Sie sind auf Seite 1von 8

Engineering Strm-tt¢re~.Vol. 19. No. 3. pp. 259 266. 1'..~.

)7
('opyrlghl .~.. 1996 [-Isc'.icr Science [.Id
Prinlcd in Great Brilain. All rights reserved
EI.SEVIER
PII: S0141-0296(96)00101-0 0141-0296/97 5.,17.(X) + ().IX)

AWS vs International Design Rules


for circular tubular K-connections
Yoshiaki Kurobane and Kenshi Ochi
Department of Architecture and Civil Engineering, Kumamoto University, KunlanlotO 860,
Japan
(Received June 1996: revised version accepted August 1996)

The American Welding Society (AWS) rules are compared with the
International Institute of Welding (IIW) rules for circular tubular
connections because of the differences between the two. Subjects
discussed include: accuracy in prediction, effects of size and
material properties, premature failures due to local buckling of
braces and due to fracture at weld toes, and applications to multi-
planar connections, focusing on the ultimate behaviour and design
of K-connections under axial brace loading for the sake of concise-
ness. One important conclusion is that the AWS rules lead to rela-
tively unsafe design for K and KK-connections with stocky chords.
Suggestions to enhance the reliability of the AWS rules are made.
Copyright © 1996 Elsevier Science Ltd.

Keywords:connection, tubular joint, multiplanar connection,


ultimate capacity, local buckling, size effect

1. Introduction nections, are outside the scope of this study. However. the
multiplanar K-connection (called the KK-connection
The American Welding Society (AWS) Code t provisions hereaftcr) shows behaviour similar to the planar K-connec-
for circular tubular K-connections are compared with simi- tion. Although study of KK-connections is currently in pro-
lar provisions proposed in the International Institute of gress, these connections will be included in the following
Welding (IlW) Design Recommendations -~. Although a few discussion.
differences in scope exist between the two codes, both
codes recommend ultimate limit state criteria for the same
tubular connections. The two criteria differ significantly in Notation
spite of the fact that the existing database is their common The following symbols are used in this paper
basis. These differences are discussed herein. A~ 7r(D - T) 7; cross-sectional area of chord
Direct comparison of the design rules between the two COV coefficient of variation
codes could complicate the points in question because dif- D outside diameter of chord
ferent values of the resistance and load factors and simpli- d outside diameter of compression brace
fying assumptions are adopted in the codes. A comparison E elastic modulus
is made between the ultimate limit state equations on which F,, local buckling strength of compression brace in
the design rules are based. The ultimate resistance equa- K-connection
tions for the IIW Recommendations have been cited by F,i, I'..K/I r r ( d - t ) t ] , shcll bending strength of K-con-
Kurobane et al. 3, while those for the AWS Code are nection
described (with their background) by Marshall 4 as well as fv yield strength of tube material
in the code itself. F, ultimate tensile strength of tube material
The IIW design equations for tubular connections have f function explaining variables in IIW equation
achieved a wide international consensus, and are being g longitudinal gap in K-connection or KK-
adopted for Eurocode 35. Essentially the same equations are connection
also used in Canada 6 and Japan 7. The IIW Design Recom- g! transverse gap in KK-conncction
mendations are fully described in the Comit6 International L longitudinal distance between footprints of two
pour le D6veloppement et I'Etude de la Construction Tubu- braces
laire (Cidect) design guide s, which is easiest to read. N axial force in chord on compression brace side
No great difference is found between the AWS and IIW t/ N/[A,F,] nondimensional axial stress in chord
equations for simple tubular connections besides K-connec- P axial load in brace
tions. These connections, which include T, Y and X con- P, ultimate limit state load for connection

