RECOMMENDED PRACTICERP-C203

FATIGUE STRENGTH ANALYSISOF

OFFSHORE STEEL STRUCTURES OCTOBER 2001

DET NORSKE VERITAS

Comments may be sent by e-mail to UXOHV#GQYFRP

For subscription orders or information about subscription terms, please use GLVWULEXWLRQ#GQYFRP

Comprehensive information about DNV services, research and publications can be found at KWWS://www.dnv.com, or can be obtained fromDNV, Veritasveien 1, N-1322 Hvik, Norway; Tel +47 67 57 99 00, Fax +47 67 57 99 11.

Det Norske Veritas. All rights reserved. No part of this publication may be reproduced or transmitted in any form or by any means, includingphotocopying and recording, without the prior written consent of Det Norske Veritas.

Computer Typesetting by Det Norske Veritas.

Printed in Norway by GCS AS.If any person suffers loss or damage which is proved to have been caused by any negligent act or omission of Det Norske Veritas, then Det Norske Veritas shall pay compensation to such person forhis proved direct loss or damage. However, the compensation shall not exceed an amount equal to ten times the fee charged for the service in question, provided that the maximum compensation shallnever exceed USD 2 million.

In this provision Det Norske Veritas shall mean the Foundation Det Norske Veritas as well as all its subsidiaries, directors, officers, employees, agents and any other acting on behalf of Det NorskeVeritas.

DET NORSKE VERITAS (DNV) is an autonomous and independent foundation with the objectives of safeguarding life, prop-erty and the environment, at sea and onshore. DNV undertakes classification, certification, and other verification andconsultancy services relating to quality of ships, offshore units and installations, and onshore industries world-wide, andcarries out research in relation to these functions.

DNV Offshore publications consist of a three level hierarchy of documents:

Provide principles and procedures of DNV classification, certification, verification and con-sultancy services.

Provide technical provisions and acceptance criteria for general use by the offshore industry as well as thetechnical basis for DNV offshore services.

Provide proven technology and sound engineering practice as well as guidance for the higher levelOffshore Service Specifications and Offshore Standards.

DNV Offshore publications are offered within the following areas:A) Quality and Safety MethodologyB) Materials TechnologyC) StructuresD) SystemsE) Special FacilitiesF) Pipelines and RisersG) Asset Operation

This Recommended Practice is developed in close co-operation with the offshore industry, research institutes and universities.All contributions are highly appreciated.

Following main changes are made:

In the calculation of SCFs for butt welds and cruciform joints the misalignment may be reuduced with a value 0 which areinherent in the S-N data. See Section 2.6,2.8.7 and 2.12.

Calculation of reduced hot spot stress when weld profiling is performed, equation (4.2.3).

!!"#$%

$&"'

DET NORSKE VERITAS

' $"!($% )1.1 General .................................................................. 41.2 Validity of standard ............................................... 41.3 Methods for fatigue analysis.................................. 41.4 Guidance to when a detailed fatigue analysis can be

omitted .................................................................... 41.5 Symbols ................................................................. 4 #$%*( #+,-%-.#-! #$%*(-$- /2.1 Introduction ........................................................... 62.2 Stresses to be considered ....................................... 62.3 S-N curves ............................................................. 72.4 Mean stress influence for non welded structures. 122.5 Effect of fabrication tolerances............................ 122.6 Stress concentration factors for plated structures 132.7 Stress concentration factors for ship details ........ 152.8 Stress concentration factors for tubular joints and

members ................................................................ 152.9 Stress concentration factors for joints with square

sections.................................................................. 22

2.10 Fillet and partial penetration welds ......................222.11 Bolts .....................................................................242.12 Pipelines...............................................................242.13 Calculation of hot spot stress by finite element

analysis ..................................................................252.14 Simplified fatigue analysis...................................28 #$%*(# #+,-%--! 0"#$("1# %-3.1 Introduction..........................................................32) 2"3 $0#$%*(%0&,#&"%#$% 4.1 General.................................................................334.2 Weld profiling by machining and grinding ..........334.3 Grinding ...............................................................334.4 TIG dressing.........................................................344.5 Hammer peening..................................................344 5$ !!0#$%*(+%0 4/ "$# $%-% #$%*(%0"!%%$ /6.1 General.................................................................366 0" -6

22 !%5'7+#--%0%#$% 0$"($("#+$#%+-888889

22 !%57:-0"(&(+#";% $-88888888849

22 !%57:-0" $"#$% -

) !!"#$%

$&"'

DET NORSKE VERITAS

' $"!($%

'' "#+This Recommended Practice presents recommendations inrelation to fatigue analyses based on fatigue tests and fracturemechanics. Conditions for the validity of the RecommendedPractice are given in section 1.2 below.

The aim of fatigue design is to ensure that the structure hasan adequate fatigue life. Calculated fatigue lives also formthe basis for efficient inspection programmes duringfabrication and the operational life of the structure.

To ensure that the structure will fulfil its intended function, afatigue assessment, supported where appropriate by adetailed fatigue analysis, should be carried out for eachindividual member, which is subjected to fatigue loading.See also section 1.4. It should be noted that any element ormember of the structure, every welded joint and attachmentor other form of stress concentration, is potentially a sourceof fatigue cracking and should be individually considered.

' #+%!%$,0-$# !#"!This Recommended Practice is valid for steel materials in airwith yield strength less than 700 MPa. For steel materials inseawater with cathodic protection or steel with free corrosionthe Recommended Practice is valid up to 500 MPa.

This Recommended Practice is also valid for bolts in airenvironment or with protection corresponding to thatcondition of grades up to 10.9, ASTM A490 or equivalent.

This RP may be used for stainless steel.

' $1!-0"0#$%*(# #+,-%-The fatigue analysis should be based on S-N data,determined by fatigue testing of the considered weldeddetail, and the linear damage hypothesis. When appropriate,the fatigue analysis may alternatively be based on fracturemechanics. If the fatigue life estimate based on fatigue testsis short for a component where a failure may lead to severeconsequences, a more accurate investigation considering alarger portion of the structure, or a fracture mechanicsanalysis, should be performed. For calculations based onfracture mechanics, it should be documented that there is asufficient time interval between time of crack detectionduring in-service inspection and the time of unstable fracture.

All significant stress ranges, which contribute to fatiguedamage in the structure, should be considered. The long termdistribution of stress ranges may be found by deterministic orspectral analysis, see also ref. /1/. Dynamic effects shall beduly accounted for when establishing the stress history. Afatigue analysis may be based on an expected stress history,which can be defined as expected number of cycles at eachstress range level during the predicted life span. A practicalapplication of this is to establish a long term stress rangehistory that is on the safe side. The part of the stress rangehistory contributing most significantly to the fatigue damageshould be most carefully evaluated. See also Appendix 4,Commentary, for some guidance.

It should be noted that the shape parameter h in the Weibulldistribution has a significant impact on calculated fatiguedamage. For effect of the shape parameter on fatigue damagesee also design charts in Figure 2.14-1 and Figure 2.14-2.Thus, when the fatigue damage is calculated based on closedform solutions with an assumption of a Weibull long termstress range distribution, a shape parameter to the safe sideshould be used.

') (%!# $

!!"#$% 4

$&"'

DET NORSKE VERITAS

SCFAS Stress concentration factor at the saddle for axialload

SCFAC Stress concentration factor at the crown for axialload

SCFMIP Stress concentration factor for in plane momentSCFMOP Stress concentration factor for out of plane momentT Thickness of chordTe Equivalent thickness of chordTd Design life in secondsQ Probability for exceedance of the stress range a Crack depthai Half crack depth for internal cracks Intercept of the design S-N curve with the log N axise- Exp(-)g Gap = a/D; factor depending on the geometry of the

member and the crack.h Weibull shape parameter, weld sizek number of stress blocks, exponent on thicknessl segment lengths of the tubularm negative inverse slope of the S-N curve; crack

growth parameterni number of stress cycles in stress block ino is the number of cycles over the time period for

which the stress range level o is defined.tref reference thicknesst plate thickness, thickness of brace membertc cone thicknesstp plate thicknessq Weibull scale parameter

gamma function usage factor the slope angle of the cone; = L/D d/D eccentricity0 eccentricity inherent in the S-N curve R/To average zero-crossing frequency Poissons rationominal nominal stresshot spot hot spot stress or geometric stressx Maximum nominal stresses due to axial forcemy and mz maximum nominal stresses due to bending about the

y-axis and the z-axis stress range

ostress range exceeded once out of n0 cycles

t/T

/ !!"#$%

$&"'

DET NORSKE VERITAS

#$%*( #+,-%-.#-! #$%*(-$-

' $"!($% The fatigue life may be calculated based on the S-N fatigueapproach under the assumption of linear cumulative damage(Palmgren-Miner rule).When the long-term stress range distribution is expressed bya stress histogram, consisting of a convenient number ofconstant amplitude stress range blocks i each with anumber of stress repetitions ni the fatigue criterion reads:

( ) ====

P

N

L

LL

N

L

L

L

11

1 =''>

where

= accumulated fatigue damage

= intercept of the design S-N curve with the log N axis

= negative inverse slope of the S-N curve

= number of stress blocks

i = number of stress cycles in stress block

i = number of cycles to failure at constant stress rangei

= usage factor

= 1 / Design Fatigue Factor from OS-C101 Section 6Fatigue Limit States.

Applying a histogram to express the stress distribution, thenumber of stress blocks, , should be large enough to ensurereasonable numerical accuracy, and should not be less than20. Due consideration should be given to selection ofintegration method as the position of the integration pointsmay have a significant influence on the calculated fatigue lifedependent on integration method.

See also section 2.14 for calculation of fatigue damage usingthe simplified method.

$"---$& -%!"!

