Proceedings of the Eighteenth (2008) International Offshore and Polar Engineering Conference

Vancouver, BC, Canada, July 6-11, 2008

Copyright © 2008 by The International Society of Offshore and Polar Engineers (ISOPE)
ISBN 978-1-880653-70-8 (Set); ISBN 1-880653-68-0 (Set)

Joints between Plates or I Sections and a Circular Hollow Section Chord

J. Wardenier
Delft University of Technology, Delft, The Netherlands and
National University of Singapore, Singapore
G.J. van der Vegte
Delft University of Technology, Delft, The Netherlands
Y. Makino
Kumamoto University, Kumamoto, Japan

ABSTRACT a exponent for n

b1 external width of brace
Recently Sub-commission IIW-XV-E has drafted new design cip coefficient for in-plane bending moment capacity
recommendations for hollow section joints. The joint strength equations cop coefficient for out-of-plane bending moment capacity
for circular hollow section (CHS) joints in these new recommendations d0 external diameter of chord
are the result of extensive re-analyses. This paper deals with design fy0 yield strength of chord
recommendations for uniplanar T and X joints between plates or I section h1 external height of brace
braces and a CHS chord. These new design recommendations are directly n non-dimensional chord stress ratio; in analysis N0 / Npl,0
related to those for CHS joints. However, small modifications are used in t0 wall thickness of chord
the constants to make the equations in line with the available test data. t1 wall thickness of brace
β width ratio between brace and chord
The design recommendations for joints loaded by in-plane or out-of-plane γ half diameter to thickness ratio of the chord, γ = d0/2t0
bending are related to those for axially loading by simple factors. γm partial factor
η brace height to chord diameter ratio h1/d0
This paper further shows that considerable differences exist between the θ1 angle between brace member and the chord
results of the various test data and additional numerical investigations are
recommended to solve these contradictions. INTRODUCTION

KEY WORDS: Plate to CHS joints, tubular joints, static strength. Recently Sub-commission IIW-XV-E has updated the design
recommendations for hollow section joints, designated as IIW (2008).
NOMENCLATURE The joint capacity equations for CHS joints in these new
recommendations are the result of extensive re-analyses discussed in
CHS circular hollow sections van der Vegte et al. (2008), Wardenier et al. (2008), and Zhao et al.
FE finite element (2008).
RHS rectangular hollow sections
One of the main differences with the previous IIW (1989) design
C1 exponent in chord stress function recommendations is that in the new IIW (2008) recommendations the
Mpl,0 chord plastic moment resistance for class 1 and 2 sections influence function for the chord stress effect on the joint capacity is
Mel,0 chord elastic moment resistance for class 3 sections based on the maximum chord stress instead of a so-called prestress, i.e.
M0 bending moment applied to the chord based on the chord loading excluding the effect of the brace loading
N1 axial brace load components. This required a full re-analysis of data, including
N0 chord axial load numerical data. Further, for tension loaded chords, a strength reduction
Npl,0 chord plastic axial load resistance due to chord stress is included whereas this was not the case in the
Nu,k characteristic axial load resistance previous IIW (1989) recommendations.
Qf chord stress function
Qu function f(β,γ,η) in the design strength equations The updated IIW (2008) design recommendations also include strength
Wel elastic section modulus (ip; in-plane; op: out-of-plane) formulae for plate and I to CHS chord joints which are the subject of
this publication.

