ssc-367 - Ship Structure Committee
ssc-367 - Ship Structure Committee
ssc-367 - Ship Structure Committee
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
SSC-<strong>367</strong> -<br />
FATIGUE TECHNOLOGY<br />
ASSESSMENT AND STRATEGIES<br />
FOR FATIGUE AVOIDANCE IN<br />
MARINE STRUCTURES<br />
This document has been approved<br />
for public release and sale; its<br />
distribution is urdimited<br />
SHIP STRUCTURE<br />
1993<br />
COMMITTEE<br />
—.——<br />
.——--———-<br />
__—————
SSC-<strong>367</strong> ‘<br />
FATIGUE TECHNOLOGY<br />
ASSESSMENT AND STRATEGIES<br />
FOR FATIGUE AVOIDANCE IN<br />
MARINE STRUCTURES<br />
This document has been approved<br />
for public release and srilq its<br />
distribution is unlimited<br />
SHIP STRUCTURE<br />
1993<br />
COMMITTEE
SHIP STRUCTURF COMMllTFF<br />
The SHIP STRLfCTURE COMMllTEE is constituted to prosecute a research program to improve the hull structures of ships and other<br />
marine structures by an extension of knowledge pertaining to design, materials, and methods of construction.<br />
RADM A. E. Henn, USCG (Chairman)<br />
Chief, Office of Marine Safety, Security<br />
and Environmental Protection<br />
U. S. Coast Guard<br />
Mr. Thomas 1-1,Peirce Mr. H. T, Hailer Dr. Donald Liu<br />
Marine Research and Development Associate Administrator for <strong>Ship</strong>- Seniqr Vice President<br />
Coordinator building and <strong>Ship</strong> Operations American Bureau of <strong>Ship</strong>ping<br />
Transportation Development Center Maritime Administration<br />
Transport Canada<br />
Mr. Alexander Malakhoff Mr. Thomas W. Allen<br />
CDR Stephen E. Sharpe, USCG<br />
Director, Structural Integrity<br />
Engineering Offmer (N7) Executtie Dkector<br />
Subgroup (SW 05P) Military Sealift Command<br />
Shi <strong>Structure</strong> <strong>Committee</strong><br />
Naval Sea Systems Command<br />
U. i . Coast Guard<br />
CONTRACTING OFFICER TEC HNICAL REPRESENTATIVE<br />
Mr. WNiam J. Slekierka<br />
SEA05P4<br />
Naval Sea Systems Command<br />
~Hl P STRUCTURF<br />
SUBCOMMIITFF<br />
The SHIP STRUCTURE SUBCOMMllTEE acts for the <strong>Ship</strong> <strong>Structure</strong> <strong>Committee</strong> on technical matters by providing technical<br />
coordination for determinating the goals and objectives of the program and by evaluating and interpreting the results in terms of<br />
structural design, construction, and operation.<br />
AMERICAN BUREAU OF SHIPPING NAVALSEA SYSTEMS COM MAND TRANSPORT CANADA<br />
Mr. Stephen G. Amtson (Chairman) Dr. Robert A Sielski<br />
Mr. John Grinstead<br />
Mr. John F. ConIon<br />
Mr. Charles L Null Mr. Ian Bayly<br />
Dr. John S. Spenmr Mr. W. Thomas Packard Mr. David L. Stocks<br />
Mr. Glenn M. M.he Mr. Allen H. Engle Mr. Peter Timonin<br />
MILITARY SFAI IFT COMMAN~<br />
E ADMINISTRATION U. S, COAST GUARD<br />
Mr. Rolwt E. Van Jones Mr. Frederick Seibold CAPT T. E. Thompson<br />
Mr. Rickard A. Anderson Mr. Norman O. Hammer<br />
CAPT W. E. Colburn, Jr.<br />
Mr. Michael W. Touma<br />
Mr. Chao H. Lin Mr. Rubin Scheinberg<br />
Mr. Jeffrey E. Beach Dr. Walter M, Maclean Mr. Il. Paul Cojeen<br />
SHIP STRUCTURE SUBCOMMlmEE LIAISON MEMBERS<br />
U. S. COAST GUARD ACADEMY<br />
NATI ONAL ACAD~Y OF SCIENCES -<br />
LCDR Bruce FL Mustain<br />
Mr. Alexander<br />
B. Stavovy<br />
U.S. MFRCHAN T MARINF ACADFMY<br />
Dr. C. B. Kim<br />
u. S, NAVAL ACADEMY<br />
Dr. Rarrrswar Bhattacharyya<br />
STATE UNIVERSITY OF NEW YORK<br />
~<br />
Dr. W. R. Porter<br />
SOCIETY OF NAVAL ARCHITECTS AND<br />
MARINE ENGINEERS<br />
~TION#.&AEC~~EMY OF SCIENCES -<br />