' +#$!-$"($("-

The procedure for the fatigue analysis is based on theassumption that it is only necessary to consider the ranges ofcyclic principal stresses in determining the fatigue endurance(i. e. mean stresses are neglected for fatigue assessment ofwelded connections).

When the potential fatigue crack is located in the parentmaterial at the weld toe, the relevant hot spot stress is therange of maximum principal stress adjacent to the potentialcrack location with stress concentrations being taken intoaccount.

For joints other than tubular joints, the joint classificationand corresponding S-N curves takes into account the localstress concentrations created by the joints themselves and bythe weld profile. The design stress can therefore be regardedas the nominal stress, adjacent to the weld underconsideration. However, if the joint is situated in a region ofstress concentration resulting from the gross shape of thestructure, this must be taken into account in calculating thenominal stress. As an example, for the weld shown in Figure2.2-1a), the relevant local stress for fatigue design would bethe tensile stress, . For the weld shown in Figure 2.2-1b),the stress concentration factor for the global geometry mustin addition be accounted for, giving the relevant local stressequal to SCFnominal, where SCF isthe stress concentrationfactor due to the hole.

nominallocal SCF= ='>

local shall be used together with the relevant S-N curves Dthrough G, dependent on joint classification.The maximum principal stress range within 45 of thenormal to the weld toe should be used for the analysis.

For detailed finite element analysis of welded connectionsother than tubular joints it may also be convenient to use thealternative hot spot stress for fatigue life assessment, seesection 2.13.3 for further guidance.

(&(+#"?% $-

For a tubular joint, i. e. brace to chord connection, t he stressto be used for design purpose is the range of idealised hotspot stress defined by: the greatest value of the extrapolationof the maximum principal stress distribution immediatelyoutside the region effected by the geometry of the weld. Thehot spot stress to be used in combination with the T-curve iscalculated as

nominalstressspothot SCF= =>

Where

SCF = stress concentration factor as given in section 2.8.

Where support plating below bearings are designed withfillet welded connection, it should be verified that fatiguecracking of the weld will not occur. Even though the jointmay be required to carry wholly compressive stresses and theplate surfaces may be machined to fit, for fatigue purposes,the total stress fluctuation should be considered to betransmitted through the welds.

!!"#$% 6

$&"'

DET NORSKE VERITAS

If it is assumed that compressive loading is transferredthrough contact, it should be verified that contact will not belost during the welding. The actual installation conditionincluding maximum construction tolerances should beaccounted for.

%*("'52+# #$% 0+#+-$"---

%++$

Throatsection

%*("52+# #$% 0-$"--- $1$1"#$-$% 0#0%++$

@ !!"#$%

$&"'

DET NORSKE VERITAS

The fatigue design is based on use of S-N curves, whichare obtained from fatigue tests. The design S-N curveswhich follows are based on the mean-minus-two-standard-deviation curves for relevant experimental data. The S-Ncurves are thus associated with a 97.6% probability ofsurvival.

The basic design S-N curve is given as

logmalogNlog = ='>

N = predicted number of cycles to failure for stressrange

= stress range

m = negative inverse slope of S-N curve

log = intercept of log -axis by S-N curve

s2alogalog = =>

where

a = constant relating to mean S-N curve

s = standard deviation of log N.

The fatigue strength of welded joints is to some extentdependent on plate thickness. This effect is due to the localgeometry of the weld toe in relation to thickness of theadjoining plates. See also effect of profiling on thicknesseffect in Section 4.2. It is also dependent on the stressgradient over the thickness. The thickness effect isaccounted for by a modification on stress such that thedesign S-N curve for thickness larger than the referencethickness reads, see also Appendix 4, Commentary:

=

k

refttlogmalogNlog

=>

where

m = negative inverse slope of the S - N curve

log = intercept of log N axis

tref = reference thickness equal 25 mm for weldedconnections other than tubular joints. For tubularjoints the reference thickness is 32 mm. For boltstref = 25 mm.

t = thickness through which a crack will most likelygrow. t= tref is used for thickness less than tref.

k = thickness exponent on fatigue strength as given inTable 2.3-1, Table 2.3-2 and Table 2.3-3.

k = 0.10 for tubular butt welds made from one side.

k = 0.40 for threaded bolts subjected to stressvariation in the axial direction.

In general the thickness exponent is included in the designequation to account for a situation that the actual size ofthe structural component considered is different ingeometry from that the S-N data are based on. Thethickness exponent is considered to account for differentsize of plate through which a crack will most likely grow.To some extent it also accounts for size of weld andattachment. However, it does not account for weld lengthor length of component different from that tested such as e.g. design of mooring systems with a significant largernumber of chain links in the actual mooring line than whatthe test data are based on. Then the size effect should becarefully considered using probabilistic theory to achieve areliable design, see Appendix 4, Commentary.

!!"#$% 9

$&"'

DET NORSKE VERITAS

("3-% #%"

S-N curves for air environment are given in Table 2.3-1and Figure 2.3-1. The T curve is shown in Figure 2.3-3.

#&+'("3-% #%" 1logD

2log>

!

"#

$%

&' (

B1 12.913 16.856 93.57 0B2 12.739 16.566 81.87 0C 12.592 16.320 73.10 0.15

C1 12.449 16.081 65.50 0.15C2 12.301 15.835 58.48 0.15D 12.164 15.606 52.63 0.20 1.00E 12.010 15.350 46.78 0.20 1.13F 11.855 15.091 41.52 0.25 1.27

F1 11.699 14.832 36.84 0.25 1.43F3 11.546 14.576 32.75 0.25 1.61G 11.398 14.330 29.24 0.25 1.80

W1 11.261 14.101 26.32 0.25 2.00W2 11.107 13.845 23.39 0.25 2.25W3 10.970 13.617 21.05 0.25 2.50T 12.164 15.606 52.63 0.25 for SCF 10.0

0.30 for SCF >10.01.00

*) see also section 1.4

10

100

1000

1.00E+04 1.00E+05 1.00E+06 1.00E+07 1.00E+081XPEHURIF\FOHV

6

W

U

H

V

V

U

D

Q

J

H

0

3

D

%

%

&

&

&

'

(

)

)

)

*

:

:

:

%*("'("3-% #%"

' !!"#$%

$&"'

DET NORSKE VERITAS

("3-% -#

!!"#$% ''

$&"'

DET NORSKE VERITAS

) ("3-0"$(&(+#"?% $-

S-N curves for tubular joints in air environment and inseawater with cathodic protection are given in Table 2.3-1,Table 2.3-2 and Figure 2.3-3.

1

10

100

1000

1.00E+04 1.00E+05 1.00E+06 1.00E+07 1.00E+08 1.00E+09

1XPEHURIF\FOHV

6

W

U

H

V

V

U

D

Q

J

H

0

3

D

Seawater with cathodic protection

In air

%*("("3-0"$(&(+#"?% $-% #%"# !% -#

' !!"#$%

$&"'

DET NORSKE VERITAS

#&+("3-% -#

!!"#$% '

$&"'

DET NORSKE VERITAS

/ $"-- $"#$% 0#$"-0"2+#$!-$"($("-

/' "#+

A stress concentration factor may be defined as the ratio ofhot spot stress range over nominal stress range.

The eccentricity between welded plates may be accountedfor in the calculation of stress concentration factor. Thefollowing formula applies for a butt weld in an unstiffenedplate or for a pipe butt weld with a large radius:

t

)(31SCF 0m

+=

=/'>

where m is eccentricity (misalignment) and t is platethickness, see Figure 2.8-4. 0 = 0.1t is misalignmentinherent in the S-N data for for butt welds.

The stress concentration for the weld between plates withdifferent thickness in a stiffened platefield may be derivedfrom the following formula:

( )

+

++=

5.1

5.10m

1

61SCF

W

7

W

W

=/>

where

m = maximum misalignment

t = ( )W7 eccentricity due to change in thickness0 = 0.1t is misalignment inherent in the S-N data for

butt welds

T = thickness of thicker plate

t = thickness of thinner plate

See also Figure 2.8-3.

The stress concentration factor for cruciform joint may bederived from following formula:

+++

+=

4

34

3

33

2

32

1

31

1

02

l

)(61SCF

O

W

O

W

O

W

O

W

W

Where

= (m + W) is the total eccentricity.

0 = 0.15t is misalignment inherent in the S-N data forcruciform joints

t = thickness of the considered plate

The other symbols are defined in Figure 2.6-1.

l3

l4

l2 l1

t2 t1

t3

t4

Stress concentration factors for rounded rectangular holesare given in Figure 2.6-2.

Where there is one stress raiser close to another detailbeing evaluated with respect to fatigue, the interaction ofstress between these should be considered. An example ofthis is a welded connection in a vicinity of a hole. Then theincrease in stress at the considered detail due to the holecan be evaluated from Figure 2.6-3.

Some guidelines on effect of interaction of different holescan be found in Peterson's Stress Concentration Factors,/15/).

!"# $

%&$$

DET NORSKE VERITAS

0.00

0.20

0.40

0.60

0.80

1.00

1.20

1.40

1.60

1.80

2.00

2.20

2.40

2.60

2.80

3.00

1.00 1.20 1.40 1.60 1.80 2.00 2.20 2.40 2.60 2.80 3.00 3.20 3.40 3.60 3.80 4.00

5HODWLYHGLVWDQFHIURPFHQWUHRIKROH[U

5

H

O

D

W

L

Y

H

V

W

U

H

V

V

r

Stress direction

x/r

Line for calculation of stressLine for calculationof stress

r

x

&

!"# $ '

%&$$

DET NORSKE VERITAS

(

Stress concentration factors for holes with reinforcementare given in Appendix 3.