319
CLASSIFICATION OF JOINTS The procedure used for the determination of the design strength follows
the approach of Wardenier (1982) and is described by van der Vegte et
Figure 1 shows the classification of plate and I to CHS chord joints. al. (2008).
The classification is similar to that used by Kurobane (1981), Makino et
al. (1991), Wardenier (1982) and Wardenier et al. (1991). First, the characteristic strength Nu,k is determined by taking account of
the dimensional tolerances and scatter in data, followed by a correction
Table 1. Classification of joints equal to 1/0.85 being fy,mean/ fy,k. The value Nu,k is further divided by a
XP-1 XP-2 XP-3 partial factor γm = 1.1 to obtain the design strength N1* . The ratio
b1 h1 Cruciform brace, i.e.
combination of XP-1 between N1* and the mean value of the test results varies, and depends
t0 t0
d0 d0 and XP-2. on the type of joint and the CoV. However, a reasonable value is about
1.25 for the XP and TP joints. Since all tests were carried out before the
deformation limit of 0.03d0 (Lu et al., 1994) was adopted by IIW, a
XP-4 XP-5 TP-1, etc. refer to T minimum value 1.25 to 1.30 is recommended.
h1 b1 h1 b1 joints with a brace at
one side of the chord. INITIAL APPROACH
t0 t0
d0 d0
Based on the foregoing philosophy, the new equation for CHS X joints
(van der Vegte et al., 2008) in combination with the f(η) function of the
CIDECT Design Guide No. 1 (Wardenier et al., 1991) is initially
assumed as the basic equation for XP-joints:
BACKGROUND
⎛ 1+ β ⎞ f t2
Kurobane (1981) was the first researcher who collected CHS and N1* = 2.6 ⎜⎜ ⎟⎟ (1 + 0.25η) γ0.15 Qf y0 0 (1)
XP/TP joint strength data in a systematic way. Based on his data and ⎝ 1 − 0.7β ⎠ sin θ1
analysis, followed by the re-analysis of Wardenier (1982), the joint
strength equations for CHS joints in the IIW (1989) recommendations Similarly, as a first assumption, the basic equation for TP-joints is
were developed. Later re-analyses by the first author, which were taken as :
confirmed by the research results of Makino et al. (1991), formed the
basis for the recommendations for plate and I to CHS chord joints given
in the CIDECT Design Guide No. 1 (Wardenier et al., 1991). ( )
N1* = 2.6 1 + 6.8β 2 (1 + 0.25η ) γ 0.2 Q f
f y0 t 02
sin θ1
(2)

Similar as in the CIDECT Design Guide No. 1, the equations for the
capacity for plate and I brace to CHS chord joints are related to those
for CHS joints. Compared to the CHS joint equations, the constants in
the Qu functions of the CIDECT recommendations have been reduced
by about 5%. This can be explained by the shorter effective chord
length in the ring model (Togo, 1967; van der Vegte, 1995).
For joints loaded by bending moments, the capacity for plate and I
brace to CHS chord joints is related to the axial load capacity by the
depth or the height of the brace (Wardenier et al., 1991).
The design equations for XP and TP joints in the updated IIW
recommendations have been developed in accordance with the
following philosophy.
PHILOSOPHY
For axially loaded XP-3 joints, which have a longitudinal and a cross
plate, the longitudinal plate will only be effective after considerable
local yielding of the cross plate connection, thus Eq. 1 overestimates
the capacity. Hence, it is recommended to relate this type of joint to
XP-1 joints, which is in line with earlier findings, e.g. Kurobane (1981).
The same applies to TP-3 joints.
Although for the XP-2 joints the coefficient of variation (CoV) is 6.8%,
the mean value is very high compared to the values for the other joints.
Therefore, the following, simplified equation for XP-2 joints used in
the CIDECT Design Guide No. 1, is adopted:
f y0 t 02
N1* = 5 (1 + 0.25η) Qf (3)
sin θ1

For a consistent approach with CHS joints and for improved
understanding by designers, the following starting points are used for
the development of the design equations:
- if possible, one design strength function is developed for all types of
axially loaded joints, and if not possible, then:
- for the axially loaded XP-1, 2, 3, 4 and 5 joints, the design strength
will be related to that for axially loaded CHS X joints.
- for the axially loaded TP-1, 2, 3, 4 and 5 joints, the design strength
will be related to that for axially loaded CHS T joints.
- the in-plane bending moment capacity will be related to the axial
load capacity by cip h1.
- the out-of-plane bending moment capacity will be related to the axial
load capacity by cop b1.
As a result, the general format shown in Table 2 in combination with
the Qu functions in Table 3 (derived from Eqs. 1 to 3) are initially used
to describe the strength of joints between plates or I sections and a CHS
chord. The reference strength function Qu, i.e. the first part in Eq. 1, is a
function of the diameter ratio β, the chord diameter to thickness ratio 2γ
and the plate height to chord diameter ratio η. The function Qf, referred
to as the chord stress function, uses the maximum chord stress as the
major variable. As shown in Table 3, the capacities of XP/TP-1, 4, 5
joints are directly related to those of axially loaded CHS X and T joints.
The in-plane bending moment capacity of the XP/TP-4, 5 joints is
related to the axial load capacity of the XP/TP-1 joint by h1. Further,
the out-of plane bending moment capacity for the XP/TP-1, 4, 5 joints
is related to the axial load capacity of XP-1 by 0.5b1. For simplicity, no

320
difference is made between the XP/TP-4 and 5 joints. For chord tension, C1 = 0.20 is adopted, equal for all CHS joints. For
chord stress ratios -0.8 ≤ n ≤ 0, Eq. 4 gives a mean = 1.01 and a CoV =
The equations given for the punching shear capacity are based on those 4.4%. Eq. 4 is identical to the expression used for CHS K gap-joints.
given in the CIDECT Design Guide No 1. These include the assumed
effective punching shear area. The n in Eq. 18 for Qf is only given for class 1 and 2 sections because
the range of validity of all equations is limited by class 1 and 2 sections
CHORD STRESS EFFECT for compression loaded chords. A further, general limit is:

For the analysis of the TP joints, the effect of the chord stress should be
known, however, for TP joints, no systematic investigation of the chord
stress effect is reported. From the investigations on CHS X and T joints,
it was concluded that the chord stress effect for X joints is marginally
more severe than that for T joints. Therefore, the chord stress function
is based on the numerical work of de Winkel (1998) for axially loaded
XP-4 joints. As e.g. shown in Figures 1a and 1b, the “de Winkel” data
fit well with Eq. 4 with C1 = 0.25.
f(n) = (1− | n |)C1
2γ ≤ 50, and for X joints: 2γ ≤ 40 (19)
ANALYSIS BASED ON EQUATIONS OF TABLES 2 AND 3
A comparison of the test data with the equations of Tables 2 and 3 is
presented in Figure 2. Figure 3 shows the comparison with the available
numerical data for XP-1 joints. The statistical data are given in Table 4.
In general, the agreement with the test data is good as illustrated in
Figure 2. However, the mean value for the numerical data in Figure 3 is
(4) rather low.

Table 4. Statistical values for XP data vs. Eq. 9 for XP-1, 3, Eq. 12 for
Table 2. Limit states criteria for XP and TP joints
XP-4, 5 and Eq. 11 for XP-2 (equations from Tables 2 and 3)
Criterion to be checked Axially loaded brace Brace bending
Joint No. of data Mean CoV (%) Remarks
moment
tests FE
Design strength: f y0 t 02 f y0 t 02 XP-1 4 1.15 5.0
chord plastification N1* = Qu Qf M1* = Qu Qf h1
sin θ1 sin θ1 XP-2 12 1.49 7.1 High mean
(5) (6) XP-3 4 1.04 8.5 Low mean
Design strength: For I sections with η ≤ 2 ( for axial loading XP-4 6 1.18 3.8
chord punching shear and out-of-plane bending) and RHS sections: XP-5 3 1.29 - Small number
(only for d1 ≤ d0 – 2t0) M ip M op
N1 t0 XP-1 21 0.91 8.8 Low mean
+ + ≤ 0.58 f y0 (7)
A1 Wel,ip Wel,op t1
All other cases: IIW (2008) vs. FE de Winkel for beta=0.25
1.2
N1 M ip M op t0
+ + ≤ 1.16 f y0 (8)
A1 Wel,ip Wel, op t1 1.0 2γ=15

0.8
Table 3. Initially assumed Qu and Qf functions for XP and TP joints
Initially assumed function Qu 0.6 2γ=30
Qf

In-plane Out-of-plane
Axial loading 0.4
bending bending
⎛ 1 + β ⎞ 0.15 b 0.2 2γ
XP-1 Q u = 2.6 ⎜⎜ ⎟⎟ γ (9) 0 0.5Q u 1
⎝ 1 - 0.7β ⎠ h 1 0.0
-1.0 -0.8 -0.6 -0.4 -0.2 0.0
TP-1 ( )
Qu = 2.6 1 + 6.8β2 γ0.2 (10) 0 0.5Q u
b1
h1
n

Fig. 1a Comparison of the “de Winkel” data for β = 0.25 with Eq. 4
XP-2
Q u = 5 (1 + 0.25η ) (11) Qu 0
TP-2 IIW (2008) vs. FE de Winkel for beta=0.65
⎛ 1+ β ⎞ Qu b 1.2
⎟⎟ (1 + 0.25η) γ
0.15
Qu = 2.6 ⎜⎜ 0.5Q u 1
XP-4/5 ⎝ 1 − 0.7β ⎠ 1 + 0.25η h1 1.0 2γ=15
(12) (14) (15)

TP-4/5
( 2
)
Qu = 2.6 1 + 6.8β (1 + 0.25η) γ 0.2 Q u b
0.5Q u 1
0.8

2γ=30
(13) 1 + 0.25η h1 0.6
Qf

Function Qf 0.4
Compression: Class 1 and 2 sections: Compression 2γ
0.2
Q f = (1 − n )0.25 (16) N0 M0
a a=1
n= + (18) 0.0
N pl,0 M pl,0 Tension: -1.0 -0.8 -0.6 -0.4 -0.2 0.0
Tension: a=2 n
Q f = (1 − n )0.20
(17)
Fig. 1b Comparison of the “de Winkel” data for β = 0.65 with Eq. 4