MARINE STRUCTURES<br />
Mr. Peter M. Palermo<br />
WFI IIING RFSFARCHCOUNC II<br />
Dr. Ma_tin Prager<br />
~ L INSTITUTE<br />
Mr. Alexander<br />
D. Wilson<br />
DEPARTMENTOF NATIONAL DEFENCE - CANADA<br />
Dr. William<br />
Sandberg<br />
Dr. Neil G. Pegg<br />
OFFICE OF NAVAI RFSEARG H<br />
Dr. Yapa D. S. Rajapaske
MemberAgencies:<br />
United States Coast Guard<br />
Naval Sea Systems Command<br />
Maritime Administration<br />
American Bureau of <strong>Ship</strong>ping<br />
Miiitaty Sealift Command<br />
Transport Canada<br />
~<br />
c<br />
<strong>Ship</strong><br />
<strong>Structure</strong><br />
<strong>Committee</strong><br />
An Interagency Advisory<strong>Committee</strong><br />
May 17, 1993<br />
AddressCorrespondence to:<br />
Executive Director<br />
<strong>Ship</strong><strong>Structure</strong> <strong>Committee</strong><br />
U.S.CoastGuard(G-Ml/R)<br />
2100SecondStreet, S.W.<br />
Washington, D.C.20593-0001<br />
PH:(202)267-0003<br />
FAX:(202)267-4677<br />
SSC-<strong>367</strong><br />
SR-1324<br />
FATIGUE TECHNOLOGY ASSESSMENT AND STRATEGIES FOR FATIGUE<br />
AVOIDANCE IN MARINE STRUCTURES<br />
This report synthesizes the state-of-the–art in fatigue<br />
technology as it relates to the marine field. Over the years<br />
more sophisticated methods have been developed to anticipate the<br />
life cycle loads on structures and more accurately predict the<br />
failure modes. As new design methods have been developed and<br />
more intricate and less robust structures have been built it has<br />
become more critical than ever that the design tools used be the<br />
most effective for the task. This report categorizes fatigue<br />
failure parameters, identifies strengths and weaknesses of the<br />
available design methods, and recommends fatigue avoidance<br />
strategies based upon variables that contribute to the<br />
uncertainties of fatigue life. The report concludes with<br />
recommendations for further research in this field.<br />
Gmdvwm<br />
A. E. HENN<br />
Rear Admiral, U.S. Coast Guard<br />
Chairman, <strong>Ship</strong> <strong>Structure</strong> <strong>Committee</strong>
—.--—<br />
1. Report No.<br />
2. Gowotnmrnt AcenszIon Na.<br />
T*chnical<br />
3. Rompiqnt’s Cataiog No.<br />
I?cport Documentation Fagc<br />
4. T,tle and $ubti~i=<br />
I<br />
FATIGUEDESIGNPROCEDURES<br />
5. ROport Dot.<br />
June1992<br />
6. Perfoming Organl*at~en coda<br />
7. Authds) 8. Porfeming Orgonitatien Report No.<br />
CuneytC.Capanoglu<br />
s.performing Or~mizotion Nemo rnd Addro~c<br />
SR-1324<br />
12. Spontsrinq A90ncy Namo and Addr**s<br />
Is,suppi~mentary<br />
Not*~<br />
EARLAND WRIGHT<br />
180HowardStreet<br />
SanFrancisco, CA 94105<br />
10. Work Unit No. (TRAIS)<br />
11.co~timet or Grant No.<br />
DTCG23-88-C-2~29<br />
1s.TYP- et Rwport ad Period Covormd<br />
<strong>Ship</strong><strong>Structure</strong> <strong>Committee</strong><br />
U.S.CoastGuard(G-M)<br />
2100 SecondStreet, SW<br />
Washington,DC 20593 1d.Sponsoring A9cnty Cod.<br />
Sponsoredbythe<strong>Ship</strong><strong>Structure</strong> <strong>Committee</strong>anditsmembersagencies.<br />
FiialRepo~<br />
G-M<br />
16. Abstract<br />
Thisreponprovidesanup-todateassessmentoffatigue technology, directed specifically toward<br />
themarineindustry. A comprehensive overviewoffatigue analysis anddesign,aglobalreviewof<br />
fatioueincluding rulesandregulations andcurrentpractices, anda fatigueanalysis anddesign<br />
criteria, areprovidedasa generalguideline tofatigue assessment.A detailed discussion ofall<br />
fatigueparametersisgroupedunderthreeanalysis blocks:<br />