At welds on reinforced rings fatigue cracking may occur atseveral locations depending on geometry of ring and weldsize:

Fatigue cracking transverse to the weld toe in theregion with a large stress concentration giving largestress parallel to the weld.

Fatigue cracking normal to the weld toe. Fatigue cracking from the weld root.

All these potential regions for fatigue cracking should beinvestigated. For stresses to be used together with thedifferent S-N curves see section 2.2.

Potential fatigue cracking transverse to the weld toe

For stresses parallel with the weld the local stress to beused together with the C curve is obtained with SCF fromAppendix 3 (t in Figure 2.6-4 c).Potential fatigue cracking parallell wih the weld toe

For stresses normal to the weld the resulting hot spot stressto be used together with the D curve is obtained with SCFfrom Appendix 3 (n Figure 2.6-4 a and b).Potential fatigue cracking from the weld root

At some locations of the welds there are stress in the platetransverse to the fillet weld, n, and stress in the plateparallel with the weld t, see Figure 2.6-4 b. Then the filletweld is designed for a combined stress obtained as

22 2.02 WQZ

+=

where

t = plate thickness

a = throat thickness for a double sided fillet weld.

This equation can be outlined from equation (2.2.3) andthe resulting stress range is to be used together with theW3 curve. The basic stress in the plate as shown in Figure2.6-4 is derived from Appendix 3. Also fatigue crackingfrom the weld root should be analysed for the stresscondition in Figure 2.6-4 a using equation (2.6.4) (witht=0).(n and t include SCFs from Appendix 3)

Q

Q

W

D E

W

F

&(

) *Stress concentration factors for ship details may be foundin Fatigue Assessment of Ship Structures (CN 30.7), ref./1/. S-N curve C from this RP may be used if theprocedure of CN 30.7 is used to determine the hot spot andKw stress. S-N curve D from this RP may be used if theprocedure of CN 30.7 is used to determine the local stress(Excluding the stress concentration factor due to the weldgeometry, Kw, from the analysis, as this factor is accountedfor in the D-curve).

+ &

&

+ *&

Stress concentration factors for simple tubular joints aregiven in Appendix 2 of this RP.

+ **&

The stresses are calculated at the crown and the saddlepoints, see Figure 2.8-1. Then the hot spot stress at thesepoints is derived by summation of the single stresscomponents from axial, in-plane and out of plane action.The hot spot stress may be higher for the intermediatepoints between the saddle and the crown. The hot spotstress at these points is derived by a linear interpolation ofthe stress due to the axial action at the crown and saddleand a sinusoidal variation of the bending stress resultingfrom in-plane and out of plane bending. Thus the hot spotstress should be evaluated at 8 spots around thecircumference of the intersection, ref. Figure 2.8-2.

mzMOPmyMIPxASAC8

mxMOPxAS7

mzMOPmyMIPxASAC6

myMIPxAC5

mzMOPmyMIPxASAC4

mxMOPxAS3

mzMOPmyMIPxASAC2

myMIPxAC1

SCF221SCF2

21)SCF(SCF

21

SCFSCF

SCF221SCF2

21)SCF(SCF

21

SCFSCF

SCF221SCF2

21)SCF(SCF

21

SCFSCF

SCF221SCF2

21)SCF(SCF

21

SCFSCF

+=

=

++=

+=

+++=

+=

++=

= +

!"# $

%&$$

DET NORSKE VERITAS

Here x, my and mz are the maximum nominal stressesdue to axial load and bending in-plane and out-of-planerespectively. SCFAS is the stress concentration factor at thesaddle for axial load and the SCFAC is the stress

concentration factor at the crown. SCFMIP is the stressconcentration factor for in plane moment and SCFMOP isthe stress concentration factor for out of plane moment.

Braced

Crown Toe

D

g

Saddle

Crown Heel

Chord

t

T

NM OP

IPM

+,

&

Axial load

z

x y1 2

3456

78

In-plane Out-of-planebending moment bending moment

N MIP MOP

+**

Influence functions may be used as an alternative to theprocedure given here to calculate hot spot stress. See e.g.Combined Hot-Spot Stress Procedures for TubularJoints, ref. /23/ and Development of SCF Formulae andGeneralised Influence Functions for use in FatigueAnalysis ref. /2/.

+ -&(

The root area of single-sided welded tubular joints may bemore critical with respect to fatigue cracks than the outsideregion connecting the brace to the chord. In such cases, itis recommended that stubs are provided for tubular jointswhere high fatigue strength is required, such that weldingfrom the backside can be performed.

Failure from the root has been observed at the saddleposition of tubular joints where the brace diameter is equalthe chord diameter, both in laboratory tests and in service.It is likely that fatigue cracking from the root might occurfor rather low stress concentrations. Thus, special attentionshould be given to joints other than simple joints, such asring-stiffened joints and joints where weld profiling orgrinding on the surface is required to achieve sufficientfatigue life. It should be remembered that surfaceimprovement does not increase the fatigue life at the root.

!"# $ )

%&$$

DET NORSKE VERITAS

Based on experience it is not likely that fatigue crackingfrom the inside will occur earlier than from the outside forsimple T and Y joints and K type tubular joints. The sameconsideration may be made for X-joints with diameterratio joints with > 0.90 it is recommended that a fatigueassessment of the root area is performed. Some guidanceon such an assessment can be found in Appendix 4,Commentary.

Due to limited accessibility for in service inspection ahigher design fatigue factor should be used for the weldroot than for the outside weld toe hot spot. Reference isalso made to Appendix 4, Commentary

+ &

Equations for joints for ring stiffened joints are given inStress Concentration Factors for Ring-Stiffened TubularJoints, ref. /3/. The following points should be notedregarding the equations:

The derived SCF ratios for the brace/chordintersection and the SCF's for the ring edge are meanvalues, although the degree of scatter and proposeddesign factors are given.

Short chord effects shall be taken into account whererelevant.

For joints with diameter ratio 0.8, the effect ofstiffening is uncertain. It may even increase the SCF.

The maximum of the saddle and crown stressconcentration factor values should be applied aroundthe whole brace/chord intersection.

The following points can be made about the use of ringstiffeners in general:

Thin shell FE analysis should be avoided forcalculating the SCF if the maximum stress is expectedto be near the brace-ring crossing point in the fatigueanalysis.

Ring stiffeners have a marked effect on thecircumferential stress in the chord, but have little orno effect on the longitudinal stress.

Ring stiffeners outside the brace footprint have littleeffect on the SCF, but may be of help for the staticstrength.

Failures in the ring inner edge or brace ring interfaceoccur internally, and will probably only be detectedafter through thickness cracking, at which the majorityof the fatigue life will have been expired. These areasshould therefore be considered as non-inspectableunless more sophisticated inspection methods areused.

+' ,&

Grouted joints have either the chord completely filled withgrout (single skin grouted joints) or the annulus betweenthe chord and an inner member filled with grout (doubleskin grouted joints). The SCF of a grouted joint dependson load history. The SCF is less if the bond between thechord and the grout is unbroken. For model testing ofgrouted joints the bond should be broken prior to SCFmeasurements. Due to the grout the tensile andcompressive SCF may be different.

To achieve a fatigue design that is to the safe side it isrecommended to use SCFs derived from tests where thebounds are broken and where the joint is subjected to atension loading. The bounds can be broken by a significantloading in tension. This load level may be determinedduring the testing by an evaluation of the forcedisplacement relationship. (When incrementing theloading into a non-linear behaviour).The grouted joints shall be treated as simple joints, exceptthat the chord thickness in the term for saddle SCFcalculation for brace and chord shall be substituted with anequivalent chord wall thickness given by

134T)/144(5DTe += +where D and T are chord diameter and thicknessrespectively.

Joints with high or low ratios have little effect ofgrouting. The benefits of grouting should be neglected forjoints with > 0.9 or 12.0 unless documentedotherwise.

+

It is recommended that finite element analysis should beused to determine the magnitude and location of themaximum stress range in castings sensitive to fatigue. Thefinite element model should use volume elements at thecritical areas and properly model the shape of the joint.Consideration should be given to the inside of the castings.The brace to casting circumferential butt weld (which isdesigned to an appropriate S-N curve for suchconnections) may be the most critical location for fatigue.

+) &&(

Due to less severe S-N curve for the outside than theinside, it is strongly recommended that tubular butt weldconnections are designed such that any thicknesstransitions are placed on the outside (see Figure 2.8-3). Forthis geometry, the SCF for the transition applies to theoutside. On the inside it is then conservative to use SCF =1.0. Thickness transitions are normally to be fabricatedwith slope 1:4.

+ !"# $

%&$$

DET NORSKE VERITAS

Outside

Inside

Neutralaxis

14

nominal

L

Tt

t

+ .&&(

Stress concentrations at tubular butt weld connections aredue to eccentricities resulting from different sources.These may be classified as concentricity (difference intubular diameters), differences in thickness of joinedtubulars, out of roundness and centre eccentricity, seeFigure 2.8-5 and Figure 2.8-6. The resulting eccentricitymay be conservatively evaluated by a direct summation ofthe contribution from the different sources. Theeccentricity due to out of roundness normally gives thelargest contribution to the resulting eccentricity .

It is conservative to use the formula for plate eccentricitiesfor calculation of SCF at tubular butt welds. The effect ofthe diameter in relation to thickness may be included byuse of the following formula, provided that T/t 2:

-

5.20t e

tT1

1t

)6(1SCF

+

++=

P

+

where

2.5

tT1

1tD

1.82L

+=

0 = 0.1t is misalignment inherent in the S-N data.

