Professional Documents
Culture Documents
OTC 17236 The New API RP 2A, 22 Edition Tubular Joint Design Practice
OTC 17236 The New API RP 2A, 22 Edition Tubular Joint Design Practice
OTC 17236 The New API RP 2A, 22 Edition Tubular Joint Design Practice
, Q
g
,
and Q
f
parameters have been fully revised as outlined below.
Validity Range. As before, the new formulation is based on
an interpretation of data, both experimental and FEA.
Therefore, a validity range has been imposed, although its
implication in general is minimal since the range covers the
wide spectrum of geometries currently used in practice. J oint
designs outside the below parametric ranges are permitted, but
require special investigation of design and welding issues.
0.2 = d/D 1.0
10 = D/2t 50
30 90
F
y
72 ksi (500 MPa)
g/D > -0.6 (for K joints)
Basic Capacity Equations. The general form of the brace
load and moment based capacity equations have been
maintained. However, the 0.8d multiplier in Edition 21s
4.3.1-4b formulation has been eliminated and absorbed into
the Q
u
coefficient. The 1.7 factor has been replaced with FS
enabling transportability between the WSD design approach
commonly used by API and the LRFD approach proposed in
the ISO 19902 Code, as follows:
P
a
=
sin
2
FS
T F
Q Q
yc
f u
API Eq. (4.3-1a)
M
a
=
sin
2
FS
d T F
Q Q
yc
f u
API Eq. (4.3-1b)
(plus 1/3 increase in both cases where applicable)
Where:
P
a
= Allowable capacity for brace axial load.
M
a
= Allowable capacity for brace bending moment.
F
yc
= The yield stress of the chord member at the joint
(or 0.8 of the tensile strength, if less), ksi (MPa)
FS = Safety factor =1.60
For axially loaded braces with a classification that is a mixture
of K, Y and X joints, take a weighted average of P
a
based on the
portion of each in the total load. Note that the FS value is
reduced from 1.7 to 1.6.
Strength Factor Q
u
varies with the joint and load type, as
given in Table 1 (4.3-1 in API RP2A Edition 22).
Chord Load Factor Q
f
, which accounts for the presence of
nominal loads in the chord, has been fully reformulated to
produce a much better representation of the chord stress
effects for a wide variety of joint types and loadings. Chord
stress effects are difficult to simulate in physical tests. The
finite element models used for the Edition 22 formulation
enabled better representation of the chord stress effects,
resulting in the following formulation:
Q
f
=
+
2
3 2 1
1 A C
M
FSM
C
P
FSP
C
p
ipb
y
c
API Eq. (4.3-2)
The new A parameter is defined as follows:
A
=
5 . 0
2 2
p
c
y
c
M
FSM
P
FSP API Eq (4.3-3)
(Where 1/3 increase applicable, FS=1.20 in API Equations
4.3-2 and 4.3-3).
P
c
and M
c
are the nominal axial load and bending resultant
(i.e. = + ) in the chord,
2
c
M
2
ipb
M
2
opb
M
P
y
is the yield axial capacity of the chord,
M
p
is the plastic moment capacity of the chord, and
C
1
, C
2
and C
3
are coefficients depending on joint and load type
as given in the following Table (RP2A Table 4.3-2).
VALUES FOR C
1
, C
2
and C
3
J oint Type and Loading C
1
C
2
C
3
K joints under brace axial loading 0.2 0.2 0.3
T/Y joints under brace axial loading 0.3 0 0.8
X joints under brace axial loading *
0.9
=1.0
0.2
-0.2
0
0
0.5
0.2
All joints under brace moment loading 0.2 0 0.4
*Linearly interpolated values between =0.9 and =1.0 for X joints
under brace axial loading.
The average of the chord loads and bending moments on either
side of the brace intersection should be used in the API
Equations 4.3-2 and 4.3-3. Chord axial load is positive in
tension, chord in-plane bending moment is positive when it
produces compression on the joint footprint. The chord
thickness at the joint should be used in the above calculations.