●<br />
●<br />
●<br />
Fatiguestressmodel,covering environmental forces, structuresponseandloading,<br />
stress<br />
responseamplitudeoperations (RAOS)andhot-spot stresses<br />
Fatiguestresshistory modelcoveringlong-term distribution ofenvironmental loading<br />
Fatigueresistance ofstructures anddamageassessmentmathodolo@es<br />
Theanalysesanddesignparametersthataffectfatigue assessmentarediscussedtogetherwith<br />
uncertainties andresearchgaps,toprovideabasisfordeveloping strategies forfatigue avoidance.<br />
Additional in-depthdiscussions ofwave environmt,stre<strong>ssc</strong>oncentration factors,etc.are<br />
presentedintheappendixes.<br />
17. Key Word,<br />
Assessmentoffatigue technology,<br />
fatioue stressmdels,fatigue<br />
stresshistory models,$atigue<br />
resistance, fatioueparamatefs<br />
andfatigue avoidancestrategies<br />
19. S* Curity ciassil. (of this toport)<br />
FormDOT F 1700.7(8G721<br />
UnCJSssifiarj<br />
.+<br />
I<br />
~. SOcutity CII<br />
lB. Distribution $tot~~t<br />
Available from:<br />
National TechnicalInformation Serv.<br />
U.S.DepartmentofCommerce<br />
Springfield, VA 22151<br />
1.(*f thi c p~o) 21.No. of Pegms<br />
Unclassified<br />
R=ptoductian O{ compktad<br />
page authorized<br />
194 EXCI.<br />
Appendixes<br />
22. Pri =~
-. —---- —-...–. .<br />
COMVEflSION<br />
FACTORS<br />
Apprmimat, Conversions toMeiricMmsures<br />
m<br />
— C4<br />
Approxinmta Conversionsfrom Mstric Mcasutes<br />
WlwnVW Mnow Multiply b~<br />
—-<br />
C4<br />
Symbol<br />
Wlmt Yom Know<br />
Muhirlf by<br />
la Find<br />
LENGTH<br />
,,,,,’<br />
in<br />
h<br />
Vd<br />
mi<br />
in2<br />
~2<br />
d<br />
mi 2<br />
02<br />
lb<br />
tap<br />
rkap<br />
N a=<br />
c<br />
P1<br />
qt<br />
LENGTH<br />
“2.5<br />
30<br />
0.9<br />
I .e<br />
AREA<br />
squwc inches 6.6<br />
square feet 0.0s<br />
cquam yards OJJ<br />
mqusmmiles 2.6<br />
■clmm 0.4<br />
Oumcea 26<br />
-~ 0,45<br />
Cflai tam 0.9<br />
12000 lb}<br />
cenlimslera<br />
centimwmc<br />
mmers<br />
kitmneters<br />
mquwe Centilrwws<br />
Wplwa<br />
matO1!J<br />
cqume inmrs<br />
aqunre kilcmetems<br />
hectwea<br />
cm<br />
cm<br />
m<br />
km<br />
*2<br />
~2<br />
~2<br />
km2<br />
hm<br />
MASS {wtight] —- =<br />
VOLUME<br />
t*mspnns 5<br />
Iab!eapwms 15<br />
lfukt Cslnces 30<br />
cups 0.24<br />
pints 0.4?<br />
qumta 0.95<br />
gallons 3.6<br />
cubic hat 0.03<br />
cubic yards 0,16<br />
grcnls<br />
kil~ann<br />
tmmes<br />
milliliters<br />
milliliters<br />
milliliters<br />
!itels<br />
liters<br />
Iilers<br />
liters<br />
cubic mews<br />
cubic rnmers<br />
n<br />
kg<br />
t<br />
ml<br />
ml<br />
ml<br />
1<br />
I<br />
)<br />
I<br />
~3<br />
~?<br />
—:<br />
-<br />
—<br />
—<br />
—-<br />
—.<br />
—.<br />
m<br />
.<br />
a<br />
c-<br />
m<br />
m<br />
— .<br />
—-<br />
—-<br />
—<br />
—-<br />
—<br />
—<br />
—<br />
—<br />
m<br />
C4<br />
0<br />
Oi<br />
m<br />
c-<br />
w<br />
mm<br />
cm<br />
km<br />
m<br />
km<br />
9<br />
kg<br />
t<br />
n-d<br />
I<br />
t<br />
I<br />
m’<br />
~3<br />
miIlinw[ws 0.04<br />
canlimmlera 0,4<br />
maters 3.3<br />
Mstara 1.1<br />
hikmwters 0.6<br />
AREA<br />
sqtmr.s intimaters 0.t6<br />
equmrs matam 1.2<br />
squsre kilowlem 0.4<br />
hactamt ~10,OCSIn?) 2.6<br />
MASS ~waight]<br />
gram3 0.033<br />
wagramm 2.2<br />
tmlms{loal kg) 1.1<br />
VOLUME<br />
millilitcms 0.03<br />
liters 2.1<br />
Iilars 1.06<br />
Iiwm 0.26<br />
cubic maters 35<br />
cubic nwlers 1.3<br />
TEMPERATURE [mwet]<br />
inches<br />
inclces<br />
fear<br />
yank<br />
miles<br />
in<br />