321
 Further, the re-analysis made clear that, with exception for the XP-4
XP- joints: experiments vs equations Tables 2 and 3 joints, the f(η) function shown in Eq. 20 gives a better fit with the
450
available data than f(η) = 1+0.25η. Hence, Eq. 20 is adopted:
400 XP-1
350 f(η) = 1+0.4η (20)
XP-2
300
Experiments

250
XP-3
However, if Eq. 20 is used for the f(η) function, the in-plane bending
200 moment capacity equation of XP-2 joints has to be modified according
150 XP-4
to:
100 0.8h1 N1* instead of h1 N1* .
50 XP-5
0
BASIC EQUATIONS FOR XP AND TP JOINTS
0 50 100 150 200 250 300 350 400 450
design strengths
Considering the results of the initial analysis, the basic equation for
Fig. 2 Comparison all XP test data with Eq. 9 for XP-1, 3, Eq. 12 for axially loaded XP joints (Eq. 1) is modified to Eq. 21, and to Eq. 22 for
XP-4, 5 and Eq. 11 for XP-2 (equations Tables 2 and 3) XP-2 joints.

XP-1 joints:
XP-1 joints: FE
FEde
de Winkel
Winkel / equation (9)
(7)
⎛ 1+ β ⎞ f t2
1.2
N1* = 2.2 ⎜⎜ ⎟ (1 + 0.4η) γ0.15 Qf y0 0
⎟ (21)
β=0.25
1.0
⎝ 1 − 0.7β ⎠ sin θ1
β=0.4
f y0 t 02
0.8 β=0.55 N1* = 5 (1 + 0.4η) Qf (22)
(7)
eq. (9)

sin θ1
FE / eq.

0.6 β=0.6

0.4 β=0.65 In a similar way, the basic equation for axially loaded TP-joint (Eq. 2)
β=0.8
is modified to Eq. 23, and to Eq. 22 for TP-2 joints.
0.2
β=0.9
0.0
0.0 0.2 0.4 0.6 0.8 1.0
( )
N1* = 2.2 1 + 6.8β 2 (1 + 0.4η)　
γ 0.2 Qf
f y0 t 02
sin θ1
(23)
β
Fig. 3 Comparison all XP-1 numerical data with Eq. 9 These equations result in the modified (and adopted) Qu functions shown
in Table 5.
Although these numerical data of de Winkel (1998) show 8% higher
results than the experiments used for calibration of the FE model, the Table 5. Adopted Qu and Qf functions for XP and TP joints
experiments were welded with a single sided V weld in combination Adopted function Qu
with a small back weld, resulting in an eccentricity, which causes In-plane Out-of-plane
slightly lower experimental results. Thus, the FE results are expected to Axial loading
bending bending
be approximately equal or slightly higher than that of a completely
comparable experiment. ⎛ 1 + β ⎞ 0.15 b
XP-1 Q u = 2.2 ⎜⎜ ⎟⎟ γ (9a) 0 0.5Q u 1
⎝ 1 - 0.7β ⎠ h1
A similar analysis for TP joints also showed large discrepancies
between various test series; in this case between the data obtained from
Akiyama et al. (1974) and all other test data. Comparison of the data in
TP-1 (
Qu = 2.2 1 + 6.8β 2 γ0.2 ) (10a) 0 0.5Q u
b1
h1
Makino et al. (1991, 1996) with those in Akiyama et al. (1974), shows XP-2
that Makino used Akiyama’s ultimate strength data for the TP-1 joints, Q u = 5 (1 + 0.4η ) (11a) 0.8Qu 0
TP-2
and the yield strength for the TP-2 joints which is a reasonable
⎛ 1+ β ⎞ Qu b
assumption considering the deformations. A further observation is that Qu = 2.2 ⎜⎜ ⎟⎟ (1 + 0.4η) γ0.15 0.5Q u 1
the capacities for a 0.03d0 deformation (Lu et al., 1994) approximately XP-4/5 ⎝ 1 − 0.7β ⎠ 1 + 0.4η h1
agree with the Akiyama yield values for TP-2 and the ultimate values (12a) (14a) (15)
for TP-1, which is also in line with the assumption of Makino et al.
(1991, 1996). However, it still does not explain the large differences in TP-4/5
( 2
)
Qu = 2.2 1 + 6.8β (1 + 0.4η) γ 0.2 Q u b
0.5Q u 1
mean value between the ultimate capacities of Akiyama’s data and (13a) 1 + 0.4η h1
those of the other researchers. Function Qf
Compression: Class 1 and 2 sections: Compression
Since no acceptable explanation can be found for the relatively low
capacities of the data obtained from de Winkel for XP-1 joints and the Q f = (1 − n ) 0.25
(16)
N0 M0
a a=1
Akiyama data for TP-1 and TP-2 joints, IIW sub-commission XV-E n= + (18)
N pl,0 M pl,0 Tension:
decided to reduce the constant in the equations such that the mean value Tension: a=2
of the deviating data of de Winkel (1998) and Akiyama et al. (1974)
Q f = (1 − n )0.20 (17)
should not be lower than 1.0, although this gives conservative
estimations for all other test data.