This formula also takes into account the length over whichthe eccentricity is distributed: L, ref. Figure 2.8-4 andFigure 2.8-3. The stress concentration is reduced as L isincreased and/or D is reduced. It is noted that for small Land large D the last formula provides stress concentrationfactors that are close to but lower that of the simplerformula for plates.

t

D

L

m

+(

!"# $ /

%&$$

DET NORSKE VERITAS

The transition of the weld to base material on the outsideof the tubular can normally be classified corresponding toS-N curve E. If welding is performed in a horizontalposition it can be classified as D.

In tubulars, the root side of welds made from one side isnormally classified as F3. This requires goodworkmanship during construction, in order to ensure fullpenetration welds, and that work is checked by non-destructive examination. It may be difficult to document afull penetration weld in most cases due to limitations in thenon-destructive examination technique to detect defects inthe root area. The F3 curve can be considered to accountfor some lack of penetration, but it should be noted that amajor part of the fatigue life is associated with the initialcrack growth while the defects are small. This may beevaluated by fracture mechanics such as described inBS 7910 Guidance on Methods for Assessing theAcceptability of Flaws in Fusion Welded Structures, ref/7/. Therefore, if a fabrication method is used where lackof penetration is to be expected, the design S-N curvesshould be adjusted to account for this by use of fracturemechanics.

For global bending moments over the tubular section it isthe nominal stress derived at the neutral axis of Figure2.8-3 that should be used together with an SCF fromequation (2.8.3) for calculation of hot spot stress.

A A

Section A-Aa) Concentricity

t

t

m

A A

b) Thickness Section A-A

T

t

= (T-t)t

+',

&&(

A

Section A-Ac) Out of roundness

A

t

t

m

mm

A A

Section A-Ad) Center eccentricity

t

t

mm

+,

&&(

++

The stress concentration at a ring stiffener can becalculated as

rArt1.56t1

shell theof inside for the0.541SCF

shell theof outside for the0.541SCF

+=

=

+=+

where

Ar = area of ring stiffener without effective shell.

r = radius of shell measured from centre to mean shellthickness

t = thickness of shell plating.

It can thus be noted that it is more efficient to place ringstiffeners on the inside of shell, as compared with theoutside. In addition, if the shell comprises longitudinalstiffeners that are ended, it is recommended to end thelongitudinal stiffeners against ring stiffeners for the inside.The corresponding combination on the outside gives aconsiderably larger stress concentration.

$ !"# $

%&$$

DET NORSKE VERITAS

The SCF = 1.0 if continuous longitudinal stiffeners areused.

In the case of a bulkhead instead of a ring, Ar is takenas( )1

tr b

, where tb is the thickness of the bulkhead.

+)

+/

The stress concentration at each side of unstiffenedtubular-cone junctions outside can be estimated by thefollowing equations (the SCF shall be used together withthe stress in the tubular at the junction for both the tubularand the cone side of the weld):

tant

)t(tDt0.61SCF 2

cj ++=

+'

tant

)t(tDt0.61SCF 2

c

cj ++=

+

where

Dj = cylinder diameter at junction (Ds, DL)t = tubular member wall thickness (ts, tL)tc = cone thickness = the slope angle of the cone (see Figure 2.8-8)

The stress concentration at a junction with ring stiffenercan be calculated as

r

j

r

j

r

j

r

j

r

j

AtD1.10t

1

andjunction,diameterlargerinsidetheat

1tan

At0.91D

0.541SCF

junctiondiameterlargeroutsidetheat

1tan

At0.91D

0.541SCF

junctiondiametersmallerinsidetheat

1tan

At0.91D

0.541SCF

junctiondiametersmalleroutsidetheat

1tan

At0.91D

0.541SCF

+=

=

+=

+=

++=+)

where

Ar = area of ring stiffener without effective shell.If a ring stiffener is placed a distance away from theintersection lines, an additional stress concentration shouldbe included to account for this eccentricity:

tant

31SCF += ++

!"# $

%&$$

DET NORSKE VERITAS

Ds

DL

ts

tL

.

tC

Ds

D L

ts

t L

tc e

++0

+$ &&1

This section applies to tubular sections welded together tolong strings and subjected to axial tension. Tethers andrisers of a TLP are examples of such structures.

The colinearity with small angle deviation betweenconsecutive fabricated tubular segments results inincreased stress due to a resulting global bending moment,see Figure 2.8-9. The eccentricity due to colinearity is afunction of axial tension in the tubular and is significantlyreduced as the axial force is increased by tension.Assuming that the moment M results from an eccentricityN where pretension is accounted for in the analysis, thefollowing derivation of a stress concentration factor isperformed:

( ) SCFttDN

=+/

where the stress concentration factor is:

tD4

1SCF N

+= +$

where N is eccentricity as function of the axial force NSdand D is outer diameter. The eccentricity for two elementsis indicated in Figure 2.8-10. With zero tension theeccentricity is . With an axial tension force NSd theeccentricity becomes:

klkltanh

N =+

where

k =EINSd

= segment lengths of the tubulars

NSd= axial force in tubulars

I = moment of inertia of tubulars

E = Youngs modulus.

The formula for reduction in eccentricity due to increasedaxial force can be deduced from differential equation forthe deflected shape of the model shown in Figure 2.8-10.Thus the non-linearity in terms of geometry is included inthe formula for the stress concentration factor.

Judgement should be used to evaluate the number ofelements to be considered, and whether deviation from astraight line is systematic or random, ref. Figure 2.8-9. Inthe first case, the errors must be added linearly, in thesecond case it may be added quadratically.

+/02**&

3450234452

!"# $

%&$$

DET NORSKE VERITAS

+$6

00

/ (7Stress concentration factors for T- and X- square to squarejoints may be found in Proposed Revisions for FatigueDesign of Planar Welded Connections made of HollowStructural Sections, ref. /27/.

Stress concentration factors for Y- and K square to squarejoints for d/Dw less or equal to 0.75 may be found fromStress Concentrations in T/Y and K Square-to-Square andSquare-to- Round Tubular Joints, ref. /8/, where d = depthand width of brace; Dw = depth and width of chord. Thesestress concentration factors may be used together with theD-curve.

The following stress concentration factors may be used ford/Dw = 1.0, in lieu of a more detailed analysis forcalculation of hot spot stress:

Axial: 1.90 In-plane bending: 4.00 Out-of plane bending: 1.35These stress concentration factors should be used togetherwith the F-curve.

$ **(Design should be performed such that fatigue crackingfrom the root is less likely than from the toe region. Thereason for this is that a fatigue crack at the toe can befound by in-service inspection while a fatigue crackstarting at the root can not be discovered before the crackhas grown through the weld. Thus the design of the weldgeometry should be performed such that the fatigue life forcracks starting at the root is longer than the fatigue life ofthe toe. Figure 2.10-2 can be used for evaluation ofrequired penetration. The lack of penetration, (2ai),obtained from this figure may be further reduced by afactor of 0.80 in order to obtain a recommended designvalue for avoidance of fatigue cracking from the root. Thenotation used is explained by Figure 2.10-1.

It should be added that it is difficult to detect internaldefects by NDE in fillet/partial penetration welds. Suchconnections should therefore not be used in structuralconnections of significant importance for the integrity.

!"# $

%&$$

DET NORSKE VERITAS

$8(**(

0

0.2

0.4

0.6

0.8

1

1.2

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1D

L

W

S

K

W

S

tp = 50 mmtp = 25 mmtp = 12 mmtp = 6mm

Weld toe failure

Weld root failure

$80(*&&07

!"# $

%&$$

DET NORSKE VERITAS

9A bolted joint connection subjected to dynamic loadingshould be designed with pretensioned bolts. Thepretension should be high enough to avoid slipping afterrelevant loss of pretension during service life. Connectionswhere the pretensioned bolts are subjected to dynamicaxial forces should be designed with respect to fatiguetaking into account the stress range in the bolts resultingfrom tension and compression range. The stress range inthe bolts may be assessed based on e.g. Maskindeler 2,ref. /23/, or Systematic Calculation of High Duty BoltedJoints, ref. /26/.

*Welds in pipelines are normally made with a symmetricweld groove with welding from the outside only. Thetolerances are rather strict compared with other structuralelements with eccentricity less than 0.1*t or maximum 3mm. (t = wall thickness) The fabrication of pipelines alsoimplies a systematic and standardised NDE of the root areawhere defects are most critical. Provided that the sameacceptance criteria are used for pipelines with larger wallthickness as for that used as reference thickness (25 mm),a thickness exponent k = 0 may be used for hot spot at theroot and k = 0.15 for the weld toe. Provided that theserequirements are fulfilled, the detail at the root side may beclassified as F1 with SCF = 1.0, ref. Table 2.12-1. The F-curve and SCF = 1.0 may be used for welding ontemporary backing, ref. Table 2.12-1.

Reference is made to Table 2.12-1 for other tolerances andwelding from both sides.

For weld grooves that are not symmetrical in shape a stressconcentration due to maximum allowable eccentricityshould be included. This stress concentration factor can beassessed based on the following analytical expression

+=0.5

0tD

expt

)-3(1SCF

where:

0 = 0.1t is misalignment inherent in the S-N data.

This stress concentration factor can also be used forfatigue assessments of the weld toes, ref. also Table2.12-1.

!"# $ '

%&$$

DET NORSKE VERITAS

-&

(**

Description

Welding Geometry and hotspot

Tolerance requirement S-Ncurve

Thicknessexponent k

SCF

min(0.1t, 3 mm) F1 0.00 1.0Single side

Hot spot > min(0.1t, 3 mm) F3 0.00 1.0

min(0.1t, 2 mm) F 0.00 1.0Single sideon backing

Hot spot > min(0.1t, 2 mm) F1 0.00 1.0

Single sideHot spot

)mm4,t15.0min( D 0.15 Eq.(2.12.1)

Double side

Hot spot

)mm4,t15.0min( D 0.15 Eq.(2.12.1)

*&00

,

From detailed finite element analysis of structures it maybe difficult to evaluate what is nominal stress to be usedtogether with the S-N curves, as some of the local stressdue to a detail is accounted for in the S-N curve.

