, there are two types of bird-beak SHS T-joints, that is, are available for the fatigue behavior of bird-beak joints. Ishida
square bird-beak joint generated by rotating only chord and [13] carried out the earliest fatigue tests of bird-beak T-joints with
diamond bird-beak joint generated by rotating both chord and the load case of brace axial force being considered. Keizer [14] and
brace, as shown in Fig. 1. These innovative joints obviously have Keizer et al. [15] investigated the stress concentration factors of
more elegant configurations, and would assist in relieving lateral diamond bird-beak joints under brace axial force. Tong et al. [16]
wind loads. presented an experimental study on stress concentration factors
Tremendous efforts including Ono et al. [1–3], Ishida et al. [4], for diamond bird-beak T-joints under axial force and in-plane
Davies et al. [5], Owen et al. [6,8], Owen and Kelly [7], Chtistitas bending on the brace. Cheng et al. [17] carried out experiments to
[9], Lei [10], Zhu and Liu [11] have been devoted to the ultimate determine the strain concentration factors of square bird-beak SHS
resistance of bird-beak joints under static loads, and it shows that T-joints under chord and brace axial forces. Compared with
the ultimate strengths of bird-beak joints are obviously higher abundant products for fatigue of conventional SHS-to-SHS joints
that have been presented in literatures [18–20] and then docu-
mented in the current design codes such as IIW [21] and CIDECT
n
Corresponding author. Tel.: þ 61 3 99054972. Design Guide No. 8 [22], existing fatigue research with regard to
E-mail address: ZXL@monash.edu (X.-L. Zhao). bird-beak joints are far from systematic and complete. That is the

http://dx.doi.org/10.1016/j.tws.2015.05.003
0263-8231/& 2015 Elsevier Ltd. All rights reserved.
436 B. Cheng et al. / Thin-Walled Structures 94 (2015) 435–445

Fig. 1. Welded SHS–SHS T-joints: (a) conventional; (b) square bird-beak; and (c) diamond bird-beak.

reason why very few bird-beak joints are used in the structures
t1
dominated by cyclic repeating live loads, e.g., bridges.
Therefore, this study further focuses on the stress concentration b1
characteristics of square bird-beak SHS T-joints by the use of finite b1
element method. Two load cases, chord axial force and brace axial
force, are considered. Refined FE models of both conventional and
bird-beak T-joints, whose meshes are highly consistent with the
strain extrapolation rules, are established. The SNCF differences
among considered hot spots are analyzed for the identification of
critical locations. The relationship between SNCF and SCF of square
bird-beak T-joints are also revealed. The influences of three major L1 Brace
non-dimensional parameters on SCFs of square bird-beak T-joints
are obtained on the basis of parametric finite element analysis. Chord
Comparisons of joint types, which aim to verify lower stress t0
concentrations for square bird-beak T-joints, are finally made.
b0

2. SHS T-joints and FE model b0

L0
2.1. Dimensions

Both conventional SHS T-joints (J) and square bird-beak SHS t1


T-joints (SBBJ) are investigated. Three types of non-dimensional
parameters, that is, brace/chord width ratio β ¼b1/b0, chord wall b1
slenderness ratio 2γ ¼ b0/t0, and brace/chord wall thickness ratio b1
τ ¼ t1/t0, are mainly considered. Here, b0 and t0 represent the
sectional width and wall thickness of the chord; b1 and t1 corre-
spond to the sectional width and wall thickness of the brace, as
shown in Fig. 2. The chord length L0 is six times its sectional width
b0, and the brace length L1 is three times its sectional width b1.
L1
Brace
2.2. Hot spots
Chord
For the conventional joints, six hot-spot locations (i.e., lines B, C
and D on the chord and E, F and A on the brace) have been b0
identified in existing literatures [18] and CIDECT design guide [22],
as shown in Fig. 3(a). Among these, four corner lines A, B, D, and E t0
b0
align perfectly with the inner surface plane of the brace wall.
For the bird-beak joints, the junction configurations which
contain the so-called crown areas and saddle areas, as used for L0
CHS tubular joints, are much more complicated. By referring to the
Fig. 2. Dimensions of SHS T-joints: (a) conventional and (b) square bird-beak.
findings from Cheng et al. [17] and adopting the similar criteria used
for conventional joints, four crown hot-spot locations (i.e., Cr-B and
Cr-C on the chord and Cr-F and Cr-A on the brace) and six saddle from the weld toe (i.e., the extrapolation region) by the use of an
hot-spot locations (i.e., Sa-B, Sa-C and Sa-D on the chord and Sa-E, extrapolation approach. In the present research, the boundaries of
Sa-F and Sa-A on the brace) were selected, as shown in Fig. 3(b). extrapolation regions defined in CIDECT Design Guide No. 8 [22]
For a hot-spot location, the stress/strain at weld toe was are adopted, as shown in Fig. 3 where Lmin takes the greater value
calculated from the stresses/strains within a fixed region away of 0.4t and 4 mm.
B. Cheng et al. / Thin-Walled Structures 94 (2015) 435–445 437

Fig. 3. Arrangement of strip gauges at hot spots (top view): (a) conventional and (b) square bird-beak.