in<br />
rt<br />
yd<br />
mi<br />
sqcme inches in2<br />
equw. ymcls<br />
&<br />
squw* milas<br />
mi2<br />
mxas<br />
amca<br />
pOLsur,<br />
dmrt<br />
m<br />
Icmm<br />
fluid ounces<br />
piwCc<br />
quacta<br />
02<br />
lb<br />
fl Oz<br />
m<br />
qt<br />
gctkma<br />
gsl<br />
cubic led fta<br />
cubic yalds<br />
vd3<br />
TEMPERATWIE<br />
‘F Fehmnheit 5/9 (atCar<br />
tempamture scdmacling<br />
321<br />
{exmtj<br />
Celsius<br />
ten-mma twe<br />
.1 in : 2.54 lEIIacllyl. Fw olher ● xacl converswm and more #i?4a#Id tables. see NBS hl,sc. P, b!J1.286,<br />
Unts d We,ghts ❑wltdeasures, PTece$2.25, SD Catatog No. C13.102O6.<br />
“c<br />
Celsius 9/6 (then<br />
tampefmclra add 32)<br />
Mwanhail *F<br />
tsmpmatum<br />
‘F<br />
‘f 32 9e.6 2r2<br />
-40 0 40 80 I20 160 zm<br />
l,, ,,, ;.:l’ll,l{<br />
l’;<br />
I 4 # I * I 1 *<br />
1 1 1 1 I<br />
-:: -20 0 20 40 60 “ eo ion<br />
3-r @c
FATIGUE TECHNOLOGY Assessment AND DEVELOPMENT OF<br />
STRATEGIES FOR FATIGUE AVOIDANCE IN MARINE STRUCTURES<br />
FINAL REPORT<br />
CONTENTS<br />
Abstract<br />
Contents<br />
List of Figures<br />
Common Terms<br />
i<br />
ii<br />
ix<br />
xi<br />
1. INTRODUCTION<br />
1.1 Background<br />
1.2 Objectives<br />
1.3 scope<br />
1-1<br />
1-2<br />
1“3<br />
2* OVERVIEW OF FATIGUE<br />
2.1 Fatigue Phenomena<br />
2.2 Fatigue Analysis<br />
2.2.1 Analysis Sequence<br />
2.2.2 Analysis Methods<br />
2.3 Significance of Fatigue Failure<br />
2.4 Fatigue Failure Avoidance<br />
2-1 .<br />
2-3<br />
2-7<br />
2-8<br />
3. FATIGUE DESIGN AND ANALYSES PARAMETERS<br />
3.1 Review of Fatigue Design Parameters<br />
3.1.1 Design Parameters<br />
3.1.2 Fabrication and Post Fabrication Parameters<br />
3.1.3 In-Service Parameters<br />
3.2 Review Of Fatigue Analysis Parameters<br />
3.2.1 Fatigue Analysis Criteria<br />
3.2.2 InteractingParameters<br />
3.2.3 Stress Model Parameters<br />
3.2.4 Stress History Model Parameters<br />
3.2.5 Fatigue Damage Computation Parameters<br />
3-1<br />
3-7<br />
ii
4. GLOBAL REVIEW OF FATIGUE<br />
4.1 Applicable Analysis Methods<br />
4.1.1 Background<br />
4.1.2 Simplified Analysis and Design Methods<br />
4.1.3 Detailed Analyses and Design Methods<br />
4.1.4 Other Methods<br />
4-1<br />
4.2 Fatigue Rules and Regulations 4-16<br />
4.2.1 Applicable Methods<br />
4.2.2 SCFS, S-N Curves and Cumulative Damage<br />
4.2.3 Fatigue Analysis Based on Fracture Mechanics<br />
4.3 Current Industry Practices 4-23<br />
4.3.1 Ordinary Designs<br />
4.3.2 Specialized Designs<br />
4.4 Sensitivity of Fatigue Parameters 4-25<br />
4.5 Fatigue Design and Analysis Criteria 4-26<br />
4.5.1 Basis for the Preparationof Criteria<br />
4.5.2 Applicable Software<br />
4.5.3 Fatigue Versus Other Design and Scheduling Requirements<br />
5. FATIGUE STRESS MODELS<br />
5.1 Review of Applicable Modeling Strategies 5-1<br />
5.1.1 Modeling Strategies<br />
5.1.2 Comparison of <strong>Structure</strong>s<br />
.<br />
5.2 Floating Marine <strong>Structure</strong>s 5-4<br />
5.2.1 <strong>Ship</strong> <strong>Structure</strong>s<br />
5.2.2 Stationary Marine <strong>Structure</strong>s<br />
5.2.3 Overview and Recommendations<br />
5.3 Bottom-SupportedMarine <strong>Structure</strong>s 5-16<br />
5.3.1 Load or HydrodynamicsModel<br />
5.3.2 Mass Model<br />
5.3.3 Motions Model and Analyses Techniques<br />
5.3.4 Stiffness Model<br />
5.3.5 Overview and Recommendations<br />
5.4 Development of Hot Spot Stresses 5-25<br />
5.4.1 Nominal Stresses and Stress RAOS<br />
5.4.2 Stress Concentration Factors and Hot Spot Stresses<br />
5.4.3 Empirical Equations<br />