259
260 Design rules for circular tubular K-connections: Y. Kurobane and K. Ochi

ultimate limit state load for K-connection accord-


I:~ = 1 + 0.305n - 0.285n"
ing to equation (3)
V2~
ultimate limit state load for KK-connection ' ." -
Pu.KK
Q function of explaining variables in A W S equations l:=("/"
~./",, /
(3)
Rm mean resistance
R. nominal resistance These equations predict ultimate limit state loads when
7" chord wall thickness K-connections fail by' localized plastic shell bending
t wall thickness of compression brace deflection of the chord walls, although the above A W S
v~ coefficient of variation equations are not applicable to connections with over-
Og chord ovalizing paramcter, or local buckling para- lapping braces. The A W S code provides a separate formula
meter applicable to overlapped K-connections. However. the lat-
/3 d/D, diameter ratio or safety index ter formula, originally used for the allowable stress design.
y D / ( 2 T ) , chord thinness ratio is not assessed herein because the A W S code shows no
0 angle between chord and compression brace reliability analysis on this forlnula. In overlapped K-con-
,t, resistance factor, or out-of-plane angle between nections shell bending deflection is also observed in the
planes in which braces lie brace walls. The IIW equations also give the ultimate
capacity of overlapped K-connections by' a continuous
curve applicable also to the range g < 0 . The primary con-
2. Accuracy in prediction
cern should be how accurate these equations are.
The ultimate strength prediction equations for K-connec- Predicted ultimate capacities are compared with existing
tions used in the A W S and llW codc provisions are desig- test results in Figure I. where the simple chord ovalizing
nated as equations ( I ) - ( 3 ) (see Notation and Figure 6 parameter is used for the A W S predictions. The ultimate
for symbols). capacity represents the nmximum load (the first peak load
The A W S equation with the simple ovalizing parameter if two peak loads exist) determined by chord wall plast-
ce is ificalion, as is defined in the past study 3. The database used
here is reliable not only because it omits inappropriate tests
1.7 0.18'1 F,T e but also because it includes all the reliable test results col-
P,, = 67r/3 + ] Q( f~7 . . . . ') Q / s i n 0 (1) lected from the open literature". The mean values and coef-
• Ot' 13
ticients of variation ( ( ' O V ) of lest-to-prediction ratios are
summarized in Table I. Clearly the A W S predictions scat-
where
ter much more than the IIW predictions.
The resistance factor for the LRFD is approximately
a = 1.0 + 0.7 g / d (1.0 <-- a < 1.7) given by
Qt~ = 1.0 (/3 -< o.6)
d} = (R,,,/R,,) exp( - ().55/3Vu) (4)
0.3
/3 > 0.6)
Q~ =/3( I - 0.833/3) If thc COV, designated by Vk in the above equation,
increases from 0.1 to 0.2, the resistance factor decreases
Qt = 1.0 - 0.03y n-'
by about 25%, which may not be a serious drawback to
overall economy.
while the alternative A W S equauon proposes using the
However, if values of COV become greater than 0.2, it
exponential ovalizing parameter ¢~ written as
is natural to suspect that the prediction models may be suf-

oe= 1 + 0 . 7 10000 I

liill '[ t l l l l i, Iiii


all h,'accs at a .iolnl ] . /i II

= = LI \ D T / 2
[P sin O],~.,~.,,.,,c~.b,,~.,,,~".,hich ,, apphc,

(2) 1000 ILl


IIIIII I
ti- t/
"i i~'~l
, ~....
1111 '
,
I
II
"II
in place of ¢~ in equation ( 1 ).
The llW equation is i till I II
P,,.A =f, f, f2 f, f~ f~ TeF,. Ill 1
f,=2.11 (I + 5 . 6 6 / 3 )
~- " llotq I I II I lll]ollW I
f, = (27)" 20,, • IIIIII I I]I],AWS I

0.00904 (2T) .24


.f2= I + .....
exp (0.508
g - 3.04
7" - 1.33) + 1
10
/ [ [1
10 100 1000 10000
P uPREDICTON (kN)
1 - 0.376cos2# Figure 1 T e s t r e s u l t s c o m p a r e d with predictions f o r all n o n -
f~ = sin O overlapping connections
Design rules for circular tubular K-connections: Y. Kurobane and K. Ochi 261

Table 1 U l t i m a t e strength: m e a n s and COVs o f t e s t - t o - p r e d i c t i o n ratios

Database Prediction equation Mean COV


(1) (2) (3) (4)

Nonoverlapping c o n n e c t i o n s w i t h D > 139 m m equation ( 1) 1.70 0.243


e q u a t i o n s (1) and (2) 1.74 0.282
equation ( 1) 1.56 0.206
All n o n o v e r l a p p i n g c o n n e c t i o n s e q u a t i o n s (1) and (2) 1.60 0.219
e q u a t i o n (3) 1.00 0.094
All c o n n e c t i o n s e q u a t i o n (3) 1.00 0.104

COV w a s c o m p u t e d using d e g r e e s o f f r e e d o m e q u a l to the n u m b e r o f s p e c i m e n s m i n u s o n e