322
The equations are first compared with the data, excluding the numerical
XP-1 joint: FE de Winkel / IIW (2008)
“de Winkel” data and the “Akiyama” test data. The “de Winkel” and the 1.4
“Akiyama” data are separately checked with the equations in Table 5. β=0.25
1.2
β=0.4
XP-JOINTS 1.0

FE / IIW (2008)
β=0.55
0.8
Table 6 shows the statistical values for the data compared to the equations β=0.6
of Tables 2 and 5. 0.6
β=0.65
0.4
Table 6. Statistical values for XP data vs. eqs. given in Tables 2 and 5 β=0.8
0.2
Joint No. of data Axial compression In-Plane Bending β=0.9
tests FE mean CoV mean CoV 0.0
XP-1 4 1.35 5.0*) 0.0 0.2 0.4 0.6 0.8 1.0
21 1.07 8.8 β

XP-2 12/6 1.29 5.6 1.26 13.7 Fig. 5 XP-1 joint FE data of de Winkel (1998) (axial compression) vs.
XP-3 4 1.23 8.5*) equations of Tables 2 and 5
XP-4 6/7 1.21 5.7 1.58 12.5
24 1.00 11.5 TP-JOINTS
XP-5 3 1.30 *)
*) number of tests too small Table 7 shows the statistical values for the data compared to the equations
of Tables 2 and 5. It is remarkable that the correlation for the TP-2 in-
Axially Loaded XP Joints plane bending data is excellent and much better than that for the XP-2
joints.
The data versus the equations of Tables 2 and 5 for axially loaded XP
joints are presented in Figure 4. This figure shows an excellent agreement
between data and equations with marginal variations in mean value and XP-4 joints: TEST & FE / IIW (2008); In-Plane Bending

CoV for the various types of XP joints. The numerical data of de Winkel 2.0
for the XP-1 joints are shown in Figure 5. As already indicated, these 1.8
TEST & FE / IIW (2008)

results are low compared to the test data. 1.6 XP-4 IPB TEST
1.4
XP Joints Loaded by In-Plane Bending 1.2
1.0
As shown in Table 6, the CoV for XP-2 joints subjected to in-plane 0.8
bending is considerably higher than that for axial loading, whereas for 0.6 XP-4-IPB; FE
0.4 de Winkel
TP-2 joints (see Table 7) this is just the opposite. The combined data
for the XP-2 and TP-2 joints (axial load and in-plane bending) are 0.2
shown in Figure 9. 0.0
0.0 0.2 0.4 0.6 0.8 1.0
β
For the XP-4 joints, both the mean value and the CoV are high, because
Fig. 6a Data as a function of β
the ultimate capacities are probably obtained at different but high
rotation levels. Thus, some conservatism between the design resistance
and test data is recommended, especially since the numerical data of de XP-4 joints: TEST & FE / IIW (2008); In-Plane Bending
Winkel shown in Figure 6, give considerably lower results with a mean
2.0
value of 1.0 and a CoV of 11.5%. The numerical data of de Winkel are
1.8
especially low for the lower 2γ ratios.
1.6 XP-4 IPB
TEST / IIW (2008)

1.4
XP- joints: experiments vs IIW (2008) 1.2
450 1.0
400 XP-1 0.8
350 0.6 XP-4-IPB; FE
XP-2 0.4 de Winkel
300
Experiments

0.2
250
XP-3 0.0
200 0.0 10.0 20.0 30.0 40.0 50.0 60.0
150 XP-4
2γ
100 Fig. 6b Data as a function of 2γ
50 XP-5
0 Fig. 6 XP-4 joint in-plane bending data vs. equations in Table 5
0 50 100 150 200 250 300 350 400 450
design strengths

Fig. 4 XP joint test data (axial compression) vs. equations of Tables 2

and 5; excluding FE data of de Winkel (1998)

323
Table 7. Statistical values for TP data vs. eqs. of Tables 2 and 5 in his papers) is used, the CoV drops from about 11% to about 8%. This
Joint No. of data Axial compression In-Plane also shows the sensitivity of the data.
Bending
tests Akiyama mean CoV mean CoV XP+TP-1 joints: TEST & FE / IIW (2008);
TP-1 4 1.68 10.7*) de Winkel+Akiyama data
1.4
9 0.98 12.1
TP-2 2 2.04 *) 1.2