In many cases it may therefore be more convenient to usean alternative approach for calculation of fatigue damagewhen local stresses are obtained from finite elementanalysis.

It is realised that it is difficult to calculate the notch stressat a weld due to a significant scatter in local weldgeometry and different types of imperfections. This scatteris normally more efficiently accounted for by use of anappropriate S-N curve. In this respect it should also bementioned that the weld toe region has to be modelledwith a radius in order to obtain reliable results for thenotch stress.

If a corner detail with zero radius is modelled thecalculated stress will approach infinity as the element sizeis decreased to zero. The modelling of a relevant radiusrequires a very fine element mesh, increasing the size of

!"# $

%&$$

DET NORSKE VERITAS

the computer model. In addition, a proper radius to be usedfor the analysis will likely be a matter of discussion.

Hence, for design analysis a simplified numericalprocedure is used in order to reduce the demand for largefine mesh models for the calculation of SCF factors:

The stress concentration or the notch factor due to theweld itself is included in the S-N curve to be used, theD-curve.

The stress concentration due to the geometry effect ofthe actual detail is calculated by means of a fine meshmodel using shell elements (or solid elements),resulting in a geometric SCF factor.

This procedure is denoted the hot spot method.

It is important to have a continuous and not too steep,change in the density of the element mesh in the areaswhere the hot spot stresses are to be analysed.

The geometry of the elements should be evaluatedcarefully in order to avoid errors due to deformed elements(for example corner angles between 60 and 120 andlength/breadth ratio less than 5 are recommended).The size of the model should be so large that the calculatedresults are not significantly affected by assumptions madefor boundary conditions and application of loads.

-&

The stress range at the hot spot of tubular joints should becombined with the T-curve.

Analysis based on thick shell elements may be used. Inthis case, the weld is not included in the model. The hotspot stress may be determined as for welded connections.

More reliable results are obtained by including the weld inthe model. This implies use of three-dimensional elements.Here the Gaussian points, where stresses are calculated,may be placed 1.0 from the weld toe (r = radius ofconsidered tubular and t = thickness). The stress at thispoint may be used directly in the fatigue assessment.

8&

The stress range at the hot spot of welded connectionsshould be combined with S-N curve D. The C-curve maybe used if machining of the weld surface to the basematerial is performed. Then the machining has to beperformed such that the local stress concentration due tothe weld is removed.

The aim of the finite element analysis is not normally tocalculate directly the notch stress at a detail, but tocalculate the geometric stress distribution in the region atthe hot spot such that these stresses can be used as a basisfor derivation of stress concentration factors. Reference ismade to Figure 2.13-1 as an example showing the stressdistribution in front of an attachment (A-B) welded to a

plate with thickness . The notch stress is due to thepresence of the attachment and the weld. The aim of thefinite element analysis is to calculate the stress at the weldtoe (hot spot) due to the presence of the attachment,denoted geometric stress, hot spot. The stress concentrationfactor due to this geometry effect is defined as,

inalnom

spothotSCF =

Thus the main emphasis of the finite element analysis is tomake a model that will give stresses with sufficientaccuracy at a region outside that effected by the weld. Themodel should have a fine mesh for extrapolation ofstresses back to the weld toe in order to ensure asufficiently accurate calculation of SCF.

FEM stress concentration models are generally verysensitive to element type and mesh size. By decreasing theelement size the FEM stresses at discontinuities willapproach infinity. It is therefore necessary to set a lowerbound for element size and use an extrapolation procedureto the hot spot to have a uniform basis for comparison ofresults from different computer programs and users. Onthe other hand, in order to pick up the geometric stress, g,increase properly, it is important that the stress referencepoints in t/2 and 3t/2 (see Figure 2.13-1) are not inside thesame element. This implies that element sizes of the orderof the plate thickness are to be used for the modelling. Ifsolid modelling is used, the element size in way of the hotspot may have to be reduced to half the plate thickness incase the overall geometry of the weld is included in themodel representation.

Element stresses are normally derived at the gaussianintegration points. Depending on element type it may benecessary to perform several extrapolations in order todetermine the stress at the location representing the weldtoe. In order to preserve the information of the direction ofprincipal stresses at the hot spot, component stresses are tobe used for the extrapolation. When shell elements areused for the modelling and the overall geometry of theweld is not included in the model, the extrapolation shallbe performed to the element intersection lines. If the(overall) weld geometry is included in the model (3Dmodel), the extrapolation is related to the weld toe asshown in Figure 2.13-1. If 8 node shell elements are usedthe hot spot is considered to be at the element intersectionline.

Two different definitions for hot spot stresses are used:

1. The stress is derived by extrapolating the stress to theweld toe (intersection line).

2. The stress at 0.5t from the considered hot spot

The stresses are first extrapolated from thegaussian integration points to the plate surface. A furtherextrapolation to the line A - B is then conducted. The finalextrapolation of component stresses is carried out as a

!"# $ )

%&$$

DET NORSKE VERITAS

linear extrapolation of surface stresses along line A - B at adistance t/2 and 3t/2 from either the weld toe, oralternatively the element intersection line (where t denotesthe plate thickness). Having determined the extrapolatedstress components at the hot spot, the principal stresses areto be calculated and used for the fatigue evaluation.

Some comments on element size are given in Appendix 4,Commentary.

It is recommended to perform a verification of theprocedure on a detail that is S-N classified and that issimilar in geometry and loading to that being analysed. Ifthe verification analysis comes out with a different SCF(SCF Verification) than that inherent in the S-N detail, ref.e.g. Table 2.3-1, a resulting stress concentration factor canbe calculated as

onVerificati

12.3TableNSAnalysis SCF

SCFSCFSCF =

where

SCFS-N Table 2.3-1 = Stress concentration in the S-Ndetail as derived by the hot spotmethod, see Table 2.3-1.

SCFAnalysis = Stress concentration factor for theanalysed detail.

It should be noted that the hot spot concept can not be usedfor fatigue checks of cracks starting from the weld root offillet/partial penetration welds. The weld should bechecked separately considering the stresses in the welditself, ref. section 2.2.3.

: The hot spot stress is derived directly from thefinite elements at a distance 0.5t from the weld toe using20 node solid elements or 0.5t from the intersection lineusing 8 node shell elements.

It is also here recommended to perform a verification ofthe procedure on a detail that is S-N classified and that issimilar in geometry and loading to that being analysed. Ifthe verification analysis comes out with a different SCF(SCF Verification) than that inherent in the S-N detail, ref.e.g. Table 2.3-1, a resulting stress concentration factor canbe calculated as shown in the example above.

&1*

61*

+ !"# $

%&$$

DET NORSKE VERITAS

*0

,

The long term stress range distribution may be presentedas a two-parameter Weibull distribution

=

h

qexp)Q(

where

Q = probability for exceedance of the stress range h = Weibull shape parameter

q = Weibull scale parameter is defined from the stressrange level, 0, as

1/h0

0)n(ln

q =

0 is the largest stress range out of n0 cycles..

When the long-term stress range distribution is definedapplying Weibull distributions for the different loadconditions, and a one-slope S-N curve is used, the fatiguedamage is given by

)hm

qa

TD md0 +=

where

Td = design life in seconds

h = Weibull stress range shape distribution parameter

q = Weibull stress range scale distribution parameter

0 = average zero-crossing frequency

)hm

+ = gamma function. Values of the gamma

function are listed in Table 2.14-1.

Use of one slope S-N curves leads to results on the safeside for calculated fatigue lives (with slope of curve atN < 106-107 cycles).For other expressions for fatigue damage see Appendix 4 ,Commentary.

-&"

2:;

0.600.610.620,630,640,650,660,670,680,690,700,710,720,730,740.750.76

120.000104.40391.35080.35871.04863.11956.33150.49145.44241.05837.23433.88630.94228.34426.04424.00022.178

0.770.780.790.800.810.820.830.840.850.860.870.880.890.900.910.920.93

20.54819.08717.77216.58615.51414.54213.65812.85312.11811.44610.82910.263 9.741 9.261 8.816 8.405 8.024

0.940.950.960.970.980.991.001.011.021.031.041.051.061.071.081.091.10

7.671 7.342 7.035 6.750 6.483 6.234 6.000 5.781 5.575 5.382 5.200 5.029 4.868 4.715 4.571 4.435 4.306

Design charts for steel components in air and in seawaterwith cathodic protection are shown in Figure 2.14-1 andFigure 2.14-2 respectively. These charts have been derivedbased on the two slopes S-N curves given in this RP. Thecorresponding numerical values are given in Table 2.14-2and Table 2.14-3.

These design charts have been derived based on anassumption of an allowable fatigue damage = 1.0 during108 cycles (20 years service life and an average waveperiod of 6.3 sec). For design with other allowable fatiguedamages, , the allowable stress from the design chartsshould be reduced by factors derived from Table 2.14-2and Table 2.14-3 for conditions in air and in seawaterwith cathodic protection respectively.

The stresses derived here correspond to the referencethickness. For thickness larger than the referencethickness, an allowable extreme stress range during 108cycles may be obtained as

N

=

tt ref

tref0,t0,

where

k = thickness exponent, see section 2.3.1 and Table2.3-1

0,tref = allowable stress as derived from Table 2.14-2 -Table 2.14-5.

