Experimental and Numerical Investigation of Geometric SCFs...
Transcript of Experimental and Numerical Investigation of Geometric SCFs...
1
* E-mail: [email protected]
Experimental and Numerical Investigation of Geometric SCFs in Internally Ring-Stiffened
Tubular KT-Joints of Offshore Structures
Ahmadi, Hamid*; Lotfollahi-Yaghin, Mohammad Ali
Faculty of Civil Engineering, University of Tabriz, Tabriz, IR Iran
Received: July 2012 Accepted: November 2012
© 2013 Journal of the Persian Gulf. All rights reserved.
Abstract Tubular KT-joints are quite common in offshore structural design and despite the crucial role of stress
concentration factors (SCFs) in evaluating the fatigue performance of tubular joints. However, the SCF
distribution in internally ring-stiffened KT-joints have not been investigated and no design equation is
currently available to determine the SCFs for this type of joint. In the present paper, results of
experimental and numerical investigations of the SCF distribution in internally ring-stiffened tubular KT-
joints are presented. In this research, experimental study has been followed by a set of parametric stress
analyses for 118 steel ring-stiffened KT-joints subjected to balanced axial loads. The analysis results are
used to present the general remarks on the effect of geometrical parameters on the SCF distribution along
the weld toe and to establish a new set of SCF parametric equations for the fatigue design of internally
ring-stiffened KT-joints.
Keywords: Fatigue, Stress concentration factor (SCF), Tubular KT-joint, Internal ring-stiffener, Parametric
design equation
1. Introduction
Offshore jacket-type platforms are mainly
fabricated from tubular members by welding one end
of the branch member (brace) to the undisturbed
surface of the main member (chord), resulting in
what are known as tubular joints. The static and
fatigue strengths of tubular joints are the governing
factors in the design of offshore steel structures.
Significant stress concentrations at the vicinity of
the welds are considerably detrimental to the fatigue
performance of the joints. Since, offshore structures
are subjected to cyclic wave loading, it is imperative
that these structures be designed for long fatigue life.
Hence, it is important to accurately determine the
magnitude of stress concentration and to reduce it to
a reasonable level. In the design practice, a parameter
called the stress concentration factor (SCF) is used to
evaluate the magnitude of the stress concentration.
The SCF is the ratio of the local surface stress to the
nominal direct stress in the brace and its value
depends on the joint geometry, loading type, weld size
and type and the considered location around the weld.
If the capacity of a joint is found to be inadequate
during the design stage proper method for its
Journal of the Persian Gulf
(Marine Science)/Vol. 4/No. 12/June 2013/12/1-12
cor
thi
of
stif
eff
inc
joi
avo
cor
int
joi
un
stu
beh
joi
we
Fig
rrection mus
ickness requ
fabricators,
ffeners to th
ficient meth
crease load-c
int, decrease
oid attractio
rrosion attac
ternally ring-
The value o
int is mainly
der any spe
udy the behav
havior easily
int, a set of
ere defined.
g. 1: Geometrica
st be applied.
uirement is b
it can be
he inside of
hod to redu
carrying cap
the bending
on of addi
ck. These ty
-stiffened joi
of SCF along
y determine
ecific loadin
vior of tubul
y to the geo
dimensionle
Figure 1 s
al notation for a
Ahmadi et al
. For exampl
beyond the f
enhanced b
f the chord
uce stress
pacity and f
g stress in tu
itional wave
ypes of join
ints (Fig. 1).
g the weld to
ed by the jo
ng condition
lar joints and
ometrical pro
ess geometric
shows an in
an internally rin
l. / Experimen
le, if the cho
forming limi
by introducin
as this is a
concentratio
fatigue life
ube walls an
e forces an
nts are calle
oe of a tubul
oint geomet
n. In order
d to relate th
operties of th
cal paramete
nternally rin
ng-stiffened KT
ntal and Nume
2
ord
its
ng
an
on,
of
nd
nd
ed
lar
try
to
his
he
ers
ng-
stiffe
param
brace
wall
A
in pr
evalu
stres
joint
equa
for t
expe
SCF
KT-j
study
analy
joint
T-joint under ba
erical Investig
ened tubula
meters (τ, γ,
e diameters
thicknesses
Although, the
ractice and
uating the fa
ss distributio
ts has not be
ation is curre
his type of j
erimental an
distribution
joints are pre
y was follow
yses for 11
ts subject to b
alanced axial loa
gation of Geom
ar KT-joint
η, β, ζ, α,
D and d, a
T and t.