For further details of static tubular joint capacity equations
development, see Pecknold OTC 17310 (2005)
Joints with Thickened Cans. While the 0.25D joint can
extension (as detailed beyond the brace footprint) still suffices
for K-joints, the following formulation is introduced for
calculating the capacity of the simple, axially loaded Y and X
joints where a thickened joint is specified:
P
a
= [r +(1 - r) (T
n
/ T
c
)
2
] (P
a
)
c
API Eq. (4.3-4)
where:
(P
a
)
c
= P
a
from API Eq. 4.3-1a based on chord can
geometric and material properties, including Q
f
calculated with respect to chord can
4 OTC 17236
T
n
= nominal chord member thickness
T
c
= chord can thickness
r = L
c
/ (2.5 D) for joints with 0.9
= (4 - 3) L
c
/ (1.5 D) for joints with >0.9.
L = effective total length. Figure 4.3-2 gives examples for
calculation of L
c
Possible mitigations for Y-joints are discussed in Pecknold
(2005).
Strength Check: The arcsine joint interaction ratio, IR,
recommended by the 21
st
and earlier editions for axial loads
and/or bending moments in the brace has been replaced by a
parabolic relation which is found to result in better correlation
with the finite element results and adopted by the draft
ISO19902 Code:
IR =
0 . 1
M
M
M
M
P
P
opb
a
2
ipb
a a
+
+
API Eq (4.3-5)
Reliability of the New Simple Joint Formulation and
Comparison to API RP2A Edition 21
The API RP 2A, Edition 21 tubular joint design formulation
statistics are compared against the new API RP2A Edition 22
in table 3 for K, X, and Y joint configurations. The table
provides Mean Bias, Coefficient of Variation (COV) and
number of data points for the physical test database used in
API Edition 21 against the Finite Element Analysis (FEA)
database used in the API Edition 22. No FEA axial tension
data is reported because joint tension failures cannot yet be
reliably predicted by numerical methods due to the
unavailability of an appropriate and accepted failure criterion
for ductile tearing. In general, when only the physical test
database are taken into consideration, both formulations result
in not too different mean biases and COVs. However, when
the FE data is taken into consideration, striking differences are
observed.
For the balanced K joints, when FEA Database is considered
the new Edition 22 formulation biases are generally higher
while the COVs reduced significantly. The most striking COV
reductions are for the balanced axial and in plane bending load
cases. A similar trend is observed for the cross X and Y joints
especially for the X joint axial compression case where the
COV is decreased almost tenfold, from 1.33 to 0.12.
The safety index (also beta=) is the ratio of total safety
margin to total uncertainty. The 1988 safety calibration of
API RP2A found that the existing RP2A had betas of 3.4 for
90% static load, and 2.1 (lifetime) for 80% storm loading
(100-year design storm). The higher safety level was deemed
appropriate for periods when the platforms are manned and
loads are under human control. Moses (1988) proposed a
target beta of 2.44 across the board for RP2A-LRFD.
In Figures 3 and 4, the safety index for the new API Edition 22
and the old Edition 20 formulation is compared against each
other and the AWS-ASCE and API-WSD target safety indices,
now calculated for the 100% Dead Load and 100% Wave
Load cases.
Because of the lower scatter (COV), huge reductions in the
safety factor would have still given acceptable betas for
known static loads. However, for further study, a modest
reduction of the WSD safety factor to 1.6 was chosen.
Whereas APIs existing WSD safety factor of 1.7
corresponded to an LRFD resistance factor of 0.95, a WSD
safety factor of 1.62 (rounded off to 1.6) would correspond to
an LRFD resistance factor of 1.0.
Figure 3- API RP2A Edition 20, with SF=1.7.
Dead load only (static) betas for compressive axial load tests
are safely in the range of 5 to 6, and most of the experimental
betas (shaded in black) meet the target criteria. The test results
are what the Edition 20 criteria were originally based upon.
The finite element results cover a wider range of chord loading
cases (Q
f
effect) than was previously considered, and contain
some bad news. There is tremendous scatter, and most of the
finite element betas fail to meet the targets.