1 ~_ ~/~Y =10
ish adoption of thickness squared strength formulation in
the Yura-based American criteria. Test-to-prediction ratios
o8 • s ---- . . . . . . . . . . . for all nonoverlapping K-connections are plotted against
AW S ~ ~ - - ~"----------- D/T (=23,) in Figure 3, which shows that the AWS equa-
IIW 06 ,,, '1 tions allow less margin of safety as D/T decreases while
the IIW predictions cluster tightly around the 1.0 line over
the whole range of D/T. The systematic errors, unless exist-
04
~'-,~. " ' \ Y :40
ing design equations are altered, have to be corrected by
applying a constant bias factor besides a resistance factor,
0.2 ,=25 - - SIMPLE (z which would make the AWS rules more uneconomical.
-- -- -- EXPONENTIAL (z Although accuracy in prediction is one of the leading
0 i i i i factors affecting the reliability of design, any design equ-
0 5 10 15 20 ation should have a simple form and wide range of appli-
g/T cation so that human errors are automatically avoided.
However, an accurate equation is capable of detecting
Figure2 Ratio o f A W S to IIW p r e d i c t i o n s (/3 = 0.5, # = 45 °) unexpected low-strength failures induced by failure modes
different from shell bending failure lbr which both equa-
fering from a lack-of-fit. To see this, the ratio or" AWS to tions already cater, as will be demonstrated later.
IIW predictions is plotted against g/l" in Figure 2 for typi-
cal Warren-type K-connections with the material property 3. Effects of size and material properties
of F,./F, = 2/3. This ratio tends to approach to the AWS
R,,/R,,, factor of 1.8/2.44 = 0.74 (see AWS D1. I Commen- The AWS equations were based on the database from
tary C2.40. I.I) as g/T gets close to about 12. because the which data for small-size specimens had been deleted. Con-
AWS formulae give lower bound predictions of ultimate tinuum mechanics shows that no size effect exists if failures
capacity. However, on the assumption that the llW equa- are governed by the plastic bending deflection of the tube
tions are accurate enough to predict the mean resistance of walls. If failures are governed by brittle failure under ten-
K-connections, the following observations can be made. sion, however, size effects should exist. In the latter case
The errors in AWS predictions possibly contain a system- capacity criteria different from those governing the plastic
atic component caused by ignoring the combined effects of capacity of connections have to be used. The observed con-
g/T and y: that is, the connection strength varies with about nection resistances governed by these two different failure
1.8 power of thickness and is more sensitive to variation modes should not be mixed up when lbrmulating resistance
of g/T as the chord becomes lighter. This was recognized prediction equations.
by Marshall 4 as the difficulty that can be traced to the slav- A possible source of size effect on the resistance of tubu-
lar connections is the fact that small-size specimens tend
to have rclatively larger weld sizes than larger specimens.
4.5 i i _ I *
AWS SIMPLE AWS However, such a size effect can be sorted out in the process
4 / EXPONENTIAL of regression analysis. As seen in equation (3), one of the
explaining variables is the nominal gap size minus
3.04 mm. The value of 3.04 mm was determined by the
regression analysis as the best estimate of one of the para-
meters included in the capacity prediction model. In this
• f \ ~ S MEAN
way the effect of a larger weld size on the gap size is elim-
inated. As seen in Figure 1, ratios of observed to predicted
capacities distribute randomly about the diagonal over the
whole range varying from 30 kN to 3000 kN when the IIW
' IIW MEAN equation is used, which proves that errors are independent
AWS NOMINAL
o of the capacity and, therefore, of the size of a connection.
o 20 40 60 80 100 120 On the contrary, the AWS equation (equation (1)) under-
D/T predicts capacities of small-size specimens, showing a sig-
Figure3 T e s t - t o - p r e d i c t i o n ratios f o r all n o n o v e r l a p p i n g con- nificant size effect.
n e c t i o n s a c c o r d i n g to e q u a t i o n s ( 1 ) - ( 3 ) The K-connections included in the database are made of
262 Design rules for circular tubular K-connections: Y. Kurobane and K. Ochi