tubular KT-j
despite the
atigue perfor
on in intern
en investigat
ntly availabl
oint. In the
nd numerica
n in internall
esented. In th
wed by a s
8 steel ring
balanced axi
ads
metric SCFs in
with the
and αB) fo
and their co
-joints are qu
crucial role
rmance of tu
nally ring-sti
ated so far an
le to determi
present pape
al investigat
ly ring-stiffe
his research,
set of param
g-stiffened t
ial loading.
n Internally…
geometrical
or chord and
orresponding
uite common
of SCFs in
ubular joints,
iffened KT-
nd no design
ine the SCFs
er, results of
tions of the
ened tubular
experimental
metric stress
tubular KT-
l
d
g
n
n
,
-
n
s
f
e
r
l
s
-
Journal of the Persian Gulf (Marine Science)/Vol .4/No. 12/June 2013/12/1-12
3
The analysis results were used to present general
remarks on the effect of geometrical parameters
including τ (brace-to-chord thickness ratio), γ (chord
wall slenderness ratio), η (ring-to-chord width ratio), β
(brace-to-chord diameter ratio), and θ (outer brace
inclination angle) on the SCF distribution along the
weld toe. Since the vertical (central) brace SCFs are
generally much bigger than the inclined (outer) brace
SCFs, the present study was focused only on the SCF
distribution along the weld toe of the intersection
between the central brace and the chord. Based on the
results of ring-stiffened KT-joint finite element (FE)
models which were verified against experimental
measurements, a comprehensive SCF database was
constructed and a new set of SCF parametric equations
was established, through nonlinear regression analysis,
for the fatigue design of internally ring-stiffened KT-
joints. An assessment study of these equations was
conducted based on the acceptance criteria proposed by
the UK DoE (1983). Significant effort has been devoted over the past
four decades to the study of SCFs in various unstiffened joints. Nevertheless, very few investigations have been reported for stiffened joints due mainly to the geometrical complexity and the vast variety of possible stiffening arrangements.
Dharmavasan and Aaghaakouchak (1988) showed
that in the case of axial and Out-of-Plane Bending (OPB)
loadings, careful positioning of the stiffeners greatly
reduced the stress concentrations and gave a more
uniform stress distribution around the intersection.
Various stiffener sizes and configurations were analyzed
and recommendations made on the optimum design.
Aaghaakouchak and Dharmavasan (1990) presented an
improved FE technique consisting of 3-D brick and
semi-loof shell elements. Ramachandra et al. (1992)
investigated the effect of geometrical parameters on the
SCFs in ring-stiffened tubular T- and Y-joints. They
proposed a set of parametric equations for the prediction
of maximum SCFs under axial, In-Plane Bendeing
(IPB), and OPB loadings. Using degenerate shell
elements, Nwosu et al. (1995) studied the stress
distribution along the intersection of internally ring-
stiffened tubular T-joints, under the action of axial,
IPB, and OPB loads. The effects of stiffener size,
location and number have been investigated in this
study. Ramachandra et al. (2000) investigated the effect
of internal ring stiffeners on the fatigue strength of
tubular T- and Y-joints. Hoon et al. (2001) studied the
SCF distributions along the intersections of a doubler-
plate reinforced T-joint subjected to combined
loadings. Myers et al. (2001) investigated the effects of
three different longitudinal stiffeners on the stress
concentration factors in jack-up chords. The constant
thickness continuous stiffener returned the best
performance when compared to the dual thickness
stiffener (rack/rib plate) and the non-continuous
stiffener. Woghiren and Brennan (2009) proposed a set
of parametric equations to predict the SCFs in multi-
planar tubular KK-joints stiffened by rack plates.
2. Experimental Investigation of the SCF Distribution in Ring-Stiffened KT-Joints
2.1. Details of Specimens
Two steel KT-joint specimens were fabricated for
the purpose of experimental tests in order to
investigate the effect of internal ring-stiffeners on the
SCF distribution along the weld toe of tubular KT-
joints. Fabricated specimens were of the same size
with identical geometrical characteristics. The only
difference between the two specimens was the
presence of internal ring-stiffeners in the second
specimen. Geometrical details of the internally ring-
stiffened specimen are presented in Figure 2. As
shown in this figure, three ring-stiffeners were used
in the present study for each brace. One of them was
placed at the saddle position and the other two were
welded at the crown positions.