Storm betas tell a similar story for the old criteria.
Compressive axial load tests are all acceptable, but some of
the experimental results, and almost all of the finite element
cases, are not.
Figure 4-OTJRC Static Strength Criteria, with SF=1.6.
For the new criteria, the dead load only (static) betas for test
and finite element results are all acceptable, and their range of
scatter is much reduced. Three cases out of 20 are less
conservative than existing API; these are the experimental
axial compression cases.
The wave load only (storm) betas are all acceptable, and fall in
a tight cluster, except for the notionally more conservative
tension test results. This is because the large storm load
uncertainty overwhelms the small COVs on joint strength,
making mean bias and safety factor (both elements of reserve
strength) more important.
Conclusion. The WSD safety factor of 1.6 has been adopted
for use with the new OTJ RC static strength criteria. Static
betas greatly exceed target values from precedent, benefiting
from reduced scatter, but they do not govern. When the one-
third increase is used for storm loadings, the safety factor
becomes 1.2. Storm betas are clustered on the safe side of the
API-WSD precedent.
Fatigue Strength Design
Stress Concentration Factors (SCF). Efthymious (1988)
SCF equations are included as part of the RP2A Edition 22.
These are based on maximum principal stress, rather than
strain normal to the weld toe, and are some 15% to 20% more
conservative than the alpha (ovalizing) Kellogg formulas
recommended in RP2A editions 11 through 21. Efthymiou is
more accurate for planar T, Y, K, X, connections and the
ovalizing term is retained in Efthymious more general
multiplanar cases.
The alpha (length) factor in the Efthymiou equations for T-
and Y-connections captures beam bending as it occurs in a test
frame or an isolated FEM analysis, but is difficult to reconcile
with bending moments realized in design frame analysis,
particularly where shears result from distributed wave or
gravity loads along the chord member. One should use either
OTC 17236 5
the chord bending from Efthymiou, or that from the frame
analysis, but not both. The effect of average chord axial load
should always be added. More details on application of the
Efthymious SCF equations to offshore tubular joint design are
provided in Marshall OTC 17295 (2005) and commentary to
the RP2A Edition 22.
SCFs for grouted tubular joints are also provided in Edition
22. These are generally same as those recommended by the
Draft ISO/CD 19902 Code. The design of cast nodes is based
on local stress rather than structural hotspot stress, and a
different S-N curve applies.
New Basic Tubular Joint Design S-N curve
In the 22
nd
Edition, the Offshore Tubular J oint Task Group
(OTJTG) replaced the welded joint X and X fatigue design
curves of Editions 11 thru 21 by a basic SN curve with a slope
m=3 that changes to 5 at ten million cycles as formulated
below.
Log
10
(N) =Log
10
(k
1
) m Log
10
(S) API Eq. 5.4.1-1
where N is the predicted number of cycles to failure under
stress range S, k
1
is a constant, and m is the inverse slope of
the basic S-N curves. These values are given below:
Basic Tubular Joint Design S-N Curves
Curve Log
10
(k
1
) m
Welded Tubular J oints
(WJ )
9.95 ksi (12.48 Mpa)
11.92 ksi (16.13 Mpa)
3 for N<10
7
5 for N>10
7
Cast J oints
(CJ )
11.80 ksi (15.17 Mpa)
13.00 ksi (16.13 Mpa)
3 for N<10
7
5 for N>10
7
The basic tubular joint design S-N curve recommended in the
22
nd
Edition was the subject of extensive OTJ TG discussions
and considerations. While all task group members were in
general agreement with the shape of the low cycle S-N curve
(k
1
=9.95 ksi, m =3), there was some disagreement on the
transition point of this curve to the high cycle (k
1
=11.92 ksi,
m =5) S-N curve. The draft ISO/CD 19902 and the earlier
MSL study recommended a transition point at N =10
8
stress
cycles. However, there was no observed tubular joint fatigue
failure for N >10
7
cycles to support this high cycle transition
point. Claim of fatigue failure of a non-joint configuration
weld (a riser butt weld connection) below the first fatigue
curve, in N >10
7
cycles was made, but no such test result was
made available to OTJ TG. Based on these observations,
OTJ TG resolved to specify the transition point at N =10
7
cycles, with the provision that the use of the Edition 22 fatigue
design curves are limited to simple tubular joint designs only.