!"# $ /

%&$$

DET NORSKE VERITAS

0

100

200

300

400

500

600

700

800

0.50 0.60 0.70 0.80 0.90 1.00 1.10 1.20

:HLEXOOVKDSHSDUDPHWHUK

$

O

O

R

Z

D

E

O

H

V

W

U

H

V

V

U

D

Q

J

H

L

Q

0

3

D

F

F1

F3

G

W1

W2

W3

E D , T C2 C1 C B2 B1

$ !"# $

%&$$

DET NORSKE VERITAS

0.0

100.0

200.0

300.0

400.0

500.0

600.0

700.0

800.0

0.50 0.60 0.70 0.80 0.90 1.00 1.10 1.20

:HLEXOOVKDSHSDUDPHWHUK

$

O

O

R

Z

D

E

O

H

V

W

U

H

V

V

U

D

Q

J

H

L

Q

0

3

D

D,T

E

F

F1

F3

G

W1

W2

W3

C2 C1 C B2 B1

!"# $

%&$$

DET NORSKE VERITAS

-&

!

"#

0.50 0.805 0.810 0.816 0.821 0.826 0.831 0.835 0.8390.30 0.688 0.697 0.706 0.715 0.723 0.730 0.737 0.742

0.10 0.497 0.512 0.526 0.540 0.552 0.563 0.573 0.581

-&'((

*

!

"#

0.50 0.821 0.831 0.840 0.847 0.853 0.857 0.861 0.8640.30 0.713 0.729 0.743 0.753 0.762 0.769 0.773 0.778

0.10 0.535 0.558 0.577 0.592 0.604 0.613 0.619 0.623

!"# $

%&$$

DET NORSKE VERITAS

0&

4Fracture mechanics may be used for fatigue analyses assupplement to S-N data.

Fracture mechanics is recommended for use in assessmentof acceptable defects, evaluation of acceptance criteria forfabrication and for planning in-service inspection.

The purpose of such analysis is to document, by means ofcalculations, that fatigue cracks, which might occur duringservice life, will not exceed the crack size correspondingto unstable fracture. The calculations should be performedsuch that the structural reliability by use of fracturemechanics will not be less than that achieved by use of S-N data. This can be achieved by performing the analysisaccording to the following procedure:

Crack growth parameter C determined as mean plus 2standard deviation.

A careful evaluation of initial defects that might bepresent in the structure when taking into account theactual NDE inspection method used to detect cracksduring fabrication.

Use of geometry functions that are on the safe side. Use of utilisation factors similar to those used when

the fatigue analysis is based on S-N data.

As crack initiation is not included in the fracturemechanics approach, shorter fatigue life is normallyderived from fracture mechanics than by S-N data.

In a case that the results from fracture mechanics analysescannot directly be compared with S-N data it might berecommended to perform a comparison for a detail whereS-N data are available, in order to verify the assumptionsmade for the fracture mechanics analyses.

The initial crack size to be used in the calculation shouldbe considered in each case, taking account of experiencedimperfection or defect sizes for various weldments,geometries, access and reliability of the inspection method.For surface cracks starting from transitions betweenweld/base material, a crack depth of 0.5 mm (e.g. due toundercuts and microcracks at bottom of the undercuts)may be assumed if other documented information aboutcrack depth is not available.

It is normally, assumed that compressive stresses do notcontribute to crack propagation. However, for weldedconnections containing residual stresses, the whole stressrange should be applied. Only stress components normal tothe propagation plane need to be considered.

The Paris equation may be used to predict the crackpropagation or the fatigue life:

( )mCdNda

=

where

K = Kmax - Kmin

N = Number of cycles to failure

a = crack depth. It is here assumed that the crackdepth/length ratio is low (less than 1:5).

C, m = material parameters, see BS 7910, ref. /7/.

The stress intensity factor K may be expressed as:

agK =

where

= nominal stress in the member normal to the crack

g = factor depending on the geometry of the memberand the crack.

See BS 7910, ref. /7/, for further guidelines related tofatigue assessment based on fracture mechanics.

!"# $

%&$$

DET NORSKE VERITAS

4*2>&0&

,It should be noted that improvement of the toe will notimprove the fatigue life if fatigue cracking from the root isthe most likely failure mode. The considerations made inthe following are for conditions where the root is notconsidered to be a critical initiation point. The effect fromdifferent improvement methods as given in the followingcan not be added.

8*&0By weld profiling in this section is understood profiling bymachining or grinding as profiling by welding only is notconsidered to be an efficient mean to improve fatiguelives.

In design calculations, the thickness effect may be reducedto an exponent 0.15 provided that the weld is profiled byeither machining or grinding to a radius of approximatelyhalf the plate thickness, (T/2 with stress direction as shownin Figure 4.3-1, B).Where weld profiling is used, the fatigue life can beincreased by a factor of 2.

As an alternative to including a factor 2 to increase fatiguelife one may take account of a reduced local stressconcentration factor achieved by weld profiling. A reducedlocal stress due to weld profiling can be obtained asfollows.

When weld profiling is performed, a reduced hot spotstress can be calculated as

%HQGLQJ0HPEUDQHUHGXFHG/RFDO

+=

where and are derived from equations (4.2.2) and(4.2.3) respectively.

5.025.0 )/()(tan17.047.0 $% +=

5.025.0 )/()(tan13.060.0 $% +=

For description of geometric parameters see Figure 4.2-1.

The membrane part and the bending part of the stress haveto be separated from the local stress as

%HQGLQJ0HPEUDQH/RFDO

+=

where

Membrane = Membrane stress

Bending = Bending stress

If a finite element analysis of the considered connectionhas been performed, the results from this can be useddirectly to derive membrane stress and bending stress.

For cruciform joints and heavy stiffened tubular joints itmay be assumed that the hot spot stress is mainly due tomembrane stress.

For simple tubular joints it may be assumed that the hotspot stress in the chord is due to bending only.

The reduced local stress in equation (4.2.1) is to be usedtogether with the same S-N curves as the detail isclassified for without weld profiling. (It is assumed thatR/T = 0.1 without weld profiling for a plate thickness T =25 mm).(The fatigue life can not be increased by a factor of 2 at thesame time as the hot spot stress is reduced due to weldprofiling).

Weld ProfilingT

R

8*

,

Where local grinding of the weld toes below any visibleundercuts is performed the fatigue life may be increasedby a factor of 2. In addition the thickness effect may bereduced to an exponent k = 0.20. Reference is made toFigure 4.3-1. Grinding a weld toe tangentially to the platesurface, as at A, will produce only little improvement infatigue strength. To be efficient, grinding should extendbelow the plate surface, as at B, in order to remove toedefects. Grinding is normally carried out by a rotary burr.The treatment should produce a smooth concave profile atthe weld toe with the depth of the depression penetratinginto the plate surface to at least 0.5 mm below the bottomof any visible undercut (see Figure 4.3-1). The grindingdepth should not exceed 2 mm or 10% of the platethickness, whichever is smaller.

!"# $

%&$$

DET NORSKE VERITAS

In general grinding has been used as an efficient methodfor reliable fatigue life improvement after fabrication.Grinding also improves the reliability of inspection afterfabrication and during service life. However, experienceindicates that it may be a good design practice to excludethis factor at the design stage. The designer is advised toimprove the details locally by other means, or to reducethe stress range through design and keep the possibility offatigue life improvement as a reserve to allow for possibleincrease in fatigue loading during the design andfabrication process, see also OS-C101 Design of SteelStructures, section 6.

It should also be noted that if grinding is required toachieve a specified fatigue life, the hot spot stress is ratherhigh. Due to grinding a larger fraction of the fatigue life isspent during the initiation of fatigue cracks, and the crackgrows faster after initiation. This implies use of shorterinspection intervals during service life in order to detectthe cracks before they become dangerous for the integrityof the structure.

Depth of grinding shouldbe 0.5mm below bottomof any visible undercut.BA

T

,(

-4,The fatigue life may be improved by a factor 2 by TIGdressing.

Due to uncertainties regarding quality assurance of thewelding process, this method may not be recommended forgeneral use at the design stage.

' ?*The fatigue life may be improved by a factor of 4 bymeans of hammer peening.

However, the following limitations apply:

Hammer peening should only be used on memberswhere failure will be without substantialconsequences, ref. OS-C101 Design of SteelStructures, section 6.

Hammer peening may only be used when minimumload of predominant load ranges is compressive orzero.

Overload in compression must be avoided, becausethe residual stress set up by hammer peening will bedestroyed.

Peening tip must be small enough to reach weld toe.

Due to uncertainties regarding quality assurance of theprocess, this method may not be recommendable forgeneral use at the design stage.

!"# $ '

%&$$

DET NORSKE VERITAS

' 61An extended fatigue life is considered to be acceptable andwithin normal design criteria if the calculated fatigue lifeis longer than the total design life times the Fatigue DesignFactor.

Otherwise an extended life may be based on results fromperformed inspections throughout the prior service life.Such an evaluation should be based on:

1) Calculated crack growth. Crack growth characteristics; i. e. crack length/depth

as function of time/number of cycles (this depends ontype of joint, type of loading, and possibility forredistribution of stress).

2) Reliability of inspection method used. Elapsed time from last inspection performed.

It is recommended to use Eddy Current or MagneticParticle Inspection for inspection of surface cracksstarting at hot spots.

For welded connections that are ground and inspected forfatigue cracks the following procedure may be used forcalculation of an elongated fatigue life. Provided thatgrinding below the surface to a depth of approximately 1.0mm is performed and that fatigue cracks are not found by adetailed Magnetic Particle inspection of the considered hotspot region at the weld toe, the fatigue damage at this hotspot may be considered to start again at zero. If a fatiguecrack is found, a further grinding should be performed toremove any indication of this crack. If more than 10% ofthe thickness is removed by grinding, the effect of this onincreased stress should be included when a new fatigue lifeis assessed. In some cases as much as 30% of the platethickness may be removed by grinding before a weldrepair is resorted to. This depends on type of joint, loadingcondition and accessibility for a repair.