2.3. Finite element model regions of square bird-beak T-joints. In the present research, finite
element analysis is first conducted for the nine orthogonally designed
A general-purpose finite element program ANSYS 12 [23] was square bird-beak SHS T-joints which have been tested by Cheng et al.
used to obtain the strain/stress concentration factors (SNCFs/SCFs) [17]. Fig. 5 plots the typical variations of calculated SNCFs along hot-
of two types of SHS T-joints. The analyses were based on the small spot lines Sa-C and Sa-A of specimen SBBJ-2 under brace axial force,
deformation and the material elasticity, i.e., neither geometrical nor and nonlinear variations are apparent. Further comparison between
material non-linearity was considered. Solid elements (SOLID95) linear and quadratic extrapolation methods were made by introdu-
with 20 nodes and three degrees of freedom for each node were cing a ratio of SNCFl/SNCFq, where SNCFl and SNCFq correspond to the
used. Geometries of fillet welds were accurately considered in FE strain concentration factors derived from linear and quadratic extra-
models. In order to take into account the complexities of stress polation, respectively. Fig. 6 illustrates the numerical SNCFl/SNCFq
distributions near the junction, the meshes in these areas were ratios at all the investigated spots of nine T-joints under brace axial
refined. The element sizes within the selected extrapolation regions force. The averaged and minimum values of the SNCFl/SNCFq ratios
were exactly specified to be 2 mm, which makes the nodal stresses/ are 0.91 and 0.78, respectively, which indicates that linear extrapola-
strains could be directly used for the extrapolation of hot spot tion may underestimate the SNCFs of square bird-beak RHS T-joints
stresses/strains at the weld toes. by up to 22%. Therefore, quadratic extrapolation, as recommended by
Fig. 4 demonstrates the typical finite element meshes of half of a Cheng et al. [17], was also used in this paper.
square bird-beak T-joint due to symmetric condition. The selected
mesh-controlling parameters had also been validated by comparing to 3.2. Comparison with experimental data
the half-size refined meshes, i.e., the elements within the extrapolation
regions had uniform lengths of 1 mm. Errors in SNCF were found to To further verify the validity of FE models, experimental SNCFs
vary between 0.1% and 4% for all hot spots considered, indicating that of nine orthogonally designed square bird-beak SHS T-joint speci-
the established FE meshes should be accurate enough for the study. mens [17] were also employed for the comparison. The measured
sectional widths, wall thicknesses, member lengths as well as fillet
weld sizes of specimens were adopted into the FE models. It can be
3. Numerical analysis results seen from Fig. 7 that in most occasions the numerical strain
concentration factors (i.e., SNCFFEM) are slightly greater than those
3.1. Extrapolation method obtained from the test (i.e., SNCFexp). The average value of SNCFFEM/
SNCFexp goes to 1.15 and 1.07, respectively for chord hot-spots and
It has been experimentally revealed by Cheng et al. [17] that brace hot-spots. The coincidence between the test and the simula-
nonlinearities of strain distributions do exist within extrapolation tion could be regarded as fairly good.
438 B. Cheng et al. / Thin-Walled Structures 94 (2015) 435–445

Fig. 4. Finite element model of square bird-beak T-joint (half-structure).

7 y=1.2x y=x
Sa-C 12
Sa-A
6 Linear extrapolation
Quadratic extrapolation 10 y=0.8x

5
8
SNCF

SNCFFEM

4
6

3
4

2
0 2 4 6 8 10 12 14 2
Distance from weld toe (mm) Sa-A
Sa-C
Fig. 5. Nonlinearity of strain distributions along hot lines of specimen SBBJ-2 under 0
brace axial force. 0 2 4 6 8 10 12
SNCFexp

Fig. 7. Comparison of experimental and FEM SNCFs of square bird-beak T-joints


1.0 under brace axial force.

0.9

3.3. SNCF magnitudes at hot spots


0.8
SNCFq
SNCFl

For the nine orthogonally designed square bird-beak SHS T-joints


0.7 [17], numerical SNCF magnitudes at all considered hot spots of nine
specimens were also compared, and the numbers of joints as
grouped by location of maximum SNCF are listed in Table 1, where
0.6 both test and FEM results have been given. It can be seen that, in
Chord Brace case of chord axial force, all braces are free from strain concentra-
0.5 tions and all maximum SNCFs occur at chord spots Sa-D definitely,
which is completely consistent with the measured data.
C

-F

F
-A
C
-B

-D

-E
Sa-
Sa-

Sa-
Cr-

Cr-
Cr

Sa
Cr

Sa

Sa

In case of brace axial force, the maximum chord SNCFs occurred


Hot Spot at spots Sa-C or Sa-D, and the maximum brace SNCFs occurred at
Fig. 6. Linear versus quadratic extrapolation for all hot spots of square bird-beak spots Sa-A or Cr-F. Fig. 8 shows an intuitive comparison of SNCFs at
T-joints under brace axial force. various hot spots. The following observations are made:
B. Cheng et al. / Thin-Walled Structures 94 (2015) 435–445 439