2.2. Test Setup and Loading Scheme
The two fabricated specimens were tested using a
hydraulically controlled test rig. Both ends of the
chord and upper ends of the outer (inclined) braces
we
two
loa
bra
Fig
Fig
ere welded on
o standing su
ading was ap
ace. In ord
g. 2: Geometrica
g. 3: Test setup
nto the flat p
upports as sh
pplied at the e
der to guar
al details of the
Table 1:
N
A
Ahmadi et al
plates and bo
hown in Figur
end of the ce
rantee the
e internally ring
: Axial loads ap
Nominal stress,
Applied load, F
l. / Experimen
lted directly
re 3. The stat
entral (vertica
accuracy an
-stiffened KT-j
pplied on the ce
σn (MPa) ±5
Fa (kN) 11
ntal and Nume
4
to
tic
al)
nd
reliab
were
for e
were
oint specimen
entral brace in s
5 ±10 ±
.95 23.9 3
erical Investig
bility of rec
e chosen for
each applicat
e recorded 10
tiffened and un
±15 ±20 ±
5.85 47.8 5
gation of Geom
corded data,
the applied
tion, output
0 times in 5 se
nstiffened specim
±25 ±30
59.75 71.7
metric SCFs in
6 different
axial load (T
data from s
ec intervals.
mens
n Internally…
magnitudes
Table 1) and
strain gauges
s
d
s
2.3. Strain
Procedure
In the pre
around the
intervals. A
were used
components
arrangemen
linear extrap
XV-E (199
distance of
point lies a
4b). The ho
extrapolatio
gauges to th
two crowns
gauges wer
along the d
measure the
extrapolatio
of the strain
extrapolatio
results to
sufficiently
avoid avera
of steep gra
After rec
strain conce
using the foSNCF =
Where from the m
are glued on
midpoint.
For the f
SCFs were
SCF = SNC
Where along the w
SCF/SNCF
Gauge Loc
esent study,
entire brac
As shown in
at each loc
s perpendicu
nt of the two
polation proc
99). The firs
0.4T from t
at 1.0T furth
ot-spot strain
on of the re
he weld toe.
s and two sa
re glued on
direction pa
e parallel co
on points. Ag
n at the wel
on of these
the weld t
small strain
aging high a
adients.
cording the
entration fact
ollowing equ
/ n
n is the no
measurement
nto the surfa
four typical
calculated us
CF ×⁄
is Poisson
weld toe, SC
ratio which
cations and
the strain ga
ce/chord int
Figure 4a, t
cation to m
ular to the
o strain gau
cedure recom
st extrapolat
the weld toe
her from the
n ( ) was o
esults from
At four typi
addles, anoth
nto the surfa
arallel with
omponents of
gain, parallel
ld toe was o
e two paral
toe. It mus
n gauges sh
and low strai
data from s
tors (SNCFs
ation:
ominal strain
of four stra
ace of the lo
positions alo
sing the follo
n’s ratio. Fo
CFs were d
h was obtain
Journal
SCF Extra
auges were p
tersection at
two strain ga
easure the s
weld toe.
uges followe
mmended by
ion point is
e, and the se
e first point
obtained by l
these two s
ical position
her pair of s
ace of the c
the weld to
f the strain a
l component
obtained by l
llel strain g
t be noted
hould be use
ins in the re
strain gauges
s) were calcu
n that is obt
ain gauges w
oaded brace a
ong the weld
owing equati
or other posi
derived using
ned by avera
of the Persian
5
action
placed
t 30˚
auges
strain
The
d the
IIW-
s at a
econd
(Fig.
linear
strain
ns, i.e.
strain
chord
oe to
at the
t ( n )
linear
gauge
that
ed to
gions
s, the
ulated
(1)
tained
which
at the
d toe,
ion:
(2)
itions
g the
aging
th
th
fo
F
2
w
st
ex
co
g
w
w
cr
co
n Gulf (Marin
he SCF/SNC
he present st
or unstiffene
Fig. 4: (a) Plan vi(b) Extrapolatio
.4. Results a
Figure 5 il
weld toe on t
tiffened KT
xperimental
ompared. D
eometry and
weld toe is s
whole 360˚
rown and
onsidered. A
ne Science)/Vo
CF ratios at th
tudy, average
d and stiffen
iew of the strain on procedure rec
and Discussi
lustrates the
he chord sur
T-joint spe
data and
Due to the
d loading, the
symmetric a
brace/chord
saddle posi
As shown in
ol .4/No. 12/Ju
he four typic
e ratios were
ned specimen
n gauge locationscommended by I
ion
e SCF distrib
rface of the
ecimens. In
FE resul
e symmetry
e SCF distrib
and only one
intersection
itions, is r
n Figure 5, t
une 2013/12/1
cal positions
e 1.20 and 1
ns, respective
s along the weld IIW-XV-E (199
bution along
unstiffened a
n this figu
ts have b
in the jo
bution along
e fourth of
n, between
required to
there is a go
1-12
. In
1.25
ely.
d toe, 99)
the
and
ure,
een
oint
the
the
the
be
ood
Ahmadi et al. / Experimental and Numerical Investigation of Geometric SCFs in Internally…
6
agreement between the test results and FE predictions
in both unstiffened and stiffened joints (Fig. 5).