It should be noted that fatigue cracks growing from theweld root of fillet welds can hardly be detected by NDT.Also, the fatigue life of such regions can not be improvedby grinding of the surface.

It should also be remembered that if renewal of one hotspot area is performed by local grinding, there are likelyother areas close to the considered hot spot region that arenot ground and that also experience a significant dynamicloading. The fatigue damage at this region is the same asearlier. However, also this fatigue damage may bereassessed taking into account:

the correlation with a ground neighbour hot spotregion that has not cracked

an updated reliability taking the reliability ofperformed in-service inspections into account asdiscussed above.

!"# $

%&$$

DET NORSKE VERITAS

@>

,Large uncertainties are normally associated with fatiguelife assessments. Reliability methods may be used toillustrate the effect of uncertainties on probability of afatigue failure. An example of this is shown in Figure6.1-1 based on mean expected uncertainties for a jacketdesign from Reliability of Calculated Fatigue Lives ofOffshore Structures, ref. /17/.

The calculated probability of failure is sensitive toassumptions made for the analysis. However, calculatedreliability values in a relative sense. Using Figure 6.1-1 inthis way, it might be concluded that a design modificationto achieve a longer calculated fatigue life is an efficientmean to reduce probability of a fatigue failure, ref. Figure6.1-1.

The effect of scatter in S-N data may be illustrated byFigure 6.1-2 where the difference between calculated lifeis shown for mean S-N data and design S-N data (which isdetermined as mean minus 2 standard deviations).

0.000000001

0.00000001

0.0000001

0.000001

0.00001

0.0001

0.001

0.01

0.1

1

0 0.2 0.4 0.6 0.8 1 1.2

&DOFXODWHGIDWLJXHGDPDJH

3

U

R

E

D

E

L

O

L

W

\

R

I

I

D

W

L

J

X

H

I

D

L

O

X

U

H

*&&0

020406080

100120140160180200

200 250 300 350 4000D[LPXPDOORZDEOHVWUHVVUDQJHLQ03D

&

D

O

F

X

O

D

W

H

G

I

D

W

L

J

X

H

O

L

I

H

L

Q

\

H

D

U

V

Design S-N data (mean minus 2 standard deviations) Mean S-N data

6

"

" $ )

A0$$$

DET NORSKE VERITAS

) /1/ Classification Note No 30.7 Fatigue

Assessment of Ship Structures. Det NorskeVeritas 1998.

/2/ Efthymiou, M.: Development of SCFFormulae and Generalised InfluenceFunctions for use in Fatigue Analysis. RecentDevelopments in Tubular Joint Technology,OTJ88, October 1988, London.

/3/ Smedley, S. and Fischer, P.: StressConcentration Factors for Ring-StiffenedTubular Joints. : Proceedings of the FirstInternational Offshore and Polar EngineeringConference, Edinburgh, August 1991. pp.239-250. Publ by Int Soc of Offshore andPolar Engineers (ISOPE), P.O.Box 1107,Golden, CO, USA.

/4/ Lotsberg, I., Cramer, E., Holtsmark, G.,Lseth, R., Olaisen, K. and Valsgrd, S.:Fatigue Assessment of Floating ProductionVessels. BOSS97, July 1997.

/5/ Eurocode : Design of steel structures. Part 1-1:General rules and rules for buildings. February1993.

/6/ Guidance on Design, Construction andCertification. HSE. February 1995.

/7/ BS7910:1999. Guidance on Methods forAssessing the Acceptability of Flaws inFusion Welded Structures. BSI. Draft 1999.

/8/ Soh, A. K. and Soh, C. K.: StressConcentrations in T/Y and K Square-to-Squareand Square-to- Round Tubular Joints. Journal ofOffshore Mechanics and Arctic Engineering.August 1992, Vol. 114.

/9/ Gulati, K. C., Wang, W. J. and Kan, K. Y.: AnAnaltyical study of Stress ConcentrationEffects in Multibrace Joints under CombinedLoading. OTC paper no 4407, Houston, May1982.

/10/ Gurney,T.R.: Fatigue Design Rules for weldedSteel Joints, the Welding Institute ResearchBulletin. Volume 17, number 5, May 1976.

/11/ Gurney, T. R.: The Basis for the RevisedFatigue Design Rules in the Department ofEnergy Offshore Guidance Notes. Paper No55.

/12/ Berge, S.: Effect of Plate Thickness in FatigueDesign of Welded Structures. OTC Paper no4829. Houston, May 1984.

/13/ Buitrago, J. and Zettlemoyer, N.: Fatigue ofWelded Joints Peened Underwater. 1997OMAE, ASME 1997.

/14/ Stacey, A., Sharp, J. V. and Nichols, N. W.:Fatigue Performance of Single-sidedCircumferential and Closure Welds inOffshore Jacket Structures. 1997 OMAE,ASME 1997.

/15/ Pilkey, W. D.: Petersons Stress ConcentrationFactors. Second Edition. John Wiley & Sons.1997.

/16/ Haagensen, P. J., Drgen, A., Slind, T. andrjaster, O.: Prediction of the Improvementin Fatigue Life of welded Joints Due toGrinding, Tig Dressing, Weld Shape Controland Shot Peening. Steel in Marine Structures,edited by C. Noorhook and J. deBack ElsevierScience Publishers B.V., Amsterdam, 1987,pp. 689-698.

/17/ Lotsberg, I., Fines, S. and Foss, G.:Reliability of Calculated Fatigue Lives ofOffshore Structures, Fatigue 84, 2nd Int.Conf. on Fatigue and Fatigue Thresholds, 3-7September 1984. Birmingham.

/18/ Haagensen, P. J.,Slind, T. and rjaster, O.:Scale Effects in Fatigue Design Data forWelded and Unwelded Components. Proc.Ninth Int. Conf. On Offshore Mechanics andArctic Engineering. Houston, February 1990.

/19/ Berge, S., Eide, O., Astrup, O. C., Palm, S.,Wstberg, S., Gunleiksrud, . and Lian,B.:Effect of Plate Thickness in Fatigue ofWelded Joints in Air ans in Sea Water. Steelin Marine Structures, edited by C. Noorhookand J. deBack Elsevier Science PublishersB.V., Amsterdam, 1987, pp. 799-810.

/20/ Razmjoo, G. R.: Design Guidance on Fatigueof Welded Stainless Steel Joints. OMAE1995.

/21/ Madsen, H. O., Krenk, S. and Lind, N. C.(1986) Methods of Structural Safety, Prentice-Hall, Inc., NJ.

/22/ Marshall, P. W.: API Provisions for SCF, S-N,and Size-Profile Effects. OTC Paper no 7155.Houston, May 1993.

/23/ Walen, . .: Maskindeler 2, Tapir, NTNU(In Norwegian).

/24/ Buitrago, J., Zettlemoyer, N. and Kahlish, J.L.: Combined Hot-Spot Stress Procedures for

+ !"# $

%&$$

DET NORSKE VERITAS

Tubular Joints. OTC Paper No. 4775.Houston, May 1984.

/25/ Lotsberg, I.: Stress Concentration Factors atCircumferential Welds in Tubulars. Journal ofMarine Structures, January 1999.

/26/ VDI 2230 Part 1: Systematic Calculation ofHigh Duty Bolted Joints. Verein DeutscheIngenieure, August 1988.

/27/ Van Wingerde, A.M., Packer, J.A.,Wardenier, J., Dutta, D. and Marshall, P.:Proposed Revisions for Fatigue Design ofPlanar Welded Connections made of HollowStructural Sections. Paper 65 in "TubularStructures V," Ed. M.G. Coutie and G.Davies, 1993 E & FN spon.

/28/ DNV Offshore Standard. OS-C101 Design ofSteel Structures

/29/ Lotsberg, I., and Rove, H.: StressConcenteration Factors for Butt Welds inStiffened Plates.OMAE, ASME 2000.

/30/ IIW. Fatigue Design of Welded Joints andComponents. Recommendations of IIW JointWorking Group XIII-1539-96/XV-845-96.Edited by A. Hobbacher Abington Publishing,1996, The International Institute of Welding.

Recommended Practice DNV-RP-C203Appendix 1October 2001

DET NORSKE VERITAS

39

APPENDIX 1 CLASSIFICATION OF STRUCTURAL DETAILSTable 1 Non-welded details

Notes on potential modes of failureIn plain steel, fatigue cracks will initiate at the surface, usually either at surface irregularities or at corners ofthe cross-section. In welded construction, fatigue failure will rarely occur in a region of plain material sincethe fatigue strength of the welded joints will usually be much lower. In steel with boltholes or other stressconcentrations arising from the shape of the member, failure will usually initiate at the stress concentration.The applied stress range shall include applicable stress concentration factors arising from the shape of themember.

Detailcategory Constructional details Description Requirement

B1 1.

2.

1. Rolled or extruded platesand flats

2. Rolled sections

1. to 2.:- Sharp edges, surface and

rolling flaws to beimproved by grinding.

- For members that canacquire stressconcentrations due to rustpitting etc. curve C isrequired.

B2 3. 3. Machine gas cut orsheared material with nodrag lines

3.- All visible signs of edge

discontinuities should beremoved.

- No repair by weld refill.- Re-entrant corners (slope

Recommended Practice DNV-RP-C203Appendix 1

October 2001

DET NORSKE VERITAS

40

Table 2 Bolted connectionsDetailcategory Constructional details Description Requirement

C1 1., 2. 1. Unsupported one-sided connections shallbe avoided or elseeffects of eccentricitiesshall be taken intoaccount whencalculating stresses.

