Professional Documents
Culture Documents
Abstract
In this paper a design procedure for fatigue assessment of welded pipe penetrations in plated
structures is presented. The background for the procedure is given. It applies to full
penetrations welds, partial penetrations welds and fillet welds. Fatigue cracking from the weld
toe and from the root of fillet welds is considered. The procedure is based on calculated stress
concentration factors using finite element analyses for relevant geometries of penetrations
through plated structures. Numerical examples using the procedure for calculation of number
of cycles to failure are presented. These numbers are compared with fatigue test data from
specimens that were fatigue tested at the Technical University of Trondheim and Kungliga
Techniska Høgskolan in Stockholm.
r 2004 Elsevier Ltd. All rights reserved.
Keywords: Fatigue; Welded pipe penetrations; Weld toes; Weld root; Full penetration welds; Partial
penetration welds; Fillet welds; Plate structures; Finite element analysis; Stress concentration factors; S–N
curves
1. Introduction
0951-8339/$ - see front matter r 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.marstruc.2004.03.002
ARTICLE IN PRESS
30 I. Lotsberg / Marine Structures 17 (2004) 29–51
Fig. 1. Hydrophone holder welded into a structural member of the Alexander Kielland platform.
platform; see also detail in Fig. 1. The fatigue design procedure for such details in
most design standards is not very precise. Therefore work was performed within
DNV during the 90s to improve the basis for a more reliable fatigue design
procedure for welded penetrations in plated structures. This procedure is now
included in DNV CN 30.7 [2] ‘‘Fatigue Assessment of Ship Structures’’ and DNV-
RP-C203 [3] ‘‘Fatigue Strength Analysis of Offshore Structures’’. The basis for this
procedure is presented in the following. Fatigue cracking from the weld toe and from
the root of fillet welds is considered.
The procedure is based on calculated stress concentration factors using finite
element analyses for relevant geometries of penetration. Stress concentration factors
were derived by certain requirements to finite element mesh and derivation of hot
spot stress [4]. The accuracy of this procedure was further investigated in the FPSO
Fatigue Capacity JIP [5]. In the same JIP this methodology was also used to derive a
design procedure for fatigue design of manholes with different reinforcement [6]. The
analysis procedure and its basis are described more in detail in the following parts of
this paper.
A number of specimens with welded penetrations were fatigue tested at SFI in
Trondheim at the laboratories of the Technical University [7]. At this time it was not
possible to link the test results to that of a general design procedure. Later Hannus
[8] performed more fatigue testing of similar specimens at KTH in Stockholm. He
also presented some design graphs for stress concentration factors similar to that
used in the present procedure.
In the following sections of this paper calculated numbers of cycles until failure
using the described fatigue design procedure are compared with the referred test
data.
45˚
the crack driving stress.
Insert
Tubular Then σ hot spot = σ 1
Also the region at crown position to be checked.
Then σ hot spot = σ np
(b)
τ σnp
p Fatigue crack in the fillet weld (initiating from
the weld root) at region with large normal stress
45˚
(c)
1. Fatigue cracking transverse to the weld toe in a region with a large stress
concentration giving large stress parallel to the weld (Flexible reinforcement). See
Fig. 2a.
2. Fatigue cracking parallel to the weld toe (Stiff reinforcement with large weld size).
See Fig. 2b.
3. Fatigue cracking from the weld root (Stiff reinforcement with small fillet weld
size). See Fig. 2c.
All these potential regions for fatigue cracking should be assessed in a design with
the use of appropriate stress concentration factors for holes with reinforcement.
The procedure for fatigue analysis is based on the assumption that it is only
necessary to consider the ranges of the cyclic principal stresses in determining the
fatigue endurance. When the potential fatigue crack is located in the parent material
at the weld toe, the relevant hot spot stress is the range of maximum principal stress
adjacent to the potential crack location with stress concentrations being taking into
account. In some design codes such as BS 5400 [9] and DNV-RP-C203 [3] it is
normal practice to use the principal stress within 745 normal to the weld toe
together with an S–N curve for the weld for fatigue assessment as shown in Fig. 3.