Hence, generated FE models could be considered to
be accurate enough to produce valid results. Details of
FE analysis are discussed in Section 4. It can be
clearly concluded from comparing the SCF
distributions in stiffened and unstiffened joints that,
internal ring-stiffeners can effectively reduce the
stress concentration along the brace/chord intersection
and consequently, improve the fatigue performance
of the tubular joint. It can also be seen that the
presence of stiffeners have resulted in the change of
peak SCF location from the saddle position to φ =
60˚. This result means that the position from which
the fatigue cracks initiate could be different in
stiffened and unstiffened KT-joints subjected to
axial loading. As depicted in Figure 5, the SCF
distribution along the weld toe of the stiffened joint
is much more uniform than the distribution in the
unstiffened joint. This observation is a result of
using the stiffeners at both saddle and crown
positions and means that the applied load on the
brace transfers to the chord in nearly equal
proportions through the intersection. The difference
between the SCFs in stiffened and unstiffened joints
is larger at locations adjacent to the saddle position.
This result implies that the stiffener located at the
saddle position is more effective than the stiffeners
welded at the crown positions in reducing the SCFs
along the weld toe.
Fig. 5: SCF distribution along the weld toe on the chord surface of the unstiffened and internally ring-stiffened KT-joint: comparison of the experimental data and FE results
3. SCF Numerical Analysis of Ring-Stiffened KT-Joints
3.1. Important Modeling Considerations
In the present study, the multi-purpose FEM-
based software package, ANSYS was used for the
numerical modeling and analysis of internally ring-
stiffened tubular KT-joints, subjected to balanced
axial loading (Fig. 1a), in order to investigate the
SCF distribution along the weld toe. During the FE
modeling of an internally ring-stiffened tubular joint,
there were two subjects that needed careful attention:
(a) accurate simulation of the weld profile and (b)
correct modeling of the interaction between the
chord and the stiffener, when direct merging should
be avoided in order to provide high-quality mesh
along the brace/chord intersection.
In this study, the welding size along the
brace/chord intersection satisfied the AWS D 1.1
specifications (2002). Modeling of the weld profile
according to AWS D 1.1 (2002) has been extensively
discussed by Lotfollahi-Yaghin and Ahmadi (2010).
The chord and stiffener were meshed separately
and then the ANSYS contact capability was used to
define the interaction between them.
Due to the XY-plane symmetry in the joint geometry
and loading, only half of the entire ring-stiffened KT-
joint is required to be modeled (Fig. 6a).
3.2. Mesh Generation
In the present study, ANSYS element type
SOLID95 was used to model the chord, brace,
stiffener, and the weld profile. The element is
defined by 20 nodes having three degrees of
freedom per node and may have any spatial
orientation. In order to guarantee the mesh quality,
a sub-zone mesh generation method was used
during the FE modeling. The mesh generated by
this method for an internally ring-stiffened tubular
KT-joint is shown in Figure 6.
1
2
3
4
5
6
7
0 10 20 30 40 50 60 70 80 90
Polar Angle φ (measure from crown position)
Str
ess
Con
cent
rati
on F
acto
r
Unstiffened Joint (EXP)Unstiffened Joint (FEA)Internally Ring-Stiffened Joint (EXP)Internally Ring-Stiffened Joint (FEA)
Fig. 6: The me
4. Effects oDistributio
4.1. Parame
In order
internally
models wer
purpose FE
objective w
geometrical
along the w
subjected t
Different v
parameter w
0.4, 0.5, 0.6
45˚, 60˚. Th
the normali
esh generated fo
of Geometricon
etric Study
to investiga
ring-stiffene
re generated
EM-based sof
was to study t
l parameter
weld toe of t
to balanced
values assig
were as follo
6; γ = 12, 18,
hese values
ized parame
for a ring-stiffen
cal Paramet
ate the stress
ed tubular
and analyzed
ftware packa
the effect of
s on the S
the central b
d axial loa
gned to eac
ows: η = 0.