(1) Four of the nine square bird-beak joints have maximum chord these stresses are usually perpendicular to the weld toes for
SNCFs at saddle spots Sa-C, and the other five joints have simple joints. It has been found that principal stresses could be
maximum chord SNCFs at saddle spots Sa-D. The SNCF higher than stresses perpendicular to the weld toe, whereas the
differences between these two spots are as small as 1–9%. extrapolated SCFprincipal is often lower than SCFperpendicular since
The average ratios of the SNCFs at spots Cr-C, Cr-B, and Sa-D to the principal stresses tend to increase less sharply towards the
the maximum values at spots Sa-C or Sa-D are 0.87, 0.86, and weld toe. As only stress components perpendicular to the weld toe
0.91, respectively, indicating more severe strain concentration are enlarged by stress concentrations caused by the global weld
for all chord spots. shape and the wall of the adjacent member and as only these
(2) There are four square bird-beak joints whose maximum brace stress components can be measured by simple strain gauges
SNCFs occurred at saddle spots Sa-A, and the maximum brace instead of strain gauge rosettes, the hot spot stress definition
SNCFs of the other five joints occurred at saddle spots Cr-F. The based on stress perpendicular to weld toe was commonly adopted
strain concentrations at saddle spots Sa-A and Sa-F are very by many researchers [18–20].
close to each other, with the SNCF differences between them Differences between these two stresses are compared by
being 3.5–5.1%. Similar situations happen to crown spots Cr-A analyzing the FEM results of some square bird-beak T-joints with
and Cr-F. SNCFs at saddle spots Sa-E were found to be only typical non-dimensional parameters, and it shows that the differ-
48–78% of the maximum brace SNCFs, which indicates less ences depended on the hot-spot locations as well as the load cases.
strain concentrations in saddle spots. In case of chord axial force, SCFprincipal at spots Cr-C, Cr-B, and Sa-D,
(3) In comparison, the maximum chord SNCFs of seven joints are whose directions were almost parallel to the load applied, were up
greater than the corresponding maximum ones in the brace, to 3% greater than SCFperpendicular, while the average percentage for
with the differences between them being 12–40%. While for spots Sa-C went to 24% due to the existence of oblique angles
the other two joints where highest stress concentrations between the hot-spot lines and the load direction. In case of brace
occurred in the braces, the maximum chord SNCFs are as large axial force, SCFprincipal normal to the weld toes (i.e., along lines Cr-
as 90% of the maximum brace SNCFs. Therefore, the chords of C, Cr-B, Cr-F, Sa-D, Sa-C, Sa-B, Sa-A) was also up to 3% larger than
square bird-beak T-joints are more prone to higher stress the corresponding SCFperpendicular. While for the hot-spot lines
concentrations than the braces. The findings match the experi- oblique to the weld toes (i.e., Cr-A, Sa-E, Sa-F), the maximum
mental conclusions from paper [17]. SCFprincipal/SCFperpendicular ratios went sharply to 1.17. It can there-
fore be considered that, for square bird-beak SHS T-joints, SCFs
based on the principal stress are similar to those based on stresses
3.4. Principal stress versus the stress perpendicular to weld toe normal to the weld toe.

There are two types of stresses, i.e., the principal stress and the
stress normal to weld toe, to be used for the definition of hot spot 3.5. Relationship between SNCF and SCF
stress. The stresses normal to weld toe are adopted by the AWS
[24], while the IIW [21] uses the principal stresses by stating that Relationships between SNCF and SCF of traditional tubular
joints have been investigated in existing literatures [25,26]. It
was found that SNCF could be converted to SCF by multiplying a
coefficient of 1.2 for CHS joints or 1.1 for RHS joints. This relation-
Table 1 ship, which depends on the joint configuration, the local position
Number of joints as grouped by location of maximum SNCF.
concerned as well as the loading case, can be deduced on the basis
Load case Max. Chord Brace of Hooke's Law in linear elasticity and then be expressed as follows
SNCF [27]:
ε !
Cr- Cr- Sa- Sa- Sa- Cr- Cr- Sa- Sa- Sa-
C B D C B F A E F A 1 þ νε ==?
SCF ¼ USNCF ð1Þ
Chord Exp. 0 0 9 0 0 N/A
1  ν2
force FEM 0 0 9 0 0 N/A
in which, ν is the Poisson's ratio, ε// is the strain component
Brace Exp. 2 1 0 6 0 1 1 0 2 5
force FEM 0 0 5 4 0 5 0 0 0 4
parallel to the weld toe, and ε┴ is the strain component perpendi-
cular to the weld toe.

SBBJ-1 SBBJ-2 SBBJ-3 SBBJ-4 SBBJ-5 SBBJ-6 SBBJ-7 SBBJ-8 SBBJ-9

Cr-F
Cr-A
Brace Sa-E
Sa-F
Sa-A

Cr-C
Cr-B
Chord Sa-D
Sa-C
Sa-B

0 5.1 0 6.4 0 8.0 0 6.6 0 5.7 0 6.0 0 4.4 0 5.4 0 4.1


SNCF

Fig. 8. SNCF magnitudes at different hot spots of square bird-beak T-joints under brace axial force.
440 B. Cheng et al. / Thin-Walled Structures 94 (2015) 435–445