ARTICLE IN PRESS
32 I. Lotsberg / Marine Structures 17 (2004) 29–51
σ hs = σ1
σ2
≤ 45°
σ1
σ1
σ2
Then the principal stress is linked to the FAT90 following the IIW [10] notation for
fatigue assessment of the weld toe location (or the D-curve following DNV-RP-C203
[3].
If the angle between the principal stress and the normal to the weld toe is larger
than 45 the fatigue life with a stress component parallel with the weld together with
an appropriate S–N curve that depends on the welding process should be used [10].
Some actual fatigue cracks can only be explained if it is assumed that it is a
principal stress at an angle larger than 45 that is initiating and driving the crack
growth. Thus IIW has suggested increasing this angle to 60 [11].
The angle a between the principal stress and the normal to the weld at the 45
position in Fig. 2b is in most cases larger than 45 . A better fit with test data is
achieved if we use the last IIW suggestion of principal stress direction together with a
weld S–N curve. Hence this definition is used in this paper.
Stress concentration factors for holes with reinforcement are given in Appendix 3
of DNV-RP-C203 [3]. Finite element analysis of selected geometries were performed
to establish stress concentration factors in this design standard and CN 30.7 [2].
Eight-node shell elements were used for the analysis of tubulars through plates,
Fig. 2. 20-node three-dimensional isoparametric elements were used for the analysis
of penetrations stiffened by flat rings, geometry in test no. 10 in Fig. 7. The welds
were not included in the shell models. Typical weld sizes were included in the three-
dimensional models. A large width of the plate were included in the FE model for
simulation of ‘‘infinite’’ width. The size of the element mesh at the hot spot region
was t t where t=plate thickness. One element was used over the main plate
thickness for one sided flat ring stiffener. For double sided flat ring stiffeners the
condition of symmetry was used which implied 2 elements over the main plate
thickness. The hot spot stress in the shell model was derived from extrapolation of
ARTICLE IN PRESS
I. Lotsberg / Marine Structures 17 (2004) 29–51 33
the surface stress at point t=2 and 3t=2 to the intersection line. The hot spot stress in
the three-dimensional model was derived from stresses at the elements nodal points
that were derived in the computer program from the Gaussian stresses. Then the
stresses at points t=2 and 3t=2 from the weld toe were used for a linear stress
extrapolation for derivation of the hot spot stress at the weld toe. Then the SCFs in
the graphs were derived by calculating the ratio of the hot spot stress and the
nominal stress as is defined in Fig. 2a. Finally the discrete data points were connected
by a curve fitting (Data points at tr =tp ¼ 0:0; 0.5, 1.0, 1.5 and 2.0, Fig. 17). Graphs
with SCFs that are used in this paper are included in the Appendix.
σ τ
τ
Throat
section τ p
σ np
is made to Lotsberg [12] for comparison with experimental data for components
subjected to a complex dynamic loading.
At some locations of the welds there are stress in the plate normal to the fillet weld,
snp ; see Fig. 2c, and a shear stress in the plate parallel with the weld tJp : The 45
position is the region selected for assessment. Equilibrium of plate in section parallel
with the weld gives:
tJ 2a ¼ tJp tp ð3Þ
where tJ is the mean nominal shear stress in the weld as shown in Fig. 4, a is the
throat thickness of weld, tp the plate thickness. The shear stress in the weld is then
obtained from Eq. (3) as
tJp tp
tJ ¼ ð4Þ
2a
And assuming a reaction force on the weld throat in the direction of snp then gives:
snp tp
t> ¼ s> ¼ pffiffiffi ð6Þ
2 2a
Then from Eq. (2) a combined stress is obtained with use of Eqs. (4) and (6). This
resulting stress range shall be used together with the W3 curve (=FAT 36 in IIW [10]
for air environment).
σ
τ
tp σ np
Fillet weld
a
3.1.1. Background
102 specimens with reinforced cut-outs were fatigue tested at SFI in Trondheim by
Skjeggestad et al. [7]. At this time the same system with S–N curves for different
details was not developed as that used for fatigue design today.