TEST & FE / IIW (2008)

TP-1
6/9 1.25 12.0 1.37 5.5 1.0 Akiyama
TP-3 2 1.62 *)
0.8
TP-4 2 1.75 *)
TP-5 1 1.52 *) 0.6
*) number of tests too small 0.4 XP-1
de Winkel
0.2
Axially Loaded TP Joints
0.0
The data versus the equations of Tables 2 and 5 for axially loaded TP 0.0 0.2 0.4 0.6 0.8 1.0
β
joints are presented in Figure 7. For TP joints, the correlation is not as
good as for axially loaded XP joints, and the mean value is rather high, Fig. 8 Test data for TP-1 (Akiyama et al., 1974) and numerical data for
indicating that the equations are too conservative. One recent test result XP-1 (de Winkel, 1998) vs. equations of Tables 2 and 5
of Voth and Packer (2007) is in the same scatter band as the data shown
in Figure 7. Further, as indicated in Table 7, the mean value for the XP/TP-2 joints: TEST / IIW (2008); IPB with factor 0.8
Akiyama TP-1 test results is only 0.98. Hence, also considering the 1.8
limited number of tests available, no further attempt is made to develop XP-2
1.6
a more accurate strength function for axially loaded TP-joints. compression
1.4

TEST / IIW (2008)

1.2 TP-2
In Figure 8, the Akiyama TP-1 test data are compared with the
1.0 compression
numerical XP-1 data of de Winkel. Although, the scatter is considerable,
the data are approximately in the same scatter band. 0.8 XP-2 IPB
0.6
TP- joints: experiments vs IIW (2008); 0.4
TP-2 IPB
excl. 2 outliers and Akiyama data 0.2
400
0.0
350 TP-1
0.0 20.0 40.0 60.0 80.0 100.0
300 2γ
TP-2
Experiments

250 Fig. 9 XP/TP-2 joint test data (axially compression and in-plane
200 TP-3 bending) compared to the equations in Tables 2 and 5
150
100
TP-4 COMPARISON XP AND TP-JOINTS
50 TP-5 The new TP joint equations (including the Qf effect) give, except for β
0 close to 1.0, a slightly larger capacity than the XP joints. This trend is
0 50 100 150 200 250 300 350 similar as that for CHS T and X joints and physically correct. Figure 10
design strengths
shows the ratio between the capacities of TP and XP joints as function
Fig. 7 TP joint test data (axial compression) vs. equations of Tables 2 of β, however, excluding the Qf effect. This means that the TP joint
and 5; excluding Akiyama data and 2 extremely high outliers data have to be reduced by the Qf effect which depends on the chord
length.
TP Joints Loaded by In-Plane Bending
TP-1: TP/XP IIW (2008); excl. Qf
The only available data for TP joints loaded by in-plane bending are
1.8
those for TP-2 joints obtained from Akiyama et al. (1974). As stated
1.6
before, the correlation with the equations of Tables 2 and 5 is excellent. TP/XP IIW 2008
1.4
For a better comparison between the various data series for XP-2 and (excl. Qf)
TP/XP; IIW (2008)

TP-2 joints, all data (i.e. for axial loading and in-plane bending) are 1.2

illustrated in Figure 9. 1.0

0.8
It is peculiar (see Tables 6 and 7) that for the data for XP-2 (axial 0.6
loading) and TP-2 (in-plane bending) the CoV is very small, whereas 0.4
for the data for XP-2 (in-plane bending) and TP-2 (axial loading), the 0.2
CoV is relatively large. Especially for these data which are based on 0.0
yield load, a similar CoV would be expected. A possible explanation 0.0 0.2 0.4 0.6 0.8 1.0
could be that different definitions are used in the database for the β
determination of the yield load. For example, if for the TP-2 data with Fig. 10 Comparison of the capacity for TP-1 vs. XP-1 joints according
axial loading, another Akiyama definition (he has given two definitions to equations of Tables 2 and 5 (excl. Qf effect)