It is normally unrealistic to obtain ε// and ε┴ from the test. likely to produce higher rates of increases or decreases in SCFs,
Therefore, the SNCF-to-SCF relationships of square bird-beak respectively, with increasing β.
T-joints subject to brace or chord axial force were investigated in (2) Case of variable 2γ. Both chord SCFs and brace SCFs are
this research by the use of finite element method. Fig. 9 shows the improved as 2γ is enhanced from 12.5 to 25, with the rates of
numerical SNCFs and SCFs at all selected hot spots where the increase within the middle variable zones where 2γ take values
principal strains/stresses were almost perpendicular to the weld between 15 and 20 being of the highest. Comparatively, the
toes. In these situations, the strains/stresses parallel to the weld larger β or τ is, the higher the rate of increase in SCF becomes.
toes were very small. A high correlation between SNCFs and SCFs (3) Case of variable τ. Increases in chord SCF with enlarging τ are
can be found from the figure. By fitting these data using the least evident, with the rates of increase being slightly slowed down
square method, the relationship between SNCF and SCF of square as τ gets greater. The curves in Fig. 12(c) can be fitted by the
bird-beak T-joints can be approximately written as the following: use of typical quadratic functions (half-curve). Variations in
SCF ¼ 1:15 USNCF: ð2Þ brace SCF with varying τ are relatively insignificant, with both
descending and declining trend with increasing τ being
observed for various cases of β and 2γ. If necessary, parabolic
curves can also be used for the fitting of these brace data.
4. Parametric studies (4) By further integrating the chord curves and the brace curves
into the same plot, it can be found that, in case of identical non-
4.1. Weld sizes dimensional parameters, the chords are prone to have severe
stress concentrations than the braces do in most situations.
It has been recognized that weld dimensions have significant Higher SCFs occur to the braces only when the brace/chord wall
influences on stress concentrations of regular SHS joints [18], thickness ratios (τ) are extremely small, e.g. less than 0.4. This
which means that the weld sizes should also be consistently finding of square bird-beak SHS T-joints is much different from
determined in accordance with a rational criterion during the the conventional SHS T-joints. According to the existing litera-
parametric analysis of bird-beak joints. In the parametric numer- tures [18,22], the maximum SCF of a conventional SHS T-joint
ical analysis where more cases of non-dimensional parameters
were considered, the prequalified weld details for PJP tubular 16
joints provided in AWS Code [24] were adopted, and the weld Chord spots, under chord axial force
sizes within different junction areas (i.e., the crown areas where Chord spots, under brace axial force
14 Brace spots, under brace axial force
the welded plates were perpendicular to each other and the saddle
Fitting: Y=1.1579X, R =0.9961
2
areas where the oblique angles between two plates were near or
12
equal to 1351) were determined as follows:

(1) For the crown areas in Fig. 10(a) 10


Weld length projected on the brace w1;Cr ¼ t 1
SCF

Weld length projected on the chord w0;Cr ¼ t 1 =2 8


(2) For the saddle areas in Fig. 10(b) pffiffiffi
Weld length projected on the brace w1;Sa ¼ t 1 U p2ffiffiffi 6
Weld length projected on the chord w0;Sa ¼ ðt 1 U 2Þ=2:

4
4.2. Case of chord axial force
2
The influences of three non-dimensional parameters on the
maximum SCFs at spot Sa-D of square bird-beak T-joints subject to 0
chord axial force are shown in Fig. 11. For the cases of variable β and 0 2 4 6 8 10 12 14 16
2γ, any increases in β or 2γ lead to SCF increase. The increasing trend SNCF
is more obvious when β varies. As for the varying τ, approximate Fig. 9. Comparison of SNCFs and SCFs for square bird-beak T-joints under brace
parabolic curves are observed from the figure, with the highest SCFs axial force.
being close to τ around 0.5 or 0.6. It should be noted that most of the
SCFs under chord axial force are below 2.0, which could be regarded
t1 t1
as fairly small when compared to the load case of brace axial force.

4.3. Case of brace axial force 45°


Brace wall
Brace wall
Figs. 12 and 13 demonstrate the typical variations of chord SCFs
at spot Sa-C and brace SCFs at spot Sa-A of square bird-beak 90°
T-joints under brace axial force, as functions of three non- w1,Cr
dimensional parameters considered. The following observations w1,Sa
can be made: w0,Cr
w0,Sa
(1) Case of variable β. Both chord SCFs at spot Sa-C and brace SCFs Chord wall Chord wall
at spot Sa-A first increase and then decrease as β reaches
about 1.0. The slopes of curves for β 4 βpk are much greater
than those symmetric ones for β o βpk, indicating that para-
bolic equations may not be accurate enough for the variation Fig. 10. Weld details used for parametric analysis: (a) crown area and
description. Comparatively speaking, a greater 2γ or τ is more (b) saddle area.
B. Cheng et al. / Thin-Walled Structures 94 (2015) 435–445 441

βsbb βsbb
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
2.8 24
22 2γ =15, τ =0.4
2.6 20 2γ =20, τ =0.6
2.4 Sa-D 18 2γ =25, τ =0.8 Sa-C

16
2.2 14

SCF
SCF

2.0 12
10
1.8 8
1.6 2γ =15,τ =0.4 6
2γ =20,τ =0.6 4
1.4 2
2γ =25,τ =0.8
1.2 0
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4
β β

2.2 20
18 β =0.5, τ =0.4
β =0.7, τ =0.6
16
2.0 β =0.9, τ =0.8
14
12
1.8
SCF
SCF

10
8
1.6
6
β =0.5,τ =0.4 4
1.4
β =0.7,τ =0.6 2
Sa-D β =0.9,τ =0.8 Sa-C
0
1.2 10 12 14 16 18 20 22 24 26 28
10 12 14 16 18 20 22 24 26 28