Reference is made to the design procedures for fatigue assessment of welded pipe
penetrations presented in Section 2.4. In the following the experimental fatigue test
data from Trondheim are reassessed in view of this design procedure.
1000
Nominal stress range (MPa)
100
10
Test data
E-curve (Mean)
1
10000 100000 1000000 10000000
Number of cycles
Fig. 6. Test data from Skjeggestad et al. [7] compared with the mean E-curve. Note that there are 2 test
data in the diagram at 89 000 cycles.
ARTICLE IN PRESS
36 I. Lotsberg / Marine Structures 17 (2004) 29–51
10 A B
10
100 10 100 30
10
10
2H 10 10
A B
A-A B-B
Table 1
Geometry of tested specimens with tubulars (mm)
5 6 7
Table 2
Geometry of tested specimen type no. 10 with plate reinforcement (mm)
Table 3
Fatigue test data from Skjeggestad et al. [7]
1000
Type 5
Type 6
Type 7
Hot spot stress range (MPa) D (FAT90) Mean
D (FAT90) Design
100
10
100000 1000000 10000000
Number of cycles
Fig. 8. Test data from Skjeggestad et al. [7] compared with the design procedure.
The design graphs for H=tr ¼ 2:0 in the Appendix are used for comparison with
test data for specimen types 5 and 6.
From Fig. 17 SCF=1.58. To be used together with the C–C2-curve.
From Fig. 19 SCF=1.28. To be used together with the D-curve.
The design graphs for H=tr ¼ 5:0 in the Appendix are used for comparison with
test data for specimens type 7.
From Fig. 17 SCF=1.55. To be used together with the C–C2-curve.
From Fig. 19 SCF=1.26. To be used together with the D-curve.
The observed fatigue cracks occurred at the 45 position. A comparison with test
data for this position is performed. The test data for test specimen types 5, 6 and 7
are shown in Fig. 8. The design procedure in Section 2.4 has been used to present the
test data such that they can be compared with the mean D curve.
The final crack size in the test data has not been defined. Based on assessment of
photos of the cracked specimens it is likely that the final crack sizes are larger than
that inherent in the design curves used for comparison. Thus, somewhat higher
fatigue lives would be expected from the tests as compared with the S–N curves
(failure in S–N curve in the design standard is defined as crack growth through the
plate thickness).
Even without the comment on failure criterion, the test data are found to be in
good agreement with the D-curve.
1000
Type 10
D (FAT90) Mean
100
10
10000 100000 1000000 10000000
Number of cycles
Fig. 9. Test data from Skjeggestad et al. [7] compared with the design procedure (there are 2 data points
together at the left point and 2 data points together at 446 000 cycles; i.e. there are 6 data points in this
figure).
The criticality of the two hot spots is found to be of the same magnitude (the
results in terms of calculated cycles to failure are similar). The fatigue initiation
points are not clear from the test report. However, based on the photos of the tests
after the testing it is very likely that the fatigue cracks were initiated at the weld toes.
The fatigue initiated at the 45 point, Fig. 2b (in lack of information about stress
normal to the weld at the 45 position the stress at the 0 position is used). The
fatigue data points are presented in Fig. 9. It is seen that the test data fits the
recommended S–N curve very well.
3.2.1. Background
Twelve specimens with reinforced cut-outs have been tested at Kungliga
Technisha Høgskolan (KTH) in Stockholm, [8]. In the following the experimental
fatigue test data from KTH are reassessed in view of the design procedures described
in Section 2.4.
300 100
90
90
250
860
100
250
5.5
8
30
Table 4
Table geometry of tested specimens (mm)
3 6–10 11–12
The number of cycles to failure following the design procedure for each type of
specimen is calculated (the graphs in the Appendix for H=tr ¼ 5:0 are used). In the
present paper the hot spot stress for each test is calculated for comparison with the
recommended S–N curve. The calculated stresses for weld toe assessment are shown
in Table 6 and for weld root in Table 7. The resulting S–N diagrams are presented in
Fig. 11 for weld toe failure and in Fig. 12 for failure from the weld root. A discussion
of the derived results is made in the following.