various steels ranging from hot-finished mild steel to high- bending strength as well as the local buckling strength
strength steel with a specified minimum tensile strength of observed in stub-colunm tests.
785 MPa. The effect of material properties is explained by The local buckling strength is given as a function of the
a function of the yield stress ratio F,//-',, of the chord nondimensional local buckling parameter ct = (Eli",) (tldl
materials in the IIW equation. In the A W S equations /", is of the compression brace and the joint resistancc to shell
limited to not exceed 2F,,/3. Therefore, both eqt, ations tend bending failure'", which can be rewritten in a simplilied
to reduce the predicted resistances of connections made of form as
materials with high-yield stress ratios, liowever, the A W S
2/3 limitation is an historic provision common to most /-';;, = 0.217 (t';a (F, +/-',;,) (5)
design codes and restricts the use of modern high-strength
steels. The IIW equations can be adapted for any steel. in which /-,, signifies the local buckling strength and F,,
including high-strength steel, having a yield stress ratio signifies the shell-bending strength, both being represented
greater than 2/3. in terms of the average stress on a cross-section of com-
pression brace, while F, used here designates the yield
stress of the compression brace.
4. Local buckling of compression braces The strength interactions between F,, and F,;, are ex-
Local buckling of compression braces in arcas adjacent to plained as t\)llows. The compression brace sustains a sec-
connections is one of the typical failure modes observed ondary bending moment with greater compression on the
in both gapped and overlapped K-connections as shown in toe side than on the heel side ()wing to uneven support
I"ig,re 4. When the brace local buckling occurs in over- given to the brace by the chord and tension brace. When the
lapped K-connections. the failure mode varies with the load approaches the brace local buckling load, the bending
diameter-to-thickness ratio of the brace. Namely. the failure moment at the brace end decreases because the moment is
mode varies from dominating local buckles in the com- redistributed owing to local buckling or yielding, and then
pression brace wall ( see bij4ure 4a ) to a mixed failure mode the compression brace carries a further increase of com-
accompanying shell bending deflection in both the chord pressive load. When the shell bending failure mode begins
and brace walls as the brace becomes heavier (see I-igure to appear, however, the redistribution of bending moment
4b). Nevertheless. local buckling and shell bending failure is prevented.
modes should be strictly distinguished because the local Although equation (5) was first proTx~sed m 1986 based
buckling strength can be significantly lower than the shell on test results for 43 specimens, a recent FE analysis was
able to accurately reproduce both failure modes and I o a d -
detk~rmation curves observed in the past tests '~. The sig-
(a) nificance of brace local buckling has also been noted by
another researcher '2. Equation (5) is the basis for an
additional limitation imposed on the design rules for K-
connections in the IIW design rules. The A W S design rules
do not provide such a safeguard against brace local buck-
ling except for ordinary limiting diameter-to-thickness
ratios. As can easily be understood from equation (5), this
may endanger K-connections with thick-walled chords and
thin-walled braces designed according to the A W S code. It
should be noted that /",, does not increase in proportion
to F,,,.

5. Fracture at weld toes


Cracks are frequently observed along the toes of welds
between a tension brace and a chord in the gap region of
the chord wall. In most cases these cracks extend slowly
with no significant effect on load-deformation curves of
connections until the chord wall sustains large plastic
(b) deflection. However, there are a few examples in which
connections lose stability as soon as the first cracks are
found, even before plastic deflection of the chord walls
becomes extensive.
One example of such premature tensile failure is two
replicate tests of identical specimens performed at the
University of Texas ~'. Specimen configuration, loading
arrangements and a load-deformation curve reproduced
from the original report on these tests are shown in l"ieure
5. These specimens are subjected to relatively high tensile
stress in the chords and have an angle of intersection of
30 ° between the tension braces and the chords. The two
specimens showed a sudden reduction in stiffness as soon
Figure4 Local buckling of compression braces: (a)gapped as cracks were found at the weld toes (at the load step 8 in
connection; (b) overlapped connection Figure 5). Although the two specimens withstood a further
Design rules for circular tubular K-connections: Y. Kurobane and K. Ochi 263