20
2.2
18 β =0.5, 2γ =15
16 β =0.7, 2γ =20
2.0 β =0.9, 2γ =25
14
12
1.8
SCF

10
SCF

8
1.6
6
β =0.5, 2γ =15 4
1.4
β =0.7, 2γ =20 2 Sa-C
Sa-D β =0.9, 2γ =25 0
1.2 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 τ
τ
Fig. 12. Variations of SCFs at chord spot Sa-C with non-dimensional parameter,
Fig. 11. Variations of SCFs at chord spot Sa-D with non-dimensional parameter, under brace axial force: (a) brace/chord width ratio β; (b) chord wall slenderness
under chord axial force: (a) brace/chord width ratio β; (b) chord wall slenderness ratio 2γ; and (c) brace/chord wall thickness ratio τ.
ratio 2γ; and (c) brace/chord wall thickness ratio τ.

could have a larger width than the chord, which is unrealistic for
subject to brace axial force may occur either on the chord or on conventional joints. In case of brace axial force, peak points are
the brace, which depends on β, 2γ, and τ values. always observed from the SCF-β curves of the two types of joints,
that is, SCF decreases from the peak value as β is decreased from
βpk to 0 or is increased from βpk to the maximum value βmax, as
4.4. Definition of brace/chord width ratio shown in Figs. 12, 13 and 15. The primary difference lies in that βpk
of square bird-beak joints are much larger than the conventional
For a square bird-beak joint, the brace/chord width ratio β ones. Therefore, a new definition of brace/chord width ratio based
could reach a maximum value of 1.4 (or more accurately square on the contact length between brace and chord walls could be
root of 2.0) with the brace wall being exactly supported by the introduced for square bird-beak joints so that the brace/chord
chord corners. In other words, the brace of a square bird-beak joint width ratio βmax of two joint types are unified to 1.0.
442 B. Cheng et al. / Thin-Walled Structures 94 (2015) 435–445

βsbb
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
24
22 2γ =15,τ =0.4 Sa-A
20 2γ =20,τ =0.6
18 2γ =25,τ =0.8
16
14
SCF

12
10
8
6
4
2
0
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 Fig. 14. Decreases/increases in global maximum SCFs (Gi) for nine comparing
groups.
β

20
18 β =0.5,τ =0.4 Sa-A
β =0.7,τ =0.6
16
β =0.9,τ =0.8
14
12
SCF

10
8
6
4
2
0
10 12 14 16 18 20 22 24 26 28

20
18 β =0.5, 2γ =15 Sa-A
16 β =0.7, 2γ =20
β =0.9, 2γ =25
14
12
SCF

10
8
6
4
2
0
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
τ
Fig. 13. Variations of SCFs at brace spot Sa-A with non-dimensional parameter,
under brace axial force: (a) brace/chord width ratio β; (b) chord wall slenderness Fig. 15. Comparisons of SCFs between square bird-beak SHS T-joints and conven-
ratio 2γ; and (c) brace/chord wall thickness ratio τ tional SHS T-joints under brace axial forces: (a) chord spot C or Sa-C and (b) brace
spot A or Sa-A.
In the new definition, the normalized pffiffiffi brace width takes the
contact length on the chord walls (i.e., 2 U b21 ) instead of original b1. the peak βsbb under brace axial forces are observed to be within
The normalized brace/chord width ratio of square bird-beak joints 0.6–0.8.
(βsbb) can thus be calculated as
pffiffiffi  pffiffiffi
b1 2 b1
βsbb ¼ 2U =b0 ¼ U ¼ 0:707β ð3Þ 5. Comparison of joint types
2 2 b0
The normalized definition of brace/chord width ratio is con- 5.1. Square bird-beak SHS T-joints versus conventional SHS T-joints
vinced to have a better reflection to the natural features of SCF
variations and thus could be used in the simplified formulae that To make a comparison covering the widest ranges of applica-
will be proposed in the authors' upcoming research. The SCF tions, nine cases of orthogonal non-dimensional parameters that
variations with βsbb have also been provided in Figs. 11–13, where have been employed by Cheng et al. [17] were considered. Therefore,
B. Cheng et al. / Thin-Walled Structures 94 (2015) 435–445 443

totally nine comparing groups, each consisting of a square bird-beak SHS T-joints based on CIDECT equation [22], as plotted against the
SHS T-joint and a conventional SHS T-joint with identical SHS varying β.
dimensions, were established, as shown in Table 2. For the group i Since the stress concentrations of a simply supported T-joint
which is composed of conventional joint J-i and square bird-beak subject to brace axial force are produced not only by the axial force
joint SBBJ-i, the percentage of difference (Gi) in SCF is calculated as on the brace but also by the in-plane bending on the chord, the
  SCFs of conventional SHS T-joints are calculated as the following:
SCFSBBJi  SCFJi
Gi ¼  100% ð4Þ
SCFJi A1 U ðL0  b1 Þ
SCF ¼ SCFBF þ SCFCIPB U U ðb0 =2Þ ð5Þ
In case of chord axial force, since all SCFs of both conventional 4 UI 0
and bird-beak joints in nine groups are lower than 2.0, i.e., the where A1 is the sectional area of the brace, I0 is the moment of
number that had been specified as the minimum value required inertia of the chord section.
for conventional RHS joints in CIDECT Design Guide No. 8 [22], SCFBF corresponds to the loading case of brace axial force, and is
comparison is unnecessary and the same requirement (i.e. mini- calculated as [22]
mum SCF of 2.0) is recommended by the authors for square bird-
beak joints. – For chord spot C
In case of brace axial force, it can be seen from Table 2 that
SCFBF ¼ ð0:077  0:129 U β þ0:061 U β  0:0003 U2γ Þ
2
 