324
UT-OF PLANE-BENDING
Although no test data exist for out-of-plane bending, the same relation
to the axial load capacity is used as that for CHS joints.
XP AND TP JOINTS LOADED IN TENSION
XP and TP joints loaded in tension exhibit a considerable larger
capacity than in compression. However, the capacity will be limited by
the deformation criterion. Further, the deformation capacity for joints
with high β ratios may drop considerably. As a conservative
assumption, the same design equations are proposed as for XP and TP
joints loaded in compression. At present, additional research is ongoing
at the University of Toronto, see Packer (2006), Willibald et al. (2006)
and Voth and Packer (2007).
COMPARISON WITH THE CIDECT (1991) EQUATIONS
Figure 11 illustrates the comparison between the capacity
recommended by the CIDECT Design Guide No. 1 (1991) for XP-1
joints and the capacity of the new IIW (2008) recommendation. The
capacities of the XP-3, 4, 5 joints are not given because these are
directly related to the capacity of the XP-1 joint. The figure shows that
for small β and low γ ratios, the new equation gives a slightly lower
capacity than the CIDECT equation. For XP/TP-2 joints, the capacity is
unchanged.
NEW CIDECT DESIGN GUIDE (2008)
In the previous sections, reference is made to the CIDECT Design
Guide No. 1 (Wardenier et al., 1991), however, this version is recently
updated and made fully consistent with the new IIW (2008) design
recommendations (Wardenier et al., 2008).
XP-1: CIDECT (1991) / IIW (2008 for 2γ =15, 30 and 45)
1.4
CIDECT (1991) / IIW (2008)
1.2
1.0
0.8
0.6
0.4
0.2
0.0
0.0 0.2 0.4 0.6
β
Fig. 11 Comparison of CIDECT (1991) capacity vs new IIW (2008) for
XP-1 joints (highest points are for 2γ = 15 (red dotted line) and lowest
for 2γ = 45 (blue line))
CONCLUSIONS
This paper summarises the analyses and resulting design equations for
plate or I sections to CHS chord joints included in the new IIW (2008)
recommendations. The conclusions can be summarized as follows (see
Table 1 for the joint classification):
- the design strength of axially loaded XP/TP-1, 4 and 5 joints is
related to that for axially loaded CHS X and T joints, however, the
constant is lower (2.2 vs. 2.6).
- the design strength of axially loaded XP-3 and TP-3 joints is taken
the same as that of XP-1 and TP-1 joints, respectively.
- for the design strength of XP/TP-2 joints, a simplified function is
adopted, similar to that in the CIDECT Design Guide No. 1 (1991).
- the design strength for XP/TP-4 , 5 joints loaded by in-plane bending
moment is related to the axial load capacity of XP/TP-1 joints by h1;
for XP/TP-2 joints, the design strength is related to the axial load
capacity of the same joints by 0.8h1.
- the design strength for XP/TP-1, 3, 4 , 5 joints loaded by out-of-plane
bending moments is related to the axial load capacity of the same
joints by 0.5b1.
For the axial load capacity of the various types of XP joints, an excellent
agreement exists between the experimental data and the equations, with
marginal variations in mean and CoV. However, the numerical data of de
Winkel (1998) show considerably lower results than all other data.
For axially loaded TP joints, in general, the correlation is not as good as
for axially loaded XP joints and the mean value is relatively high,
indicating that the equations are too conservative. However, the data of
Akiyama et al. (1974) completely deviate from all other data and show
lower results.
The “de Winkel” data and the Akiyama test data are separately checked
with the equations and these are generally in line with each other. The
differences with the other data cannot be explained with the evidence
available and the deformations at the ultimate load capacity. Therefore,
for these data, a minimum mean value of approximately 1.0 is required.
Considering the scatter in data and deviations in dimensions and yield
stress, for the other data, the required mean value between data and
design strength is taken as 1.25 to 1.30 (van der Vegte et al., 2008).
It is further peculiar that the data for XP-2 (axial loading) and TP-2 (in-
plane bending), here both based on the yield load, have a very small CoV,
whereas the data for TP-2 (axial loading) and XP-2 (in-plane bending)
show a relatively large CoV. Especially for these data which are based on
yield load, a similar CoV would be expected. A possible explanation
could be the use of different definitions in the database for the