ARTICLE IN PRESS
I. Lotsberg / Marine Structures 17 (2004) 29–51 41
Table 5
Test results [8]
3 187.5 227,000
6 187.5 232,000
7 187.5 267,000
8 187.5 268,000
9 108.3 1,574,000
10 108.3 1,324,000
11 187.5 333,000
12 187.5 209,000
Table 6
Failure from the weld toe
Specimen no. Ref. figures in the Appendix SCF sHotspot (MPa) S–N curve
Table 7
Failure from the weld root
Specimen no. Ref. figure in SCF for snp SCF for tJp Equivalent stress range
the Appendix in weld sw (MPa) Eq. (2)
Use figures for
H=tr ¼ 5:0
3 Fig. 18 0.16
Fig. 20 0.48 49.86
SCFs are derived by extrapolation to line for r=tp =5.55 in the figures in the Appendix.
Specimen no. 3:
As the stiffness of the insert tubular is rather small for specimen no. 3, the
fatigue initiation may most likely occur at the point with maximum tangential stress
ARTICLE IN PRESS
42 I. Lotsberg / Marine Structures 17 (2004) 29–51
1000
Specimen no 6-10
Specimen no 11-12
D (FAT90) Mean
Hot spot stress range (MPa) D (FAT90) Design
100
10
10000 100000 1000000 10000000
Number of cycles
Fig. 11. Fatigue test data from Hannus compared with the assessment procedure for toe failure. Note that
there are two data points in the diagram at 2.7 105 cycles (specimens 7 and 8).
1000
Specimen no 6-10
Specimen no 11-12
W3 (FAT36) Mean
Stress range in weld (MPa)
W3 (FAT36) Design
100
10
10000 100000 1000000 10000000
Number of cycles
Fig. 12. Fatigue test data from Hannus compared with the assessment procedure for root failure. Note
that there are two data points in the diagram at 2.7 105 cycles (specimens 7 and 8).
(Fig. 2a). Here S–N curves C–C2 applies depending on start stop positions of the
welds. A SCF=2.15 is obtained for this position from Fig. 17. The test result is
compared with the mean of these S–N curves in Fig. 13. This hot spot will be the
most likely initiation point for fatigue cracks as the fatigue data point is well above
the mean S–N curve. This is also where the fatigues cracking occurred based on the
ARTICLE IN PRESS
I. Lotsberg / Marine Structures 17 (2004) 29–51 43
1000
Specimen no 3
C2 (FAT100) Mean
100
10
10000 100000 1000000 10000000
Number of cycles
Fig. 13. Fatigue test data from Hannus compared with the presented design procedure.
photo of the cracked specimen. (The data point would be slightly to the left of the
mean D-curve considering fatigue cracking from the weld toe, Fig. 11. The data
point would be significantly to the left of the design S–N curve for crack growth from
the weld root, Fig. 12. Thus, these are less likely initiation points for fatigue
cracking.)
Specimen no. 6–10:
From Figs. 11 and 12 it is observed that the scatter in the test data for specimens
6–10 is small. From Fig. 11 it is seen that a fatigue crack growth starting from the
weld toe is likely for these specimens (with the thickest insert tubular tested). From
ARTICLE IN PRESS
44 I. Lotsberg / Marine Structures 17 (2004) 29–51
Fig. 12, it is seen that a fatigue crack growth in the fillet weld is also likely for these
specimens. According to Hannus, the fatigue cracks have initiated at the weld toe for
this specimen with the large insert thickness tubular. Based on the photo of the tested
specimen it might also be a question if there has been some root cracking as the
fatigue crack did not grow into the tubular (Fig. 14).
Specimen no. 11 and 12:
From Figs. 11 and 12 it is observed that the scatter in the test data for specimens
11 and 12 is also small. From the same figures it is seen that a fatigue crack growth
starting from the weld toe as well as from the weld root is likely for these specimens.