(1) All the maximum brace SCFs of square bird-beak joints are   1:565 þ 1:874 U β  1:028 U β2
U 2γ U τ0:75 ð6Þ
smaller than those of conventional joints by 24–74%.
(2) For most comparing groups, the maximum chord SCFs of
square bird-beak joints are found to be 6–54% smaller than – For brace spot A
those of conventional ones. Increase in the maximum chord
SCFBF ¼ ð0:013 þ 0:693 U β 0:278 U β Þ
2
SCFs (i.e., positive Gi values) also happens to the groups where  
  0:790 þ 1:898 U β  2:109 U β2
β takes the value of 0.75 (e.g., group 1 and group 2). U 2γ ð7Þ
(3) Different from the fact that the maximum chord SCF of a bird-
beak T-joint is always greater than the maximum brace SCF,
SCFCIPB corresponds to the loading case of chord in-plane
the global maximum SCF of a conventional T-joint may occur
bending, and is calculated as [22]
either on the chord or on the brace. Therefore, the global
– For chord spot C
maximum SCF of the whole joint, i.e., the bigger one between
 0:248 U β 0:19
the maximum chord SCF and the maximum brace SCF, was SCFCIPB ¼ 0:725 U 2γ Uτ ð8Þ
further compared. Results show that the global maximum SCFs
of eight out of nine bird-beak T-joints are smaller than those of
conventional T-joints compared. Only one square bird-beak
joint (i.e., SBBJ-2) had the global maximum SCF of about 11% Table 2
greater than that of the conventional joint (J-2), the reasons of Comparison of local maximum SCFs between conventional SHS T-joints and bird-
which will be explained next. beak T-joints under brace axial force.

Group Non-dimensional parameter SCF


To illustrate the impact of non-dimensional parameters, the
global maximum SCF reductions (Gi) of these groups are plotted β 2γ τ Chord max. Brace max. Global max.
in Fig. 14, where β, 2γ and τ are employed as the three-dimensional
SCFJ-1 0.75 16.67 0.500 5.88 8.02 8.02
coordinates and Gi values are comparatively represented by the size SCFSBBJ-1 7.93 6.01 7.93
of the ball symbols. It is evident that all Gi within the shaded G1 34.8%  25.1%  1.1%
trapezoid prism has negative values. By linking three data G9, G5,
SCFJ-2 0.75 20.00 0.600 8.34 9.80 9.80
and G1 having identical τ, or linking three data G9, G6, and G3 having SCFSBBJ-2 10.86 7.47 10.86
identical 2γ, as shown in Fig. 14, decreases in ball sizes with G2 30.2%  23.8% 10.8%
increasing β are also evident. It may be deduced that the smaller SCFJ-3 0.75 25.00 0.750 15.87 12.48 15.87
the chord/brace width ratio is, the more significant the stress SCFSBBJ-3 14.79 9.27 14.79
concentration relief derived from the square bird-beak configura- G3  6.8%  25.7%  6.8%
tion becomes. This finding is highly consistent with the experi- SCFJ-4 0.60 16.67 0.833 10.84 8.02 10.84
mental observations from Cheng et al. [17], and could be explained SCFSBBJ-4 10.16 4.53 10.16
on the geometry induced stiffness discontinuity which had been G4  6.3%  43.5%  6.3%
considered as the natural cause of stress concentration. For a square SCFJ-5 0.60 20.00 0.500 9.95 10.41 10.41
bird-beak joint with smaller β, the brace walls are closer to the SCFSBBJ-5 8.01 5.54 8.01
chord's top corner where the stiffness in the direction of brace axis G5  19.5%  46.8%  23.0%
are fairly high, indicating that the stiffness differences between the SCFJ-6 0.60 25.00 0.625 17.38 13.69 17.38
brace walls and the chord walls supporting them are less signifi- SCFSBBJ-6 11.32 6.77 11.32
G6  34.8%  50.5%  34.8%
cant, and therefore lower stress concentration factors appear. The
situations are completely reversed for conventional joints, and as a SCFJ-7 0.45 16.67 0.667 10.54 8.18 10.54
result, smaller SCFs occur to the conventional joint with a greater β SCFSBBJ-7 6.77 3.15 6.77
G7  35.8%  61.5%  35.8%
whose brace walls are closer to the chord's side corners. For a
conventional joint where the brace's lateral walls are completely SCFJ-8 0.45 20.00 0.800 16.38 10.11 16.38
SCFSBBJ-8 8.70 3.73 8.70
supported on the chord's lateral wall in the same plane rather than
G8  46.9%  63.1%  46.9%
on the chord's top wall which is perpendicular to the brace wall (i.e.,
β ¼1.0), the SCFs are expected to reach the minimum values. SCFJ-9 0.45 25.00 0.500 15.51 14.66 15.51
SCFSBBJ-9 7.14 3.75 7.14
Fig. 15 simultaneously shows the variations of SCFs of square G9  54.0%  74.4%  54.0%
bird-beak SHS T-joints based on FE results and those of conventional
444 B. Cheng et al. / Thin-Walled Structures 94 (2015) 435–445