5.4.4 Illustrationof a T-Joint SCFS<br />
5.4.5 Overview and Recommendations<br />
. . .<br />
111<br />
( .-. ,)
6. FATIGUE STRESS HISTORY MODELS<br />
6,1 Determination of Fatigue Environments<br />
6.1.1 Data Sources<br />
6.1.2 Wave and Wind Spectra<br />
6.1.3 Scatter Diagram<br />
6.1.4 Directionalityand Spreading<br />
6-1<br />
6.2 Stress Spectrum 6-10<br />
6.2.1 Stress RAOS<br />
6.2.2 Response Analysis<br />
6.2,3 Uncertainties and Gaps in Stress Spectrum Development<br />
6.2.4 Decompose into Stress Record<br />
6.3 Time-Domain Analyses 6-14<br />
6.3.1 Stress Statistics<br />
6.3.2 70 PercentileSpectra<br />
6.4 Overview and Recommendations 6-15<br />
7. FATIGUE DAMAGE ASSESSMENT<br />
7.1 Basic Principles of Fatigue Damage Assessment<br />
7.2 S-N Curves<br />
7.2.1 Design Parameters<br />
7.2.2 Fabrication and Post-FabricationParameters<br />
7.2.3 EnvironmentalParameters<br />
7.3 Fatigue Damage Computation<br />
7.3.1 Miner’s Rule<br />
7.3.2 Alternative Rules<br />
7.4 Stress History and Upgraded Miner’s Rule<br />
7.4.1 Background<br />
7.4.2 Miner’s Rule IncorporatingRainflow Correction<br />
7.4.3 Other Alternatives<br />
7.5 Overview and Recommendations<br />
7.5.1 Application of S-N Curves<br />
7.5.2 Fatigue Damage Computation<br />
7-1<br />
7-2<br />
7-11<br />
7-14<br />
7-19<br />
iv<br />
-7
8. FATIGUE DUE TO VORTEX SHEDDING<br />
8.1 Vortex Shedding Phenomenon<br />
8.1.1 Background<br />
8.1.2 Vortex Induced Vibrations (VIV)<br />
8.2 Analyses and Design For Vortex Shedding<br />
8.2.1 Susceptibilityto Vortex Shedding<br />
8.2.2 VIV Response and Stresses<br />
8.3 Fatigue Damage Assessment<br />
8.4 Methods of Minimizing Vortex Shedding Oscillations<br />
8.5 Recommendations<br />
8-1<br />
8-3<br />
8-5<br />
8-6<br />
8-6<br />
9. FATIGUE AVOIDANCE STRATEGY<br />
9.1 Review of Factors Contributing to Failure<br />
9.2 Basic Fatigue Avoidance Strategies<br />
9.2,1 Basic Premises<br />
9.2.2 Fatigue Avoidance Strategies<br />
9.3 Fatigue Strength ImprovementStrategies<br />
9.3.1 Fabrication Effects<br />
9.3.2 Post-FabricationStrength Improvement<br />
9.3,3 Comparison of Strength ImprovementStrategies<br />
9.4 Fatigue Analysis Strategies<br />
9.4.1 Review of Uncertainties,Gaps and Research Needs<br />
9.4.2 Recent Research Activities<br />
9.4.3 Cost-EffectiveAnalysis Strategies<br />
9.5 Recommendations<br />
9.5.1 Research Priorities<br />
9.5.2 Rules and Regulations<br />
10. REFERENCES<br />
9-1<br />
9-2<br />
9-7<br />
9-14<br />
9-22<br />
1o-1<br />
v
APPENDICES<br />
A. REVIEW OF OCEAN ENVIRONMENT<br />
A.1<br />
A.2<br />
A.3<br />
IRREGULARWAVES<br />
PROBABILITYCHARACTERISTICSOF WAVE SPECTRA<br />
A.2.1 CharacteristicFrequencies and Periods<br />
A.2.2 CharacteristicWave Heights<br />
WAVE SPECTRA FORMULAS<br />
A.3.1 Bretschneiderand ISSC Spectrum<br />
A.3.2 Pierson-MoskowitzSpectrum<br />
A.3.3 JONSWAP and Related Spectra<br />
A.3.4 Scott and Scott-WiegelSpectra<br />
A-1<br />
A-5<br />
A-n<br />
A.4<br />
SELECTING A WAVE SPECTRUM A-16<br />
A.4.1 Wave Hindcasting<br />
A.4.2 Direct Wave Measurements<br />
A.5<br />
A.6<br />
A.7<br />
A.8<br />
WAVE SCATTER DIAGRAM<br />
WAVE EXCEEDANCE CURVE<br />
WAVE HISTOGRAM AND THE<br />
EXTREME VALUES AND THE<br />
A-19<br />
A-21<br />
RAYLEIGH DISTRIBUTION A-22 .<br />
WEIBULL DISTRIBUTION A-22<br />
A.9<br />
A.1O<br />
WIND ENVIRONMENT<br />
A-23<br />
A.9.1 Air Turbulence, Surface Rouqhness and Wind Profile<br />
A.9.2 Applied, Mean and Cyclic Velocities<br />
A.9.3 Gust Spectra “<br />
REFERENCES A-28<br />
vi<br />
(-’<br />
7
B. REVIEW OF LINEAR SYSTEM RESPONSE TO RANDOM EXCITATION<br />