More information on the justification of the basic S-N curves
are provided in Marshall OTC17295 (2005), Commentary to
the RP2A Edition 22, Bomel (1995), and Dimitrakis (1995).
Fatigue life correction factors for seawater, thickness, and use
of weld profile control, grinding, and peening have been
introduced as summarized below. For more details see
Marshall OTC17295 Paper (2005) and commentary to the
RP2A Edition 22.
Seawater Effects. The basic design S-N curves given in Table
4 are applicable for joints in air and submerged coated joints.
New RP2A Edition 22 recommends that the basic allowable
cyclic stresses should be corrected empirically for seawater
effects (Hart, 1981). For welded joints in seawater with
adequate cathodic protection, the m=3 branch of the S-N curve
should be reduced by a factor of 2.0 on life, with m=5 branch
remaining unchanged and the position of the slope change
adjusted accordingly. For free corrosion, the reduction factor
is 3.0 on the m=3 branch, with no slope change. These
recommendations are similar to that recommended in the Draft
ISO/CD 19902 Code. Cathodic protection does not
completely restore in-air fatigue performance in the low cycle
range, where significant crack growth is involved.
Thickness effect. The basic as-welded S-N curve is for a
reference thickness of t
ref
=5/8 inch (16 mm), which obscures
the fact that the criteria become more onerous for typical joint
can thickness used offshore. For material thickness above the
reference thickness, the following thickness effect is applied
for as-welded joints:
S =S
o
if t
ref
< 5/8 inch
S =S
o
(t
ref
/t)
0.25
if t
ref
> 5/8 inch
where S =allowable stress range,
S
o
=the allowable stress range from the S-N curve, and
t =member thickness for which fatigue life is predicted.
If the weld has profile control as defined in API RP2A Figure
11.1.3d, the exponent in the above equation may be taken as
0.20. If the weld toe has been ground and peened or a cast
joint is used, the exponent in the above equation may be taken
as 0.15. For cast joints, reference thickness t
ref
is 1.5in
(38mm).
Weld improvement Techniques. For welded joints,
improvement factors on fatigue performance can be obtained
by a number of methods, including controlled burr grinding of
the weld toe, hammer peening, or as-welded profile control to
produce a smooth concave profile which blends smoothly with
the parent metal. Below table shows improvement factors that
can be applied, provided adequate quality control procedures
are followed. The grinding improvement factor is not
applicable for joints in seawater without cathodic protection.
Factors for Weld Improvement factors
Weld Improvement
Technique
Improvement
Factor on S
Improvement
Factor on N
Profiled to merge smoothly
- 0.1
Varies
Weld toe burr grind 1.25 2
Hammer peening 1.56 4
For welds with profile control, where the weld toe has been
profiled (by grinding if required) to merge smoothly with the
parent metal, and magnetic particle inspection demonstrates
the weld toe is free of surface and near-surface defects, the
improvement on fatigue performance can be considered as
shown in the Table, where is the ratio of branch/chord
thickness. This improvement is in addition to the use of
hotspot stress at the actual weld toe location, and the reduced
6 OTC 17236
size effect exponent. Either the factor on S or on N should be
used, but not both.
Welding assemblages with fully ground radius profiles and
stress relief may be considered the equivalent of castings with
weld repairs.
Cumulative Damage. As in previous editions, API RP 2A
Edition 22 specifies the Miners Rule for the calculation of
cumulative fatigue damage. This subject is further examined
in Marshall (2005).