– For brace spot A (3) The stresses within crown and saddle areas are of the highest.
SCFCIPB ¼ 0 ð9Þ In case of chord axial force, stress concentrations on the braces
are very insignificant, and the maximum SCFs occur at the
saddle spot Sa-D on the chords. In case of brace axial force, the
maximum chord SCFs appear at the saddle spots Sa-C or Sa-D,
It can be seen that SCF variations of conventional T-joints are while the maximum brace SCFs belong to the saddle spots Sa-
much different from those of square bird-beak T-joints. Obviously, A or Cr-F. In case of brace axial force, the difference in SCF is
as β is increased from 0.5 to 1.0, decreasing SCFs of conventional not evident between hot-spots Sa-C and Sa-D or Sa-A and Sa-F
joints and increasing SCFs of square bird-beak joints were or Cr-A and Cr-F.
observed for all cases. There always exists an intersection between (4) In case of chord axial force, the maximum SCFs tend to
the conventional curve and the square bird-beak curve, after increase as β or 2γ increases, but decrease with an increased
which square bird-beak SCFs are higher than conventional SCFs. τ. All the SCFs under chord axial forces are much smaller than
The critical brace/chord width ratios βcr corresponding to the those of joints subject to brace axial forces.
intersections of chord SCF curves, which take values between (5) In case of brace axial force, both SCFs at chord spots Sa-C and
0.7 and 0.8, are considerably depressed as 2γ or τ is decreased. As SCFs at brace spots Sa-A generally increase as any one of the
for the comparing group 2, two T-joints have the same β greater three non-dimensional parameters (β, 2γ or τ) increases. The
than the βcr value of chord SCF curves with 2γ ¼20 and τ ¼0.6, as exceptions are that an increase in very large β (e.g., β 4 0.9)
shown in Fig. 15(a). Therefore, it is easy to understand why the may significantly reduce the chord SCFs, and that the increase
global maximum SCF (i.e., the maximum chord SCF) of square of τ may also slightly reduce the brace SCFs.
bird-beak joint SBBJ-2 is larger than that of conventional joint J-2. (6) In case of brace axial force, the maximum SCFs on the chords are
usually greater than those on the braces. A higher SCF occurs in
5.2. Square bird-beak SHS T-joints versus CHS T-joints the brace only when τ is extremely small (e.g., τ o0.4).
(7) Stress concentrations of square bird-beak SHS T-joints under
To some extent, the square bird-beak configurations are similar chord axial force are always slightly more severe than conven-
to the circular hollow section (CHS) joints, since both conjunctions tional SHS T-joints, with the SCF differences being as small as
are composed of saddle areas and crown areas. Therefore CHS 0.3–0.5. While for the loading case of brace axial force, the square
T-joints, whose SCFs have been proved to be obviously smaller bird-beak SHS T-joints have smaller SCFs than the conventional
than SHS T-joints when non-dimensional parameters are identical, SHS T-joints with identical non-dimensional parameters, and
were also employed for comparison. A series of equations pro- those SCF reduction percentages are amplified as β is decreased.
posed in CIDECT Design Guide No. 8 [22] were directly used to
determine the SCFs of CHS T-joints. By introducing θ ¼901 and
α ¼12 into the CIDECT formulations for CHS T-joints with pinned
chord ends, the SCFs of CHS T-joints under brace axial forces are
calculated as follows:

– For chord saddle


h  2 i
SCF ¼ γ U τ1:1 1:11  3 U β  0:52 ð10Þ

– For brace saddle


h i
1:1 
SCF ¼ 1:3 þ γ U τ0:52 U α0:1 0:187 1:25 U β U β 0:96 ð11Þ

Fig. 16 presents the variations of two types of SCFs with varying


β. It can be seen that intersections between two types of curves, as
observed in Fig. 15, are also evident. However, the critical brace/
chord width ratios βcr corresponding to the intersections in Fig. 16
are obviously smaller than those measured from Fig. 15. The values
of βcr increase as 2γ or τ increases. In general, square bird-beak
T-joints is more favorable than CHS T-joints only in cases of small
β, large 2γ and τ.

6. Conclusions

This paper presented a numerical investigation of the stress


concentration factors of square bird-beak SHS T-joints subject to
chord and brace axial forces. Based on the finite element analysis,
the following conclusions can be made:

(1) Stresses normal to the weld toes can be used for the definition
of hot spot stress.
(2) SCFs can be converted from SNCFs by multiplying a coefficient Fig. 16. Comparisons of SCFs between square bird-beak SHS T-joints and CHS
of 1.15. T-joints under brace axial forces: (a) chord saddle and (b) brace saddle.
B. Cheng et al. / Thin-Walled Structures 94 (2015) 435–445 445