ARTICLE IN PRESS
I. Lotsberg / Marine Structures 17 (2004) 29–51 45
tr
tp
tr A A
r H
A A
3.50
3.00
r/t p
100
2.50
SCF
50
2.00 100
50
20
1.50
10
1.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00
tr /t p
Fig. 17. SCF at welded pipe penetration. Stress in plate, parallel with weld.
From the test report it is observed that one of the specimens failed from a fatigue
crack starting at the weld toe that grew into the insert tubular. The other also started
at the weld toe but followed the fillet weld without growing into the insert tubular
(Figs. 15 and 16).
4. Conclusions
For a number of years it has been normal practice to use the principal
stress range within an angle 745 with the normal to the weld for assessment of
fatigue.
IIW has suggested to increase this angle from 745 to that of 760 . Fatigue tests
of welded pipe penetrations in plates that are reassessed in terms of the design
ARTICLE IN PRESS
46 I. Lotsberg / Marine Structures 17 (2004) 29–51
0.50
0.45
r/t p
0.40
10
0.35
0.30
20
SCF
0.25
50
0.20
100
0.15
50
0.10
0.05 100
0.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00
t r /t p
Fig. 18. SCF at welded pipe penetration. Stress in plate, normal to weld, Fig. 2c.
1.40
r/t p
1.35
1.30 10
1.25
20
1.20
SCF
1.15 50
1.10 100
1.05 50
1.00
100
0.95
0.90
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00
t r /t p
Fig. 19. SCF at welded pipe penetration. Principal stress in plate, Fig. 2b.
1.00
0.90 r/t p
0.80 10
0.70 20
0.60
50
SCF
0.50 100
0.40 50
0.30
0.20 100
0.10
0.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00
t r /t p
Fig. 20. SCF at welded pipe penetration. Shear stress in plate, Fig. 2c.
* The fatigue test data by Skjeggestad et al. [7] are found to comply well with the
design S–N curve that are used for the toes of cruciform fillet welded connections
today.
* The design procedure presented in Section 2.4 compare well with the test data for
reinforcement with fillet welded tubular insert (symmetric reinforcement). For
these tests a weld toe failure is the most likely failure mode.
* The design procedure compare well with the test data for plate reinforcement
welded to the main plate (non-symmetric reinforcement). For these tests a weld
toe failure is the most likely failure mode.
* The present work also shows that the test data derived by Skjeggestad et al. [7]
more than 30 years ago are valid in terms of today’s design practice. The test data
shows small scatter and the data are presented in such a manner that they can be
used for a reassessment.
* The design procedure compare very well with the test data for reinforcement with
fillet welded thin walled tubular insert. For these tests a weld toe failure is the
most likely failure mode.
ARTICLE IN PRESS
48 I. Lotsberg / Marine Structures 17 (2004) 29–51
tR
B A A tp
R SCF
A A
3.50
3.40 t R /t p
0.5
3.30
1.0
SCF
The following
3.20 relation applies 1.5
(a = throat-
thickness):
t R /t p a/t R
3.10 0.5 0.71
1.0 0.40
1.5 0.33
3.00
0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80
B/R
Fig. 21. SCF at circular reinforcement around hole. Maximum stress concentration in base material.
* The design procedure compare also very well with the test data for reinforcement
with fillet welded thick walled tubular insert. For these tests the fatigue failures
may be explained both with crack initiation from the weld toe as well as the weld
root as the data points in both cases are close to the recommended mean S–N
curve.
Thus the presented design procedure for welded pipe penetrations in plates is found
to be in general good agreement with fatigue test data for the considered test
specimens. This also indicates that the methodology used to derive stress
concentration factors for these details are sound. The hot spot stress is derived
ARTICLE IN PRESS
I. Lotsberg / Marine Structures 17 (2004) 29–51 49
SCF tR
B A A tp
R
A A
3.00
1.0
2.40
1.5
2.20
2.00
1.80
0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80
B/R
Fig. 22. SCF at circular reinforcement around hole. Maximum stress concentration at weld.
from finite element analysis with a linear stress extrapolation of read out stress t=2
and 3t=2 from the intersection line/weld toe from an element mesh of t t (t=plate
thickness). Eight-node shell elements and 20-node isoparametric solid elements were
used for these analyses.