, 24; τ = 0.4,
cover the pr
eters typicall
Journal
ned tubular KT-
ters on the S
s concentrati
KT-joints,
d using the m
age, ANSYS
f non-dimens
SCF distrib
brace in the j
ading (Fig.
ch dimensio
1, 0.15, 0.2
0.7, 1.0; θ =
ractical rang
ly found in
of the Persian
7
-joint using the
SCF
on in
118
multi-
S. The
sional
bution
joints
1a).
onless
; β =
= 30˚,
ges of
ring-
st
st
b
th
ra
0
0
1
0
3
4
stco3
n Gulf (Marin
sub-zone meth
tiffened tub
tructures. Th
race in all m
he other two
The 118 ge
anges of the
.1 ≤ η ≤ 0.2
.4 ≤ β ≤ 0.6
2 ≤ γ ≤ 24
.4 ≤ τ ≤ 1.0
0˚ ≤ θ ≤ 60˚
.2. Effect of
The SCF dtiffened andompared in F60˚ curve o
ne Science)/Vo
od
bular joints
hree ring-stif
models: One
at the crown
enerated mo
geometric pa
internal ring
distribution d unstiffeneFigure 7a. Thf the weld p
ol .4/No. 12/Ju
of offsho
ffeners were
e at the sadd
n positions.
odels spanned
arameters:
g-stiffeners o
along the wed tubular he polar angpath is mea
une 2013/12/1
ore jacket-ty
e used for e
dle position a
d the follow
on SCFs
weld toe in tKT-joints
gle (φ) along asured from
1-12
ype
ach
and
wing
(3)
two are the the
crocroobis stifalsstifThinvthecrounat stifdesjoistifanddevstifdifuntheagr
4.3
We
thimolea
at of at tBu0.7pea
4.4
We
cho
own positionowns and 9served that ilocated at
ffened joint so clear thatffened and u
hese two obvestigating the SCFs at typown. As shstiffened KTthe same
ffened joint.sign equatioints were uffened jointsd highly convelop SCF fffened tubulafference betstiffened joine saddle preement with
3. Effect of th
eld Toe
Since the pickness of thodels havingads to the decThe increaseall positionsthe τ (say τ =the saddle po
ut for interme7, 1.0), changak SCF posit
4. Effect of th
eld Toe
The parameord thicknes
n. Hence, val90˚/270˚ at in the unstif
the saddleit is located t the shapesunstiffened j
bservations hhe SCF distripical positio
hown in FigT-joint is 4.5 position in This result
ons developesed for thes, obtained nservative. Hformulae spar KT-joints.tween the nts was largposition. Thh the experim
he γ on the S
parameter γ he chord, thg fixed valucrease of thee of the γ leas along the w= 0.4), the position regarediate and bge of the γ ction.
he τ on the S
ter τ is the rss and the γ
Ahmadi et al
lues of φ arethe saddles
ffened joint, e position, at the crowns of SCF dijoints are q
highlight theibution inste
ons, such as tgure 7b, thetimes biggerthe corresp
indicated thed for the une SCF calcu
SCFs ould Hence, it waecially desig It can also bSCFs in
ger at locatiohis observa
mental results
SCF Distribu
is the ratioe increase o
ue of the ch chord thicknads to an incweld toe. Fopeak SCF is ardless of the big values of can result in
SCF Distribut
ratio of bracγ is the ratio
l. / Experimen
0˚/180˚ at th. It could bthe peak SCwhile in th
n position. It istributions
quite differene necessity ad of studyinthe saddle ane SCF in ar than the SCponding rin
hat, if the SCnstiffened KTulation in thbe unrealists necessary gned for rinbe seen that thstiffened an
ons adjacent ation was
(Section 3.4
ution along th
o of radius of the γ in thhord diametness. crease in SCFr small valualways locatevalue of the
f the τ (say τdisplacing th
tion along th
ce thickness o of radius
ntal and Nume
8
he be
CF he is in
nt. of ng nd an
CF ng-CF T-he tic to
ng-he nd to in
4).
he
to he ter
Fs ues ed γ. = he
he
to to
thickthe mincre
That allthe iτ is geomτ connot hSCF
Fig. 7and sproperSCF positio
4.5. E
Weld
R
SCF
the p
erical Investig
kness of the cmodels havinease of the brhe increase ol positions ancrease in Shighly rema
metrical paransiderably afhave remarkdistribution
7: (a) SCF distristiffened tubularties (η = 0.15, βin the consideron in the corresp
Effect of the
d Toe
esults of inv
distribution
present secti
gation of Geom
chord. Henceng fixed valrace thicknesof the τ leadlong the welCF values drkable in cometers. Althffects the makable influen
along the we
ibutions along thar KT-joints hβ = 0.4, γ = 12, τred unstiffened ponding stiffened
e η on the SC
vestigating th
along the w
on. In this s
metric SCFs in
e, the increaslue of the γ ss.