Simplified Fatigue Design Procedure. The Simplified
Fatigue Design Procedure has been maintained, with the
allowable peak hot spot stresses rechecked for the new SN
curve. Its re-calibration as a function of the shape of the long-
term stress distribution is described by Marshall (2005),
Safety Factors. The API RP2A Editions 11 through 21 in
general required that the design fatigue life of each joint
should be at least twice intended service life of the structure
(i.e., Safety Factor =2.). It also recommended that, for critical
elements whose sole failure could be catastrophic, larger
safety factors should be considered, In concert with the draft
ISO 1992 Code and recent safety studies, RP2A Edition 22
recognized the effect of failure consequence (i.e. criticality)
and the in-service inspectability of a tubular joint design in
more detail. Critical elements are those whose sole failure
could be catastrophic. For failure-critical and non-inspectable
connections, increased safety factors are recommended as per
Table below. Reduced safety factors can be used for Category
L-2 and L-3 (Low consequence and hurricane evacuated or
unmanned) conventional steel jacket structures on the basis of
in-service performance data (for redundant framing inspected
by divers or ROV, SF of 1.0 and half the numbers in the table
for the other cases).
Fatigue Life Safety Factors
Failure Critical Inspectable Not Inspectable
No 2 5
Yes 5 10
For more details see Marshall OTC17295 (2005), commentary
to the RP2A Edition 22, Bomel (1995), Dexter (1997),
Dimitrakis (1995), Hart (1981, 85, 89), Marshall (1989, 92),
Trembath (1995), and Vosikovsky (1991).
Conclusions
The upgraded tubular J oint design procedures are expected to
result in more reliable designs under static and fatigue
loadings. The reduced scatter in the new static design
formulation justifies a reduction of the static load safety factor
from 1.7 to 1.6. Further reductions may be considered in the
future, with more reliability calibration.
The new API RP 2A tubular joint design procedures improve
tubular joint design procedures and provide better alignment
with the ISO/CD 19902 draft design procedures.
Acknowledgements
Authors acknowledge API and the members of the API SC2
on Offshore Structures who encouraged and made most funds
available for the research and development of the new API
tubular design provisions. George Rodenbush the Chairman
and Andy Radford, The API Senior Associate, are specifically
acknowledged for their support. Shell International E&P,
Paragon Engineering Services Inc., MSL, and ExxonMobil
provided considerable manpower and logistical support during
the preparation of the API revisions and this manuscript.
References
American Petroleum Institute, API (2000), Recommended Practice
for Planning, Designing, and Constructing Fixed Offshore
Platforms-Working Stress Design, API RP2A-WSD, Edition 22,
December 2000.
American Petroleum Institute, API (1990), Proposed API RP2A
Upgrade Plan, 1990 - 1999, for Joint Strength and Fatigue
Provisions, API Committee Chaired by N Zettlemoyer, 1990.
American Welding Society. Structural Welding Code, AWS D1.1,
ANSI Document.
Back, J de (1981), Strength of Tubular joints, Special and Plenary
Sessions, PS7, Proc of the 2
nd
Intl Conf Steel in Marine Structures,
Paris.
BOMEL (1995), Design and Reassessment of Tubular Joints for
Offshore Structures, Chapter 5: Fatigue life assessment, S-N
approach, BOMEL report C6060R09.07 Rev A, February 1995
Dexter, RJ and Fisher J W (1997), Fatigue and Fracture, Chapter 24 in
Chen, WF, Handbook of Structural Engineering, CRC Press
Dier, A. F (1995). and Lalani, M. Strength and Stiffness of Tubular
Joints for Assessment/Design Purposes, Paper OTC 7799, Offshore
Technology Conference, Houston, May 1995.
Dimitrakis, SD, Lawrence, FV, and Mohr, WC (1995), S-N curves
for tubular joints, final report to OTJ RC, American Petroleum Inst.
Efthymiou, M (1988), Development of SCF formulae and generalized
influence functions for use in fatigue analysis, Recent
Developments in Tubular J oint Technology, OTJ'88, October
1988, London, plus updates.
Hartt, WH (1981) Fatigue of welded structural steel in seawater, OTC
3962, Proc Offshore Tech Conf, May 1981
Hartt, WH and Lin, N (1985), Variable deflection fatigue properties
of welded steel as applicable to offshore structures, Florida
Atlantic Univ. final report to API.