determination of the yield load.
CIDECT 1991 Although, no test data exist for out-of-plane bending, a similar relation to
vs IIW 2008
the axial load capacity is used as for CHS joints.
Compared to the equations in the previous version of the CIDECT
Design Guide No. 1 (1991), the new equations give for small β and low
γ ratios, slightly lower capacities. For XP/TP-2 joints, the capacity is
unchanged.
ACKNOWLEDGEMENT
0.8 1.0
Appreciation is extended to our colleagues Prof. J.A. Packer from
University of Toronto, Canada and Prof. X-L. Zhao from Monash
University, Australia for the continuous discussions in the framework of
IIW Sub-commission XV-E.
REFERENCES
Akiyama, N, Yajima, H, and Ohtake, A (1974). “Experimental Study
on Strength of Joints in Steel Tubular Structures,” Journal of
Society of Steel Construction, Vol 10, No 102, pp 37-68 (in
Japanese).
Ariyoshi, M, and Makino, Y (2000). “Load-Deformation Relationships
for Gusset Plate to CHS Tube Joints under Compression,” Int
Journal of Offshore and Polar Engineering, Vol 10, No 4, pp 292-
300.
325
IIW (1989). “Design Recommendations for Hollow Section Joints –
Predominantly Statically Loaded,” 2nd Edition, International
Institute of Welding, Sub-commission XV-E, Annual Assembly,
Helsinki, Finland, IIW Doc XV-701-89.
IIW (2008). “Static Design Procedure for Welded Hollow Section
Joints - Recommendations,” 3rd Edition, International Institute of
Welding, Sub-commission XV-E, IIW Doc XV-1281-08.
Kurobane, Y (1981). “New Developments and Practices in Tubular
Joint Design (+ Addendum),” International Institute of Welding,
Annual Assembly, Oporto, Portugal, IIW Doc XV-488-81.
Lu, LH, Winkel, GD de, Yu, Y, and Wardenier, J (1994). “Deformation
Limit for the Ultimate Strength of Hollow Section Joints,” Proc 6th
Int Symposium on Tubular Structures, Melbourne, Australia,
Tubular Structures VI, Balkema, Rotterdam, The Netherlands.
Makino, Y, Kurobane, Y, Paul, JC, Orita, Y, and Hiraishi, K (1991).
“Ultimate Capacity of Gusset Plate-to-Tube Joints under Axial and
In Plane Bending Loads,” Proc 4th Int Symposium on Tubular
Structures, Delft University Press, Delft, The Netherlands.
Makino, Y, Kurobane, Y, Ochi, K, Vegte, GJ van der, and Wilmshurst
SR (1996). “Database of Test and Numerical Analysis Results for
Unstiffened Tubular Joints,” IIW Doc XV-E-96-220.
Packer, JA (2006). “Tubular Brace Member Connections in Braced
Steel Frames,” Houdremont Lecture, Proc 11th Int Symposium on
Tubular Structures, Quebec City, Canada, Tubular Structures XI,
Taylor & Francis Group, London, UK.
Togo, T (1967). “Experimental Study on Mechanical Behaviour of
Tubular Joints,” PhD Thesis, Osaka University, Japan, (in
Japanese).
Vegte, GJ van der (1995). “The Static Strength of Uniplanar and
Multiplanar Tubular T and X Joints,” PhD Thesis, Delft University
of Technology, Delft, The Netherlands.
Vegte, GJ van der, Wardenier, J, Zhao, X-L, and Packer, JA (2008).
“Evaluation of the New CHS Strength Formulae to Design
Strengths”, Proc 12th Int Symposium on Tubular Structures,
Shanghai, China, Tubular Structures XII, Taylor & Francis Group,
London, UK.
Voth, AP, and Packer, JA (2007). “Branch Plate Connections to Round
Hollow Sections,” CIDECT Interim Report 5BS-5/07.
Wardenier, J (1982). “Hollow Section Joints,” Delft University Press,
The Netherlands, ISBN 90-6275-084-2.
Wardenier, J, Kurobane, Y, Packer, JA, Dutta, D, and Yeomans, N
(1991). “Design Guide for Circular Hollow Section (CHS) Joints
under Predominantly Static Loading,” Published by Verlag TUV
Rheinland GmbH, Köln, Germany, ISBN 3-88585-975-0.
Wardenier, J, Kurobane, Y, Packer, JA, Vegte, GJ van der, and Zhao,
X-L (2008). “Design Guide for Circular Hollow Section (CHS)
Joints under Predominantly Static Loading,” Published by Verlag
TUV Rheinland GmbH, Köln, Germany.
Willibald, S, Packer, JA, Voth, AP, and Zhao, X-L (2006). “Through
Plate Joints to Elliptical and Circular Hollow Sections,” Proc 11th
Int Symposium on Tubular Structures, Quebec City, Canada,
Tubular Structures XI, Taylor & Francis Group, London, UK.
Winkel, GD de (1998). “The Static Strength of I-Beam to Circular
Hollow Section Column Connections,” PhD Thesis, Delft
University of Technology, Delft, The Netherlands.
Zhao, X-L, Wardenier, J, Packer, JA and Vegte, GJ van der (2008).
“New IIW (2008) Static Design Recommendations for Hollow
Section Joints,” Proc 12th Int Symposium on Tubular Structures,
Shanghai, China, Tubular Structures XII, Taylor & Francis Group,
London, UK.

326