400
I:~ -~L~.88 kips capacity multiplied by correction terms. These equations
12
were based on the database ~-2~ obtained as of 1990. The
~FUL STRENGTH289 kips Y / Cidect design guide recommends a simple constant value
3oo s ~ ~ ,o......-----'~ I
of 0.9 as the correction factor for KK-connections. Thus,
~ i r s a E ~ wir. ~SRA~-~.~I
P,o" ~=cE CRACK e,Y~.O,~ 3o" I the ultimate capacity is given by
[kips) 'IE'LDING A BRANCH B
2O0 P..KK = 0.9P..K (6)
PgO"
t ~ C~I~ B THRU Since then extensive investigations on the XX, "rr, TX and
IO0 KK-connections have teen performed ~--~-'". Subsequent
~1- , - - f - ~ - - - ~ P3o'" ZPgo" developments in this area are quite significant and have
2 A reached the stage that warrants devising new useful design
I I I I formulae for multiplanar connections. Later in this section,
1.0 2.0 3.0 4.0
AXIAL OEFLECTION (in.) the AWS equations will be assessed in the light of the most
recent research results.
Figure 5 Test results f o r K - c o n n e c t i o n a c c o r d i n g t o Yura ~3 (1 in The AWS code shows definite design criteria for multi-
= 25.4 m m , 1 kip = 4.45 kN)
planar connections. Moreover, the AWS equation (equation
(1) with cr given by equation (2)) is the only exception
increase in load after large plastic deflection of the chord
that proposes general design criteria applicable to any type
walls, the load-deformation curve clearly indicates that the
first peak load (this load corresponds to the ultimate of nonoverlapping multiplanar connections without a need
for joint classification.
capacity according to the definition in both the AWS and
IIW equations) was controlled by tensile failure. The parameter o~ in equation (2) plays a role of incorpor-
ating not only a multiplanar effect due to chord wall ovaliz-
The IIW equation suggests that these specimens sus-
tained a premature low-strength failure. The predicted ing (circumferential bending) but also a membrane shell
capacity according to the I1W equation is equal to 1620 kN effect due to loads at position L distant from the reference
(364 kips) as compared with 1291 kN (289 kips) observed brace. The cosine term in equation (2) represents the influ-
in the test. The test-to-prediction ratio is 0.8, which shows ence of braces in other positions around the circumference.
For KK-connections under symmetrical axial brace loading
a significant reduction in ultimate resistance. On the con-
as shown in Figure 6, the AWS criteria predict that chord
trary, the predicted capacity according to the AWS equation
wall ovalizing is suppressed as the out-of-plane angle d~
(equation ( 1 )) is 739 kN ( 166 kips), giving the test-to-pre-
diction ratio of 1.75. Thus, the AWS equations cannot cap- increases from 0 ° to 90 ° allowing the maximum beneficial
effect when ~b = 90 °. However, this is not the case in actual
ture an occurrence of premature failure, which may result
in overlooking this important difference in failure mode. KK-connections.
The capacity of KK-connections increases with ~b when
Another example of tensile failure includes a test of large
K-connections performed at Sumitomo Metal Industry ~4. In d~ is small. When qb is small, the two compression braces
these tests cold-tent high-strength steels with measured act as one member and penetrate the chord wall together
(see Figure 7a). As 'b increases further and reaches a cer-
yield stresses varying from 739 to 833 MPa were used for
tain value (about 50°), a different failure mode starts to
the chords.
A very simple analysis was attempted to predict the criti- appear: the chord wall sustains radial deflection between
cal load at which tensile failure of this nature started L~. the two compression braces, eventually creating a fold
between them (see Figure 7b). Paul 22 distinguished
Namely, a component of axial load in the tension brace
was assumed to be carried by an imaginary tensile strip between the former and latter failure modes as failure types
accommodated in the gap region. A tentative conclusion 1 and 2, respectively. As soon as the failure mode changes
drawn from this study was to limit the minimum gap size from type 1 to type 2, the capacity of KK-connections sud-
to prevent low-strength tensile failure. A typical value of denly changes. The AWS criteria ignore this dramatic
the limiting gap size for large K-connections like those change in failure mode.
Finite element analysis results were found to reproduce
tested at Sumitomo and at the University of Texas was
found to be g/T = 5. This result implies that an American well the behaviour of KK-connections observed in tests 26.
practice of using the minimum gap size of 2 in ~" could incur Accurate ultimate capacity prediction equations have re-
a danger of low-strength tensile failure in large K-connec-
tions. N .P
Many factors are involved in the early developments of
cracks at the weld toes. Initiating cracks were due to ductile
tensile failures in all the existing tests results, although
these cracks may change to fast unstable cracks if adverse
effects are combined ~7. The use of clean, low-carbon and
fine-grained steel may be effective in preventing crack
extension TM. No definite criteria to prevent these cracks are
given in either the AWS or the IIW code provisions except
for general cautions on material selection.

6. A p p l i c a t i o n to m u l t i p l a n a r c o n n e c t i o n s
The design equations for multiplanar connections in the D
Cidect design guide have a format of uniplanar joint Figure 6 Typical Warren type KK-connection
264 Design rules for circular tubular K-connections: Y. Kurobane and K. Ochi