B.1 GENERAL<br />
B.1.l Introduction<br />
6.1.2 Abstract<br />
B.1.3 Purpose<br />
B.2 RESPONSE TO RANDOM WAVES<br />
B.2.1 Spectrum Analysis Procedure<br />
B.2.2 Transfer Function<br />
6.2.3 Wave Spectra<br />
B.2.4 Force Spectrum<br />
B.2.5 White Noise Spectrum<br />
6.3 EXTREME RESPONSE<br />
6.3.1 Maximum Wave Height Method<br />
B.3.2 Wave Spectrum Method<br />
B.4 OPERATIONAL RESPONSE<br />
B.4.1 Special Family Method<br />
B.4.2 Wave Spectrum Method<br />
B-1<br />
B-2<br />
B-15<br />
B-17<br />
c. STRESS CONCENTRATION FACTORS<br />
C*1 OVERVIEW<br />
C*2 STRESS CONCENTRATION FACTOR EQUATIONS<br />
C.2.1 Kuang with Marshall Reduction<br />
C.2.2 Smedley-Wordsworth<br />
C.3 PARAMETRIC STUDY RESULTS<br />
C.3.1 Figures<br />
C.3.2 Tables<br />
C*4 FINITE ELEMENT ANALYSES RESULTS<br />
C.3.1 Column-GirderConnection<br />
C*5 REFERENCES<br />
c-1<br />
c-4<br />
C-6<br />
C-15<br />
C-17<br />
vii
D. VORTEX SHEDDING AVOIDANCE AND FATIGUE DAMAGE COMPUTATION<br />
NOMENCLATURE<br />
i<br />
D.1<br />
D.2<br />
D.3<br />
D.4<br />
D.5<br />
D*6<br />
D.7<br />
D*8<br />
0.9<br />
INTRODUCTION D-1<br />
VORTEX SHEDDING PARAMETERS D-2<br />
SUSCEPTIBILITYTO VORTEX SHEDDING D-7<br />
D.3.1 In-Line Vortex Shedding<br />
D.3.2 Cross-Flow Vortex Shedding<br />
D.3.3 Critical Flow Velocities<br />
AMPLITUDES OF VIBRATION D-9<br />
D.4.1 In-Line Vortex Shedding Amplitudes<br />
D.4.2 Cross-Flow Vortex Shedding Amplitudes<br />
STRESSES DUE TO VORTEX SHEDDING D-13<br />
FATIGUE LIFE EVALUATION D-14<br />
EXAMPLE PROBLEMS D-17<br />
D.7.1 Avoidance of Wind-Induced Cross-Flow Vortex Shedding<br />
D.7.2 Analysis for Wind-Induced Cross-Flow Vortex Shedding<br />
METHODS OF MINIMIZING VORTEX SHEDDING OSCILLATIONS D-23 —.<br />
D.8.1 Control of Structural Design<br />
D.8.2 Mass and Damping<br />
D.8.3 Devices and Spoilers<br />
REFERENCES D-27<br />
. . .<br />
Vlll
LIST OF FIGURES<br />
LIST OF FIGURES<br />
(cent.)<br />
FIGURE<br />
TITLE<br />
9-1 Typical Methods to Improve Fatigue Strength<br />
9-2 Typical Weld Toe Defects and Corrective Measures<br />
9-3 Fatigue Life ImprovementDue to Weld Toe Abrasive<br />
Water Jet Erosion<br />
9-4 Comparison of Fatigue Strength ImprovementTechniques<br />
9-5 Summary of Relevant Research Activities<br />
.—. . . . ...—
COMMON TERMS<br />
USED<br />
IN FATIGUE AND IN THIS REPORT<br />
BTM<br />
: Bottom turret mooring system for a tanker. Can<br />
be permanent or disconnectable.<br />
CAPEX<br />
: Capital expenditures incurred prior to<br />
structure commissioning and beginning<br />
operation.<br />
CATHODIC PROTECTION<br />
: An approachto reduce material corrosive action<br />
by making it the cathode of an electrolytic<br />
cell. This is done by utilizing sacrificial<br />
anodes (i.e. couplingwith more electropositive<br />
metal) or impressed current.<br />
COMPLEX JOINT<br />
: An intersection of several members, having a<br />
subassemblageof componentmembers. Applicable<br />
to a column-to-pontoon joint of a<br />
semisubmersible or a large leg joint of a<br />
platform containing stiffened bulkheads,<br />
diaphragms and other tubulars.<br />
CRUCIFORM JOINT<br />
: A transverse load carrying joint made up two<br />
plates welded on to either side of a<br />
perpendicularplate utilizing full penetration<br />
welds.<br />
DYNAMIC AMPLIFICATION FACTOR : The maximum dynamic and static load ratios,<br />
(DAF)<br />
such as the DAF applicable to base shear or<br />
overturningmoment.<br />
HEAT AFFECTED ZONE (HAZ)<br />
: The area of parent plate material susceptible<br />
to material degradationdue to welding process.