(8) In case of brace axial force, square bird-beak SHS T-joints have [10] Lei LY. The analysis of square bird-beak SHS joint subject to in-plane bending
smaller SCFs than CHS T-joints only for small β and large 2γ moment [M.S. thesis]. Xi'an, China: Xi'an University of Architecture and
and large τ.
Technology; 2009.
[11] Zhu ZQ, Liu SL. Nonlinear finite element analysis of diamond bird-beak TX-
joints. Adv Mater Res 2012;446:151–5.
[12] Hall SK, Owen JS. The behaviour of hollow section connections under seismic
loading. Struct Eng 2004;82:22–4.
Acknowledgments [13] Ishida K. Experimental research on fatigue behavior of diamond bird-beak
joint. In: Proceedings of the symposium on structural engineering. Tokyo:
Architectural Institute of Japan; 1992 [in Japanese].
The present research was undertaken with support from the [14] Keizer R. Stress concentration factors in diamond bird beak T-joints. [M.S.
special project of Science and Technology Research and Develop- thesis]. Delft, Netherlands: Delft University of Technology; 2003.
ment Plan funded by the Ministry of Railway of P.R.C. (No. [15] Keizer R, Romeijn A, Wardenier J. The fatigue behaviour of diamond bird beak
T-joints. In: Proceedings of the international symposium on tubular structures.
J2011G002). London; 2003. p. 303–10.
[16] Tong LW, Fu YG, Liu YQ, Zhao XL. Stress concentration factors of diamond bird-
References beak SHS T-joints under brace loading. Thin-Walled Struct 2014;74:201–12.
[17] Cheng B, Qian Q, Zhao XL. Tests to determine stress concentration factors for
square bird-beak SHS joints under chord and brace axial forces. ASCE J Struct
[1] Ono T, Iwata M, Ishida K. An experimental study on joints of new truss system Eng 2014;140:04014088.
using rectangular hollow sections. In: Proceedings of the international [18] van Wingerde AM. The fatigue behaviour of T- and X-joint made of square
symposium on tubular structures. London; 1991. p. 344–53. hollow section. HERON 1992;37:1–180.
[2] Ono T, Iwata M, Ishida K. Local failure of joints of new truss system using [19] van Wingerde AM, Packer JA, Wardenier J. SCF formulae for fatigue design of
rectangular hollow sections subjected to in-plane bending moment. In: K-connections between square hollow sections. J Constr Steel Res 1997;43:
Proceedings of the international symposium on tubular structures. London; 87–118.
1993. p. 503–10. [20] Packer JA, Wardenier J. Stress concentration factors for non-901 X-connections
[3] Ono T, Iwata M, Ishida K. Local failure of joints of new truss system using
made of square hollow sections. Can J Civ Eng 1998;25:370–5.
rectangular hollow sections subjected to out-of-plane bending moment. In:
[21] International Institute of Welding (IIW). Recommended fatigue design proce-
Proceedings of the international symposium on tubular structures. London;
dure for welded hollow section joints-part 1: recommendations, part 2:
1994. p. 441–8.
commentary, XV-1035-99. Cedex, France; 2008.
[4] Ishida K, Ono T, Iwata M. Ultimate strength formula for joints of new truss
[22] Zhao XL, Herion S, Packer JA, Puthli RS, Sedlacek G, Weynand K, et al. Design
system using rectangular hollow sections. In: Proceedings of the international
symposium on tubular structures. London; 1993. p. 511–8. guide for circular and rectangular hollow section joints under fatigue loading.
[5] Davies G, Owen JS, Kelly R. Bird beak T-joints in square hollow sections: a TÜV-Verlag, Köln, Germany: Comité International pour le Développement et
finite element investigation. In: Proceedings of the international offshore and l’Etude de la Construction Tubulaire; 2001.
polar engineering conference. Mountain View, CA: International Society of [23] ANSYS. User's manual (version 10.0). Houston: Swanson Analysis System Inc.;
Offshore and Polar Engineers; 1996. p. 1–6. 2005.
[6] Owen JS, Davies G, Kelly RB. A comparison of the behaviour of RHS bird beak [24] American Welding Society (AWS). Structural welding code-steel, ANSI/AWS
T-joints with normal RHS and CHS systems. In: Proceedings of the interna- D1.1/D1.1M:2010. Miami: American Welding Society; 2010.
tional symposium on tubular structures. London; 1996. p. 173–80. [25] van Delft DRV, Noordhoek C, Da Re ML. The results of the European fatigue
[7] Owen JS, Kelly RB. The effect of purlin loads on the capacity of overlapped tests on welded tubular joints compared with SCF formulas and design lines.
bird-beak K joints. In: Proceedings of the international symposium and Euro In: Proceedings of the steel in marine structures. Delft, The Netherlands; 1987.
conference. Boca Raton, FL: CRC; 2001. p. 229–38. p. 565–77.
[8] Owen JS, Davies G, Kelly RB. The influence of member orientation on the [26] Frater GS. Performance of welded rectangular hollow structural section
resistance of cross joints in square RHS construction. J Constr Steel Res trusses. [Ph.D. thesis]. Canada: University Toronto; 1991.
2001;57:253–78. [27] Sing PC, Chee KS, Nai WW. Experimental and numerical stress analyses of
[9] Chtistitas AD, Pachoumis DT, Kalfas CN, Galoussis EG. FEM analysis of tubular XT-joints. ASCE J Struct Eng 1999;125:1239–48.
conventional and square bird-beak SHS joint subjected to in-plane bending
moment – experimental study. J Constr Steel Res 2007;63:1361–72.

You might also like