ds to an increld toe. The m
due to the incomparison whough, the incagnitude of Snce on the s
weld toe.
the weld toe in having the samτ = 1.0, θ = 45˚),joint to the SCd joint
CF Distributio
he effect of
weld toe are
study, the in
n Internally…
se of the τ inleads to the
ease in SCFsmagnitude ofcrease of theith the othercrease of the
SCFs, it doesshape of the
two unstiffenedme geometrical, (b) The ratio ofCF at the same
on along the
the η on the
discussed in
nteraction of
n e
s f e r e s e
d l f e
e
n
f
the η with t
investigated
stiffener wi
providing t
remains unc
increase of
presented in
distribution
analyzed m
the η and i
Figure 8 s
different va
the paramet
presented c
0.4, 0.7, and
For smal
SCF is al
regardless o
η, from 0.1
positions alo
from value
Although th
Fig. 8: Effects o
the other geo
d. The para
dth (height)
that the va
changed, the
f the stiffen
n Figure 8,
of SCFs alo
models. These
its interactio
shows the
alues of η (0
ters β, γ, an
harts and th
d 1.0, respec
ll values of
lways locat
of the value
to 0.2, lead
ong the weld
es of the o
he increase of
of the η on the S
ometrical par
meter η is
to the chord
alue of the
e increase of
ner width. T
as an exam
ong the weld
e charts illus
on with the
SCF distrib
0.1, 0.15, an
nd θ are iden
e value of τ
tively.
the τ (say τ
ted at the
of the η. Th
s to a decrea
toe. This res
other geomet
f the η affects
SCF distribution
Journal
rameters was
the ratio o
diameter. H
chord diam
the η leads t
Three charts
mple, showin
d toe for the
trate the effe
τ. Each cha
butions for
d 0.2). Valu
ntical in all
in Figure 8a
τ = 0.4), the
saddle pos
he increase o
ase in SCFs
sult is indepe
trical param
s the magnitu
along the weld
of the Persian
9
s also
of the
Hence,
meter
to the
s are
ng the
e 9 of
ect of
art in
three
ues of
three
a-c is
peak
sition
of the
at all
endent
meters.
ude of
S
th
4
W
dinle3osa
thtoththredanSjo
toe and its intera
n Gulf (Marin
CFs, it usual
he shape of th
.6. Effect of
Weld Toe
Since the iameter to thn a model weads to an in6 comparativf the β. Theake of brevit
The increahe increase ooe. This conche other geomhe γ (say γemarkable iistribution and big valueCF distribu
oints having
action with the τ
ne Science)/Vo
ly does not h
he SCF distrib
the β on the
parameter βhe chord diamwith fixed vancrease in theve charts wese charts arety. se of the β, of SCFs at aclusion was metrical para= 12), incre
influence onlong the wels of the γ (sations could different val
τ
ol .4/No. 12/Ju
have remarka
bution along
SCF Distrib
β is the ratiameter, the inalue of the e brace diamere used to se not present
from 0.4 to all positions independent
ameters. For ease of the βn the shapeld toe. But fay γ = 18, 24
be quite dlues of the β.
une 2013/12/1
able influence
the weld toe.
bution along
io of the brncrease of thchord diame
meter. A totalstudy the effted here for
0.6, resultedalong the w
t from valuessmall valuesβ did not he of the Sfor intermedi4), the shapesdifferent in .
1-12
e on
the
race he β eter l of fect the
d in weld s of s of ave
SCF iate s of the
Ahmadi et al. / Experimental and Numerical Investigation of Geometric SCFs in Internally…
10
4.7. Effect of the θ on the SCF Distribution along the
Weld Toe
It can be concluded by investigating the effect of
the outer brace inclination angle (θ) on the SCF
distribution that, as it was expected, this parameter did
not have remarkable influence on either the magnitude
or the distribution pattern of SCFs along the weld toe
of the central brace. This conclusion was independent
from values of the other geometrical parameters.
5. Parametric Design Equations
After performing a large number of nonlinear
regression analyses by the statistical software
package SPSS, following geometrically parametric
equations are proposed for predicting the SCF
distribution along the weld toe and the peak weld-toe
SCF in internally ring-stiffened tubular KT-joints
subjected to balanced axial loading:
SCF distribution along the weld toe:
(4)
where functions ξ1 – ξ6 are expressed as follows:
1( , ) 0.198 0.009 0.310 0.920 (5.1)
(5.2)
(5.3)
(5.4)
(5.5)
(5.6)
Peak SCF along the weld toe:
(6)
In Eqs. (4) and (6), R2 denotes the coefficient of
correlation and its value is considered to be
acceptable for both equations regarding the complex
nature of the problem. The parameter φ (0˚ ≤ φ ≤
180˚) in Eq. (5.4) and the parameter θ in Eqs. (5.5)
and (6) should be inserted in radians. The validity
ranges of dimensionless parameters for the proposed
equations have been given in Eq. (3).