2. Beam splices or boltedcover plates.

1. and 2.:- Stresses to be calculated

in the gross section.- Bolts subjected to

reversal forces in shearshall be designed as aslip resistant connectionand only the membersneed to be checked forfatigue.

3. 3. Bolts and threadedrods in tension.

F1 Cold rolled threads withno following heattreatment like hotgalvanising

W3 Cut threads

3.:- Tensile stresses to be

calculated using thetensile stress area of thebolt.

- For preloaded bolts, thestress-range in the boltdepends upon the levelof preload and thegeometry of theconnection, see e.g.Maskindeler 2, ref./23/.

Recommended Practices DNV-RP-C203 41Appendix 1October 2001

DET NORSKE VERITAS

Table 3 Continuous welds essentially parallel to the direction of applied stressNotes on potential modes of failure.With the excess weld material dressed flush, fatigue cracks would be expected to initiate at weld defectlocations. In the as welded condition, cracks may initiate at start-stop positions or, if these are not present, atweld surface ripples.

General comments

a) Backing strips

If backing strips are used in these joints, they must be continuous. If they are attached by welding, such weldsmust also comply with the relevant joint classification requirements (note particularly that tack welds, unlesssubsequently ground out or covered by a continuous weld, would reduce the joint to class F)

b) Edge distance

An edge distance criterion exists to limit the possibility of local stress concentrations occurring at unweldededges as a result, for example, of undercut, weld spatter, or accidental overweave in manual fillet welding (seealso notes in Table 7). Although an edge distance can be specified only for the width direction of anelement, it is equally important to ensure that no accidental undercutting occurs on the unwelded corners of,for example cover plates or box girder flanges. If undercutting occurs it should subsequently be groundsmooth.

Detailcategory Constructional details Description Requirement

1. Automatic butt weldscarried out from bothsides. If a specialistinspectiondemonstrates thatlongitudinal welds arefree from significantflaws, category B2may be used.

C 1.

2.2. Automatic fillet welds.

Cover plate ends shallbe verified using detail5. in Table 8

1. and 2.:

- No start-stop positionis permitted exceptwhen the repair isperformed by aspecialist andinspection carried outto verify the properexecution of the repair.

Recommended Practice DNV-RP-C203Appendix 1

October 2001

DET NORSKE VERITAS

42

Detailcategory Constructional details Description Requirement

3. Automatic fillet or buttwelds carried out fromboth sides butcontaining stop-startpositions.

C1 3.

4.

4. Automatic butt weldsmade from one sideonly, with a backingbar, but without start-stop positions.

4.:- When the detail

contains start-stoppositions use categoryC2

5. Manual fillet or buttwelds.

C2

6. Manual or automaticbutt welds carried outfrom one side only,particularly for boxgirders

6.:- A very good fit

between the flange andweb plates is essential.Prepare the web edgesuch that the root faceis adequate for theachievement of regularroot penetration without brake-out.

C2 7. Repaired automatic ormanual fillet or buttwelds

7.:- Improvement methods

that are adequatelyverified may restorethe original category.

Recommended Practices DNV-RP-C203 43Appendix 1October 2001

DET NORSKE VERITAS

Table 4 Intermittent welds and welds at cope holesDetailcategory Constructional details Description Requirement

E 1. 1.Stitch or tack weldsnot subsequentlycovered by acontinuous weld

1.:- Intermittent fillet

weld with gap ratiog/h 2.5.

F 2. 2.Ends of continuouswelds at copeholes.

2.:- Cope hole not to be

filled with weldmaterial.

3. 3.Cope hole andtransverse buttweld.

3.:- For butt weld in

material with copehole advice onfatigue assessmentmay be found inCN 30.7.

- The SCF (or K-factor) from CN30.7 may be usedtogether with the Ccurve.

Recommended Practice DNV-RP-C203Appendix 1

October 2001

DET NORSKE VERITAS

44

Table 5 Transverse butt welds, welded from both sidesNotes on potential modes of failureWith the weld ends machined flush with the plate edges, fatigue cracks in the as-welded condition normallyinitiate at the weld toe, so that the fatigue strength depends largely upon the shape of the weld overfill. If theoverfill is dressed flush, the stress concentration caused by it is removed, and failure is then associated withweld defects.

Design stresses

In the design of butt welds that are not symmetric about the root and are not aligned, the stresses must includethe effect of any eccentricity (see section 2.5 to 2.9).

With connections that are supported laterally, e.g. flanges of a beam that are supported by the web,eccentricity may be neglected.

Detailcategory Constructional details Description Requirement

C1 1.

2.

3.

41

4

1

1. Transverse splices inplates flats and rolledsections

2. Flange splices in plategirders.

3. Transverse splices inplates or flats taperedin width or in thicknesswhere the slope is notgreater than 1:4.

1. and 2.:- Details 1. and 2. may

be increased toCategory C when highquality welding isachieved and the weldis proved free fromsignificant defects bynon-destructiveexamination (it isassumed that this isfulfilled by inspectioncategory I).

1., 2. and 3.:- All welds ground flush

to plate surface parallelto direction of thearrow.

- Weld run-off pieces tobe used andsubsequently removed,plate edges to beground flush indirection of stress.

- All welds welded inhorizontal position inshop.

Recommended Practices DNV-RP-C203 45Appendix 1October 2001

DET NORSKE VERITAS

Detailcategory Constructional details Description Requirement

D 4.

5.

6.

41

4 1

4.Transverse splices inplates and flats.

5.Transverse splices inrolled sections orwelded plate girders

6.Transverse splices inplates or flats taperedin width or inthickness where theslope is not greaterthan 1:4.

4., 5. and 6.:- The height of the weld

convexity to be notgreater than 10% of theweld width, with smoothtransitions to the platesurface.

- Welds made in flatposition in shop.

- Weld run-off pieces tobe used andsubsequently removed,plate edges to be groundflush in direction ofstress.

Recommended Practice DNV-RP-C203Appendix 1

October 2001

DET NORSKE VERITAS

46

Detailcategory Constructional details Description Requirement

E 7.

41

4 1

7.Transverse splices inplates, flats, rolledsections or plategirders made at site.(Detail category Dmay be used forwelds made in flatposition at sitemeeting therequirements under4., 5. and 6.)

7.:- The height of the weld

convexity to be notgreater than 20% of theweld width.

- Weld run-off pieces tobe used andsubsequently removed,plate edges to be groundflush in direction ofstress.

8.

F1 0.16hr

F311.0

hr

8.Transverse splicebetween plates ofunequal width, withthe weld ends groundto a radius.

8.:- The stress concentration

has been accounted forin the joint classification.

- The width ratio H/hshould be less than 2.

Recommended Practices DNV-RP-C203 47Appendix 1October 2001

DET NORSKE VERITAS

Table 6 Transverse butt welds, welded from one sideNotes on potential modes of failureWith the weld ends machined flush with the plate edges, fatigue cracks in the as-welded condition normallyinitiate at the weld toe, so that the fatigue strength depends largely upon the shape of the weld overfill. If theoverfill is dressed flush, the stress concentration caused by it is removed, and failure is then associated withweld defects. In welds made on permanent backing strip, fatigue cracks most likely initiate at the weldmetal/strip junction.

Design stresses

In the design of butt welds that are not symmetric about the root and are not aligned, the stresses must includethe effect of any eccentricity (see section 2.5 to 2.9).

With connections that are supported laterally, e.g. flanges of a beam that are supported by the web,eccentricity may be neglected.

Detailcategory Constructional details Description Requirement

W3 1. 1.Butt weld madefrom one side onlyand withoutbacking strip.

1.:With the root proved freefrom defects larger than 1-2mm (in the thicknessdirection) by non-destructivetesting, detail 1 may berecategorised to F3 (it isassumed that this is fulfilledby inspection category I). If itis likely that larger defectsmay be present after theinspection the detail may bedowngraded from F3 based onfatigue life calculation usingfracture mechanics. Theanalysis should then be basedon a relevant defect size.

F 2. 2.Transverse buttweld on apermanent backingstrip without filletwelds.

G 3. 3.Transverse buttweld on a backingstrip fillet weldedto the plate.

Recommended Practice DNV-RP-C203Appendix 1

October 2001

DET NORSKE VERITAS

48

Table 7 Welded attachments on the surface or the edge of a stressed memberNotes on potential modes of failureWhen the weld is parallel to the direction of the applied stress, fatigue cracks normally initiate at the weldends. When the weld is transverse to direction of stressing, cracks usually initiate at the weld toe; forattachments involving a single, as opposed to a double, weld cracks may also initiate at the weld root. Thecracks then propagate into the stressed member. When the welds are on or adjacent to the edge of the stressedmember the stress concentration is increased and the fatigue strength is reduced; this is the reason forspecifying an edge distance in some of this joints (see also note on edge distance in table Table 3).

Detailcategory Constructional details Description Requirement

1.

E l 50mm

F 50 < l 120mmF1 120 < l 300mmF3 l > 300mm

1.Welded longitudinalattachment

1. The detail category isgiven for:- Edge distance 10mm- For edge distance

< 10mm the detailcategory shall bedowngraded with oneSN-curve

2.

D150mmr ,

Wr

31

F

31

Wr

61

F1

61

Wr

101

F3

101

Wr

161

G

161

Wr

251

2.Gusset plate with aradius welded to theedge of a plate or beamflange.

Recommended Practices DNV-RP-C203 49Appendix 1October 2001

DET NORSKE VERITAS

Detailcategory Constructional details Description Requirement

3.

G l 150mm

W1 150 < l 300mm

W2 l > 300mm

3.Gusset plate welded tothe edge of a plate orbeam flange.

4.

t

5.

6.

E t 12mm

F t > 12mm

4.Transverse attachmentswith edge distance 10mm

5.Vertical stiffener weldedto a beam or a plat