D / T=42-80
(a) 1.4
•-A-.....~I~ ' / /
1.2 ,~ -.
=,[ r I~OlT=80

u_ 0.8

o 06
U.l
I:I::
rr IT= 18-30
0 0.4 l
0
FAILURE I FAILURE
TYPE 1 l TYPE 2
(b) 0.2 I
] / / 0.215
r
0.0
0.0 0.2 0.4 0.6
g~/ O
Figure8 Correction factor for KK-connections

results showed that: the multiplanar effect due to chord ova-


lizmg was strongly correlated with the geometrical variable
,~,/I): and it was difficult to relate the multiplanar effect
with & only. Therefore, the A W S equations involve sig-
nilicant errors in the evalt, ation of multiplanar effects.
Figure7 Failure modes for KK-connections: (a)failure type 1; However. these errors due to inappropriate modelling of
(b) failure type 2
multiplanar effects are even less inlluential as compared
with systematic errors due to the thickness squared strength
cently been developed for KK-connections by combining fl~rmulation m the A W S equations.
both test and FEA results > >. Of these, the simplest poss- Test and numerical results divided by A W S predictions
ible prediction equations, although less accurate, are selec- are plotted against DIT in Figure 9. Again this figure
ted and proposed herein. reveals that the A W S predictions not only scatter widely
When a type I failure occurs, the ultimate capacity of but also show a systematic component erring on the unsafe
KK-connections is predicted by side as I)/T decreases. The systematic component is even
steeper for KK-connections than for K-connections: the
P,,.~,,~ = 0.233(27) "*~-' P,,.j, (7 ) capacity of KK-connections actually' varies as the 1.3 to 1.5
power of thickness. Test and numerical results divided by
with COV = 0 . 0 6 1 9 . predictions according to equations {7 ) and ( 8 ) are also plot-
When a type 2 failure occurs, the capacity prediction ted in Figure 9. No doubt these proposed equations com-
equation is pare well with the database, without any systematic errors.
The test and numerical results are divided simply by' the
corresponding K-connection capacities according to equ-
P,,.*,i, = 0.539 (. 1 - 0.31 'g' -7)" 's t,,.~
D I ] ('~ ~ (8) ation (3). The mean and sample standard deviation of these
ratios are calculated at 0.94 and 0.127. This demonstrates
with COV = 0.0960. that the current Cidect design equations for KK-connections
The border between type 1 and 2 failure modes is at ,~,11) are, although less accurate than equations (71 and (8),
= 0.215 according to Lee e t a / . -~7 The correction factor still reasonable.
given by equations (7) and (8) becomes discontinuous at Study on the KK-connections under antisymmetrical
the border. Correction factors when D/T = 20. 40 and 80 loads is now in progress-'"~°. Data obtained so far shows
are plotted against g/D in Figure 8. Numerical results, that the KK- to K-connection capacity ratios become about
classified into four groups according to D/T and the failure 0.85 over a wide range of gJD ratio. This shows that the
type, arc compared with the correction factors as shown in Cidect correction factor of 0.9 is also roughly applicable to
I"igure 8. Test results are not included here because they KK-connections under antisymmetrical axial loads.
scatter more than numerical results, although both have the
same mean values. Fi,~ure 8 shows how abruptly the
7. Conclusions
capacity changes when the failure mode varies from type
1 to type 2. Both the A W S and IIW ultimate strength equations for non-
Multiple regression analyses on both numerical and test overlapping K-connections have been compared with all
Design rules for circular tubular K-connections: Y. Kurobane and K. Ochi 265