HOT-SPOT STRESS<br />
.<br />
The hot-spot stress is the peak stress in the<br />
immediate vicinity of a structural<br />
discontinuity,such as the stiffener edge or a<br />
cutout. On a tubular joint, the hot-spot<br />
stress usually occurs at the weld toe of the<br />
incoming tubular (brace) or the main tubular<br />
(chord).<br />
FATIGUE LIFE<br />
.<br />
●<br />
The number of stress cycles that occur before<br />
failure, typically corresponding to either<br />
first discernible surface cracking (Nl) or the<br />
first occurrence of through thickness<br />
cracking.<br />
FATIGUE STRENGTH<br />
●<br />
✎<br />
The stress range corresponding<br />
to<br />
a number of<br />
cycles at which failure occurs.<br />
FPSO<br />
.<br />
Floating production, storage<br />
and offloading<br />
tanker.<br />
IRREGULARITYFACTOR<br />
●<br />
✎<br />
The ratio of mean crossings with positive<br />
slopes to the number of peaks or valleys in the<br />
stress history.<br />
KEULEGAN-CARPENTERNUMBER, Kc :<br />
A parameter used to define the flow properties<br />
around a cylinder. Equal to the product of the<br />
amplitude of velocity and oscillation period,<br />
divided by the cylinder diameter.<br />
MEAN ZERO-CROSSING PERIOD :<br />
The mean zero-crossing period is the average<br />
time between successive wave crossings with a<br />
positive slope (up-crossing) of the zero axis<br />
in a time history.<br />
xii
MODELING ERROR (Xme)<br />
: Typically defined as the ratio of actual<br />
behavior of the structure to the one predicted<br />
by the model. It is often used to assess the<br />
accuracy of excitational loads, motions, and<br />
stresses.<br />
MODELING UNCERTAINTY<br />
NARROW-BAND LOADING<br />
: The random component of the modeling error,<br />
x, and defined by its coefficient of<br />
v%iation, (C.O.V.)X .<br />
me<br />
: The stress cycles are readily identifiable,<br />
making the choice of counting method of stress<br />
cycles immaterial.<br />
NOMINAL STRESS<br />
: The nominal stress is the stress obtained by<br />
dividing the member generalized forces (forces<br />
and moments) by member section properties<br />
(cross-sectionalarea and section modulus).<br />
OPEX<br />
: Operating expenditures due to maintenance,<br />
inspection, repairs as well as cost of fuel,<br />
variables,personnel,etc. during the life of a<br />
structure.<br />
PLASMA DRESSING<br />
: Application of plasma arc welding technique to<br />
remelt the weld toe (similarto TIG dressing)<br />
POST WELD HEAT TREATMENT<br />
(PWHT)<br />
: A procedure of heating a welded joint to<br />
relieve residual fabrication stresses.<br />
Typically, the joint is heated to 1076 1150°F<br />
(580-620”C), held at that temperaturefor about<br />
an hour for each one inch (2.5 rein/mm)<br />
thickness, and cooled in air.<br />
xiii
QA/Qc<br />
: Quality Assurance/QualityControl<br />
Quality assurance generally refers to the<br />
procedures and methods put into effect to<br />
ensure quality a priori, while quality control<br />
generally refers to reviews and checks afterthe-fact<br />
to implement corrective measures, as<br />
necessary.<br />
: The term random waves is used to characterize<br />
the irregular sea surface and associatedwater<br />
particle kinematics that occur in the ocean.<br />
Analytically random waves are represented as a<br />
summation of sinusoidal waves of different<br />
heights, periods, phases and directions.<br />
REGULAR WAVES<br />
: Regularwaves are unidirectionaland associated<br />
water particle kinematics and sea surface<br />
elevations are periodic.<br />
S-N CURVE<br />
: The S-N curves define the fatigue strength of a<br />
detail/joint by representing test data in an<br />
empirical form to establish a relationship<br />
between stress ranges and the number of cycles<br />
of stress range for fatigue failure.<br />
SEA STATE<br />
: An oceanographicenvironmentwith a wave height<br />
range characterized as a stationary random<br />
process for a specific duration.<br />
SIGNIFICANT WAVE HEIGHT<br />
: A statistic typically used to characterize the<br />
wave heights in a sea state. It is defined as<br />
the average height of the heighestone-third of<br />
all the individual waves present in a sea<br />
state.<br />
xiv
SIMPLE JOINT<br />
: An intersection of two or more structural<br />
members. Also applicableto an intersectionof<br />
unstiffenedor ring-stiffenedcylinders.<br />
STEADY STATE<br />
: Generally refers to the periodic response of a<br />
dynamic system after initial starting<br />
transients have decayed to negligible<br />
amplitude.<br />
STRESS CONCENTRATION FACTOR : The ratio of hot-spot stress to the nominal<br />
(SCF)<br />
stress (in neighborhoodof hot-spot) and often<br />
max. mized at geometric discontinueties.<br />
STRIP THEORY<br />
: App” ied to various strip methods to determine<br />
the<br />
bod<br />
hydrodynamic loadings on “ ong slender<br />
es and can account for the effect of<br />
diffracted and radiated waves.<br />
TIG DRESSING<br />
: Tungsten-inert-gas dressing is applied to<br />