Hartt, WH (1989), Weld Profile and Plate Thickness Effects in
Fatigue as Applicable to Offshore Structures, API 87-24 progress
report, Florida Atlantic University, May 1989
ISO/CD 19902, Draft E J une 2001, International Standards
Organization, Petroleum and Natural Gas Industries Offshore
Structures Part 2: Fixed Steel Structures.
ISO/DIS 14347 (2002), Fatigue Design Procedure for Welded Hollow
Section Joints Recommendations, International Standards
Organization, Geneva (as proposed by IIW-XV-E).
Lalani, M (1993), Nichols, N.W. and Sharp, J.V. The Static Strength
and Behavior of Joints in Jack-Up Rigs, Conference on Jack-up rigs,
City University, London, August 1993.
Marshall, PW (1989), Recent developments in fatigue design rules in
the USA, Fatigue Aspects in Structural Design, Delft Univ. Press
Marshall, PW (1992), API Provisions for SCF, S-N, and Size-Profile
Effects, OTC 7155, Proc Offshore Tech Conf, May 1992.
Marshall P. W (2005), Bucknell, J , Mohr W.C, Background to New
RP2A Fatigue Provisions, OTC05 Paper No: 17295, Houston TX,
2005
Marshall P. W (1974), Toprac, A. A., Basis for Tubular J oint Design)
Welding Journal, Research Supplement, May 1974
OTC 17236 7
Moment Loads, Report to the American Petroleum Institute, EWI
Project No. 42705-CAP, Edison Welding Institute, 2003.
Moses, F (1988), and Larabee, R. D., Calibration of the Draft API
RP2A-LRFD for Fixed Platforms, Proc OTC 5699, May 1988.
Pecknold D. A (2005), Marshall, P.W, Bucknell, J , New API RP2A
Tubular Joint Strength Design Provisions, OTC05 Paper No:
17295, Houston TX, 2005
MSL Engineering Limited (1996). Assessment Criteria, Reliability
and Reserve Strength of Tubular Joints, Doc. Ref. C14200R018,
Ascot, England, March 1996.
Pecknold, D.A (2000), Ha, C.C. and Mohr, W.C. Ultimate Strength
of DT Tubular Joints with Chord Preloads, Proceedings of the 19
th
International Conference on Offshore Mechanics and Arctic
Engineering, New Orleans, 2000.
Trembath, V (1995), Review of thickness effect in profiled welded
joints, MaTR 0238, Material Tech Support Unit (UK), J une 1995.
Yura, J .A. (1980), Zettlemoyer, N. and Edwards, I.F. Ultimate
Capacity Equations for Tubular Joints, OTC 3690, Houston
Pecknold, D.A (2001), Park, J .B. and Koeppenhoefer, K.C. Ultimate
Strength of Gap K Tubular Joints with Chord Preloads,
Proceedings of the 20
th
International Conference on Offshore
Mechanics and Arctic Engineering, Rio de J aneiro, 2001.
Vosikovsky, 0 and Bell, R (1991), Attachment Thickness and Weld
Profile Effects on the Fatigue Life of Welded J oints; Proc. 1991
OMAE, Stavangar
Pecknold, D.A (2003), Chang, T-Y, and Mohr, W.C. Static Strength
of T Tubular Joints with Chord Preloads under Brace Axial and
The following notes apply to Table 1:
TABLE 1- API RP 2A EDITION 22 TABLE 4.3-1-VALUES FOR Q
u
Brace Load
J oint
Classification Axial
Tension
Axial
Compression
In-plane
Bending
Out-of-Plane Bending
K (16+1.2)
1.2
Q
g
but 40
1.2
Q
g
T/Y 30
2.8 +(20+0.8)
1.6
but 2.8+36
1.6
X 23 for 0.9
20.7 +( - 0.9)
(17 - 220) for
>0.9
[2.8 +(12+0.1)]Q
(5+0.7)
1.2
2.5+(4.5+0.2)
2.6
(a) Q
=
03
1 0833
.
( . )
for >0.6
Q