The UK Department of Energy (UK DoE) (1983)
recommended the following assessment criteria
regarding the applicability of the commonly used
SCF parametric equations (P/R stands for the ratio of
the predicted SCF from a given equation to the
recorded SCF from test or analysis):
For a given dataset, if % SCFs under-predicting
25%, i.e. [%P/R < 1.0] 25%, and if % SCFs
considerably under-predicting 5%, i.e. [%P/R <
0.8] 5%, then accept the equation. If, in
addition, the percentage SCFs considerably over-
predicting 50%, i.e. [%P/R > 1.5] 50%, then
the equation is to be regarded as generally
conservative.
If the acceptance criteria were nearly met i.e. 25%
< [%P/R < 1.0] 30%, and/or 5% < [%P/R < 0.8] 7.5%, then the equation was regarded as
borderline and engineering judgment must be used
to determine acceptance or rejection.
Otherwise reject the equation as it is too optimistic.
In view of the fact that for a mean fit equation,
there is always a large percentage of under-prediction,
the requirement for joint under-prediction, i.e. P/R <
1.0, can be completely removed in the assessment of
parametric equations (Bomel Consulting Engineers,
1994). Assessment results according to the UK DoE
criteria [1] are presented in Table 2.
Table 2: Results of equation assessment according to UK DoE (1983) acceptance criteria
Equation Conditions Decision
%P/R < 0.8 %P/R > 1.5
Eq. (4) 17.4% > 5 % 2.5% < 50 % OK. Revise
Eq. (6) 2.8% < 5 % OK 0% < 50 % OK. Accept
= 1 , + 2 , 3 4 5 6 , , , 2 = 0.915
2 , = 1.203 0.001 − 1.202 ∙ 1.146 1.426 − 61.095
3 = 1.416 0.450 + 1.784 ∙ 2.892 2 + 2.874 + 1.904 4 = −0.260 + 0.397 1.218 − 4.302 0.180 + 4.577
5 = 1 + 6.559 −0.303 + 2.402 +8.022 6 , , , = 1 + 0.841 14.616 − 1.493 −0.049 + 0.320 − 0.291
max = 0.304 0.768 0.994 0.244 −0.307 −0.060 1 − 0.954 − 0.096 + 2.447 + 0.025 ∙ 2 = 0.909
Journal of the Persian Gulf (Marine Science)/Vol .4/No. 12/June 2013/12/1-12
11
As can be seen in Table 2, Eq. (6) satisfied the UK
DoE criteria, but Eq. (4) required revision. In order
to revise the Eq. (4), SCF values obtained from this
equation were multiplied by a coefficient in such a
way that resulted SCF set would satisfy the UK DoE
acceptance criteria. This idea could be expressed as
follows:
Design factor = SCFDesign / SCFEq. (4) (7)
where values of SCFEq. (4) were calculated from
the proposed equation and the values of SCF Design
were expected to satisfy the UK DoE criteria.
Multiple comparative analyses were carried out to
determine the optimum value of design factor.
Following equations should be used for design
purposes:
SCFDesign = 1.32 × SCFEq. (4) (8)
SCFDesign = 1.00 × SCFEq. (6) (9)
6. Conclusions
Results of experimental and numerical investigations
of the SCF distribution in internally ring-stiffened
tubular KT-joints were presented in this paper.
Numerical results from analyzing 118 FE models were
used to present general remarks on the effect of
geometrical parameters on the SCF distribution along
the weld toe of the central brace and to establish
parametric SCF equations for the fatigue design. Main
conclusions are summarized as follows:
The presence of internal ring-stiffeners, depending
on the values of geometrical parameters, might lead to
the disposition of the peak SCF along the weld toe.
The shapes of SCF distributions in stiffened and
unstiffened joints could also be quite different. The
SCF distribution along the weld toe of a stiffened joint
was usually much more uniform than the distribution
in its corresponding unstiffened joint. This behavior
was a result of using the stiffeners at both saddle and
crown positions. These results highlighted the
necessity of investigating the SCF distribution instead
of studying the SCFs at typical positions such as the
saddle and crown. SCFs in an unstiffened KT-joint are
generally much bigger than the SCFs at the same
positions in the corresponding ring-stiffened joint.