2.0
terns interactions between connection behaviour and frame
behaviour. This subject affords a basis for establishing
1.8 sound design rules for connections and was dealt with in
some detail in recent studies -~1-32. The authors proposed
three different design criteria for tubular connections: the
~ 1.6
strength design against static loads; and strength and duc-
tility design against dynamic loads like earthquake loads.
~ 1.4 Additional safeguards were required to prevent crack
r - growth along the weld toes in connections or a sequential
failure of connections tbllowing buckling of members to
~ 1.2
ego provide for the effects of dynamic loads. A similar assess-
ment is also mandatory with regard to the AWS design
~ 1.0 I |El rules.
< El
tu
ii 0.8 Acknowledgments
or This paper is based on work partly supported by the
O Japanese Ministry of Education under Grant B05452256
~- 0.6
09 and by Cidect under Research Project 5BF. The authors
I.IJ
I-- wish to thank Professor J. A. Packer, University of Toronto
0.4 t~ Eqs.7 & 8
for valuable comments on this manuscript.
• AWS
0.2 ~ 1/0.74 References
.... l I "Structural welding codeAteel', ANSI/AWS D I.I, Arnerican Weld-
ing Society. Miami. F[,. 1996
0.0 2 "Design recommendations for hollo~ section joints predominantly
0 20 40 60 80 statically loaded (2nd edn)'. IIW D~¢'. XV-701-89'. International
D/T Institute of Welding. Cambridge, UK, 1989
3 Kurobane. Y.. Makino. Y. and Ochi, K. "Uhimate resistance of
Figure9 Test-to-prediction ratios f o r KK-connections accord- unstiffcned tubular joints'. J. Struct. Engng. AS('E 1984. I t 0 (21.
ing to AWS and p r o p o s e d f o r m u l a e 385-4(X)
4 Marshall, P. W. lh,sign of welded tubular connection,~: hasi,~ and use
~?iA WS code provisions. Elsevier, Amsterdam. The Netherlands, 1992
the existing reliable test results. The AWS equations show
5 "Eurocodc 3: l)csign of steel structures', ENV 1993-1-1, C6mitc Eur-
errors signiticantly greater than the ilW equations, suggest- op6an dc Normalisation. British Standards Institution. I,ondon. UK.
ing the presence of systematic errors in the former equa- 1992
tions. These errors can bring about relatively unsafe design 6 t"ackcr, J. A. and Henderson. J. I'. "Design guide for hollo~, structural
section connections'. Canadian Institute of Steel Construction. 1992
lbr connections with stocky chords and very conservative
7 "Recommendationx fi)r the design and fitbrication of tubuhtr ~truc-
design for connections with thin-walled chords and small tttres in steel'. Architectural Institute of Japan, Tokyo, Japan. 1990
gaps. Systematic errors were even more pronounced in KK- (in Japanese )
connections than in K-connections. New simple capacity 8 Wardcnicr. J., Kurobanc. Y.. Packer, J. A.. Durra. D. and Yocmans,
prediction formulae for KK-conncctions are proposed, N. "Design guide Ior circular hollow section (CtIS) joints under pre-
dominantly static loading'. Cidect (Ed.) Vcrlag TIJV Rheinland
which show no systematic error. GmbH. K61n. Germany, 199 I
The AWS design equation (equations (1) and (2)) has 9 Ochi. K.. Makino, Y. and Kurobanc. Y. "Basis li)r design of unstiff-
a format so that the design strength is automatically calcu- cncd tubular joints under axial brace loading'. IIW Doc. XV-561-84.
lated for any nonoverlapping connection that has many International Institute of Welding, Cambridge, L!K, 1984
10 Kurobane, Y., Ogwa, K.. Ochi, K. and Makino. Y. "l,¢x:al buckling
members framing into it by using its unique chord ovalizing
of braces in tubular K-joints', Thin-Walled Struct. 1986. 4. 23-40
parameter, whereas, in the IIW format, engineering judg- II I)cxtcr, E. M.. l,cc. M. K. K. and Kirkwcx~. M. G. "Effect of overlap
ments have to be exercised when designing complex con- on strength of K-joints in CHS tubular members', in "l)tbular .~true-
nections other than those specified in the code. However, tures VI, P. Grundy, A. tlolgatc and G. B. Wong (Eds). A. A. Balk-
no attempt is made in this paper to enhance the reliability cma. Rotterdam, The Netherlands, 1994. pp. 581 5 8 8
12 Hcaly, B. E. "A numerical investigation into the capacity of over-
of the AWS criteria, while retaining the attractive AWS
lapped circular K-joints'. "lubular Structure~ VI. 1994. P. Grundy,
format. A. Holgatc and G. B. Wong (Eds). A. A. Balkema. Rotterdam. The
Low-strength failure events induced by failure mcxtes Netherlands. 1994, pp. 563-571
other than chord wall plastification were found to be hidden 13 Yura. J. A. and Frank. K. It. "Uhimatc load tests on tubular connce-
behind the scatter of data when test results were compared lions', CESRE Report. 78-I, University of Texas, Austin, TX, 1974
14 Ohtakc, F., Sakamoto, S., Tanaka, T., Kai, T., Nakazato. T. and Taki-
with the AWS equations. This fact demonstrates that zawa, T. 'Static and fatigue strength of high tensile strength steel
strength governing factors should vary with the failure tubular joints for offshore structures', Prof. Offshore Technology
mode. Additional safeguards which are required of the Conf. 1978, Houston, TX. pp. 1747-1755
AWS design rules are those against local buckling of com- 15 Kurobane. Y., Makino, Y. and Ogav, a, K. "Further ultimate limit state
criteria for design of tubular K-joints', "lubuhtr Structures. 1989. E.
pression braces in areas adjacent to connections and also
Nicmi and P. M~ikclainen (Eds). Elsevier, London. UK, pp. 65-72
against tensile failure from cracks at weld toes. Since, how- 16 'Recommended practice .fi~r phmning, designing alld constructing
ever, the latter failure mode is difficult to analyse, no defi- fixed offshore platfi~rm.s '. 16th edn. API PR2A, American Petroleum
nite criterion has yet been established for either the AWS Institute, I')allas. TX. 1986
17 Machida. S., Hagiwara, Y. and Kaiimoto, K. 'Evaluation of brittle
or the IIW design rules. This is an issue requiring intensive
fracture strength of tubular joints of offshore structures', Proc. 6th
study in the future. Int. Off~hore Mechanics and Amtic Engineering Syrup.. 1987, Vol.
Another important issue for the connection design con- 3, ASME, New York, pp. 231-237
266 Design rules for circular tubular K-connections: Y. Kurobane and K. Ochi