Acknowledgements
The author would like to thank the participants of the FPSO Fatigue Capacity JIP
for making this work possible: Hyundai Heavy Industries (HHI), Daewoo
Shipbuilding & Marine Engineering (DSME), Bluewater, Total, Conoco Phillips,
Shell Deepwater Develop. Systems, Statoil, BP Exploration Operating Company
Limited, Petrobras, Navion, Bureau Veritas, Det Norske Veritas, Health and Safety
ARTICLE IN PRESS
50 I. Lotsberg / Marine Structures 17 (2004) 29–51
Table 8
Principal stress direction a for H=tr ¼ 5; Figs. 19 and 2b
0.0 90 90 90 90
0.5 66 72 80 85
1.0 54 58 65 72
1.5 49 52 56 62
2.0 46 48 52 56
Appendix A. Stress concentration factors for welded pipe penetrations and circular
reinforcements in plated structures
Graphs with stress concentration factors for welded pipe penetrations are
presented in Figs. 17–20.
Full lines applies to geometry with H=tr ¼ 5:
Broken lines applies to geometry with H=tr ¼ 2:
The principal stress direction corresponding to the SCF values in Fig. 19 for
H=tr ¼ 5 is shown in Table 8. The principal stress direction refers to the angle a
shown in Fig. 2b.
Graphs with stress concentration factors for welded circular reinforcements are
presented in Figs. 21–22.
References
[1] The Alexander Kielland Accident. NOU 1981:11 (in Norwegian with summary in English).
[2] DNV Classification Note 30.7. Fatigue Assessment of Ship Structures. Høvik: Det Norske Veritas;
2001.
[3] DNV-RP-C203. Recommended practice. Fatigue strength analysis of offshore structures. Høvik: Det
Norske Veritas; 2001.
[4] Lotsberg I, Cramer EH, Holtsmark G, Løseth R, Olaisen K, Valsg(ard S. Fatigue assessment of
floating production vessels. BOSS’97, Delft University of Technology, 1997.
[5] Fricke W. Recommended hot spot analysis procedure for structural details of FPSO’s, ships based on
round-robin FE analyses. ISOPE Stavanger. Proceedings of the 11th ISOPE, Stavanger, 2001.
International Journal of Offshore and Polar Engineering 2002;12(1).
ARTICLE IN PRESS
I. Lotsberg / Marine Structures 17 (2004) 29–51 51
[6] Chen W, Landet E. Stress analysis of cut-outs with and without reinforcement. OMAE Rio de
Janeiro, 2001.
[7] Skjeggestad B, Ringard M, Bakke E. Fatigue tests of plates with circular cutouts. Skipsteknisk
Forskningsinstitutt. Report No 76, February 1969.
[8] Hannus, H. On the fatigue design procedure of marine structural details. Report No TRITA-SKP
1056. The Naval Architecture at the Royal Institute of Technology, Stockholm, December 1985.
[9] BS 5400 Part 10. Code of practice for fatigue. Steel, concrete and composite bridges, British Standard
Institution, 1980.
[10] Hobbacher A. Fatigue design of welded joints, components. Recommendations of IIW. Abington
Publishing, Cambridge, England, 1996.
[11] Niemi E. Structural stress approach to fatigue analysis of welded components. Designer’s Guide.
XIII-1819-00, XV-1090-01, XIII-WG3-06-99.
[12] Lotsberg I. Fatigue capacity of fillet welded connections subjected to axial and shear loading. IIW
Document No XIII-2000-03 (XV-1146-03).
[13] Lotsberg I, Larsen PK. Developments in fatigue design standards for offshore structures. ISOPE,
Stavanger, June 2001.