This observation implied that the use of SCF design
equations which were developed for the unstiffened
KT-joints generally led to over-predicting and highly
conservative SCFs for stiffened joints. Hence, it was
necessary to develop SCF formulae specially designed
for ring-stiffened tubular KT-joints.
The increase in the γ and/or τ and also the increase
of the β led to the increase of SCFs at all positions
along the weld toe. The magnitude of the increase in
SCF values due to the increase of the τ was highly
remarkable in comparison with the other geometrical
parameters. For small values of the τ, the peak SCF
was always located at the saddle position regardless
of the value of the γ. But for intermediate and big
values of the τ, change of the γ could result in
displacing the peak SCF position. For small values of
the γ, increase of the β did not have remarkable
influence on the shape of the SCF distribution along
the weld toe. But for intermediate and big values of
the γ, the shapes of SCF distributions could be quite
different in the joints having different values of the β.
The increase of the η led to a decrease in SCFs at all
positions along the weld toe. The parameter θ did not
have remarkable influence on either the magnitude or
the distribution pattern of SCFs along the weld toe of
the central brace.
Relatively high coefficients of correlation and the
satisfaction of acceptance criteria recommended by
UK DoE guaranteed the accuracy of the proposed
equations. Hence, the developed formulae could
reliably be used for the fatigue design of offshore
jacket-type structures.
References
Aaghaakouchak, A.A. and Dharmavasan, S., 1990.
Stress analysis of unstiffened and stiffened tubular
joints using improved finite element model of
Ahmadi et al. / Experimental and Numerical Investigation of Geometric SCFs in Internally…
12
intersection. Proceedings of the 9th OMAE
Conference, Houston, USA, 321–328.
American Welding Society, 2002. Structural Welding
Code: AWS D 1.1, Miami, USA.
Bomel Consulting Engineers, 1994. Assessment of
SCF equations using shell/KSEPL finite element
data. C5970R02.01 REV C.
Dharmavasan, S. and Aaghaakouchak, A.A., 1988.
Stress concentrations in tubular joints stiffened by
internal ring stiffeners. Proceedings of the 7th
OMAE Conference, Houston, USA, 141–148.
Hoona, K.H., Wong, L.K. and Soh, A.K., 2001.
Experimental investigation of a doubler-plate
reinforced tubular T-joint subjected to combined
loadings. Journal of Constructional Steel Research.
57: 1015–1039.
IIW-XV-E, 1999. Recommended Fatigue Design
Procedure for Welded Hollow Section Joints. IIW
Docs, XV-1035-99/XIII-1804-99, International
Institute of Welding, France.
Lotfollahi-Yaghin, M.A. and Ahmadi, H., 2010.
Effect of geometrical parameters on SCF
distribution along the weld toe of tubular KT-
joints under balanced axial loads. International
Journal of Fatigue. 32: 703–719.
Myers, P.T., Brennan, F.P. and Dover, W.D., 2001.
The effect of rack/rib plate on the stress
concentration factors in jack-up chords. Marine
Structures. 14: 485–505.
N’Diaye, A., Hariri, S., Pluvinage, G. and Azari, Z.,
2007. Stress concentration factor analysis for
notched welded tubular T-joints. International
Journal of Fatigue. 29: 1554–1570.
Nwosu, D.I., Swamidas, A.S.J. and Munaswamy, K.,
1995. Numerical stress analysis of internal ring-
stiffened tubular T-joints. Journal of Offshore
Mechanics and Arctic Engineering. 117: 113–125.
Ramachandra, D.S., Gandhi, P., Raghava, G. and
Madhava Rao, A.G., 2000. Fatigue crack growth in
stiffened steel tubular joints in seawater environment.
Engineering Structures. 22: 1390–1401.
Ramachandra, D.S., Madhava Rao, A.G., Ghandi, P.
and Pant, P.K., 1992. Structural efficiency of
internally ring stiffened steel tubular joints.
Structural Engineering. 118: 3016–3035.
UK Department of Energy, 1983. Background Notes
to the Fatigue Guidance of Offshore Tubular
Joints. UK DOE, London, UK.
Woghiren, C.O. and Brennan, F.P., 2009. Weld toe
stress concentrations in multi planar stiffened
tubular KK joints. International Journal of Fatigue.
31: 164–172.
Ahmadi et al. / Experimental and Numerical Investigation of Geometric SCFs in Internally…
Journal of the Persian Gulf (Marine Science)/Vol .4/No. 12/June 2013/12/1-12
Journal of the Persian Gulf
(Marine Science)/Vol. 4/No. 12/June 2013/12/1-12