18 Packer. J. A. "Overview of current international design guidance on 26 Wilmshurst, S. R. 'The ultimate load behaviour of multiplanar KK-
hollow structural section connection', Proc. 3rd Int. Offshore and joints subjected to symmetrical and anti-symmetrical axial loading',
Polar Engng. Conf, 1993, Singapore, Vol. IV, pp. I-7 PhD thesis. University of Wales, Swansea. UK. 1995
19 Makino, Y.. Kurobane, Y. and Ochi, K. 'Ultimate capacity of tubular 27 Lee, M. M. K. and Wilmshurst, S. R. "A parametric study of strength
double K-joints', Proc. Int. Conf IIW on Welding of Tubular Struc- of tubular multiplanar KK-joints'. J. Stru('t. Engtlg. AS('E 1996, 122
tures, 1984, Pergamon, New York, NY. pp. 451-458 ( 8 ), 893-904
20 Paul, J. C.. van der Valk. C. A. C. and Wardenier. J. 'The static 28 Kurobane, Y.. Makino, Y. and Ochi, K. 'Analysis of existing and
strength of circular multiplanar X-joints', Tubular Structure,~, 1989, forthcoming data for multi-planar KK-joints with circular hollow sec-
E. Niemi and P. M~ikel~iinen (FAs), 1989. Elsevier. l,ondon, UK, tions', Cidect Report 5BF-11-96. 1996
pp. 73-80 29 Makino. Y.. Kurobane. Y., Wilrnshurst. S. R. and Lcc, M. M. K.
21 Scola, S., Redwood. R. G. and Mitri, H. S. 'Behaviour of axially "Proposed ultimate capacity equations lk~rCHS KK-joints under anti-
loaded tubular V-joints', J. Const. Steel Res. 1990, 16, 89-109 ,.ymrnetrical loads'. Prm'. 5th Int. OJf~hore and Polar L)~gineering
22 Paul. J. C. 'The ultimate behavior of multiplanar TT and KK-joints ('otiS, 1995, The Hague. The Netherlands, Vol. 4, pp. 6 - I I
made of circular hollow sections'. PhD thesis, 1992. Kumamoto Uni- 30 l,ee. M. M. K. and Wilmshurst. S. R. 'Strength of multiplanar tubular
versity. Kumamoto. Japan KK-joints under anti-syrnmetrical axial loading', "l~bular Structure,~
23 Paul. J. C., Makino. Y. and Kurobane. Y. "Ultimate resistance of VII. 1996, J. Farkas and K. Jarmai (Eds). A. A. Balkema, Rotterdam.
unstiffened multiplanar tubular TT- and KK-joints'. J. Struct. Engng. The Netherlands, pp. 149-156
ASCE 1994, 121) (10), 2853-2870 31 Kurobanc. Y. and Ogawa. K. 'New criteria for ductility design of
24 van der Vegte. J. G. "The static strength of uniplanar and multiplanar joints based on complete CHS truss tests', l'uhular Structures V,
tubular T- and X-joints', PhD thesis, Delft University. Delft. The 1~,.~13. Coutie and Davies (Eds). E & FN Spon. l,ondon. UK,
Netherlands, 1995 pp. 570-58 I
25 van der Vegte. G. J. and Wardenier, J. 'The static behaviour of multi- 32 Kurobane. Y. and Ogawa, K. 'Earthquake resistant design provisions
planar tubular steel "IT-joints excluding the effect of chord bending', lor tubular structures', Int. Conf on Tuhular Structures. 1996, Amer-
Int. J. Offshore Polar Engng 1995, 5 (4), 308-316 ican Welding Society. Miami, FI.. pp. 74-85

Das könnte Ihnen auch gefallen