remelt the weld toe material to reduce both the<br />
SCF by minimizingdiscontinuitiesand to remove<br />
defects such as slag inclusions.<br />
TRANSFER FUNCTION<br />
: A transfer function defines the unitized<br />
structural response as a function of frequency<br />
(eg ratio of structural response to the wave<br />
amplitude applicable for each frequency).<br />
WELD TOE<br />
: The point of intersection of the weld profile<br />
and parent plate.<br />
WIDE-BAND LOADING<br />
: The smaller stress cycles are interspersed<br />
among larger stress cycles, making the<br />
definitionof stress cycle more difficult. The<br />
use of different counting methods will result<br />
in different fatigue damage predictions.<br />
xv
1.<br />
INTRODUCTION<br />
1.1<br />
BACKGROUND<br />
The detailed design of a structure focuses largely on sizing the<br />
structurescomponentmembers and on developingthe details to resist<br />
extreme functionaland environmentalloads. The analysisand design<br />
to resist extreme loading conditions is intended primarily to<br />
prevent material yield and buckling failures; the details are also<br />
chosen to help prevent fatigue failures due to cyclic loading.<br />
,-.<br />
The use of proven details and selection of steel with material<br />
properties resistingpropagationof defects are longstandingdesign<br />
practices. Analysis and design to ensure that fatigue life is<br />
substantiallyin excessof the design life becamegenerally accepted<br />
in the late 1960s. Initial simplistic analysis methods have<br />
gradually become more sophisticated. Oceanographic data collected<br />
over the last twenty years now allow better definition of wind and<br />
wave data over many parts of the world. Several test programs have<br />
allowed comparison of actual and analytically computed loads on<br />
marine structures. Laboratorytest data anddata from structures in<br />
servicenow allowbetterdefinitionof defect (crack)propagation in<br />
an ocean environment.<br />
Although engineers have progressed beyond simplified deterministic<br />
analyses, occasionallyventuring into full probabilistic analysis,<br />
substantialuncertaintiesstill are associatedwith fatigue analysis<br />
and design. Fatigue life may change dramatically with a small<br />
change in any of many variables,requiringthat the fatigue analysis<br />
and design of a marine structure be conducted as a series of<br />
parametric studies. The results of these studies, used to upgrade<br />
fatigue-sensitiveareas/details of the structure, allow development<br />
of a design that will provide a satisfactory level of confidence<br />
against fatigue failure.<br />
Review of past fatigue failures shows that it is often difficult to<br />
determine whether a failure was due to poor design, material<br />
1-1
imperfections, fabrication defects, improper inspection or<br />
maintenance, unpredicted loads or, more likely, a combination of<br />
these interactingvariables. As the complexityof marinestructures<br />
increases,betterunderstandingof the variablescontributingto the<br />
integrity of structure components and the global response of the<br />
structure becomes very important. Although several excellent<br />
documents on fatigue are available, most address fatigue design of<br />
either ship or offshore platform structures (References1.1 through<br />
1.8). Thus the engineer may have difficulty in assessing the<br />
significance of fatigue within the context of overall design of<br />
marine structures. It is also difficult to evaluatethe sensitivity<br />
and interactionof variables affecting fatigue life or the relative<br />
uncertaintiesthat are built in. The UEGReconnnendations(Reference<br />
1.8), although applicable to only tubular joints, provides a<br />
detailed discussion of various design requirements and code<br />
recommendations.<br />
Fatigue analysis and design must be carried out while the structure<br />
is being designed and revised to satisfy numerous other pre-service<br />
and in-service loading conditions. Thus, to achieve an effective<br />
design the overall design strategy should incorporatefatigue as an<br />
integral part of design, with primary impact on design details,<br />
redundancy, material and fabrication specifications, operational<br />
performance, inspection program and cost. Because structures’<br />
susceptibility to fatigue and the severity of fatigue environment<br />
varies, the chosen fatigue design and analysis methodology, the<br />
sequence, and the extent of the fatigue design effort should be<br />
compatible with the overall design program and should be carefully<br />
planned and monitored to prevent construction delays or costly<br />
modifications during construction.<br />
1.2<br />
OBJECTIVES<br />
This document was prepared to provide the engineer with an up-todate<br />
assessment of fatigue analysis and design. It may be used<br />
either as a comprehensive guideline or a quick reference source.<br />
The first four sections of the report provide an overview and<br />
1-2
general assessmentof fatiguewhile the Iatterfive sectionsprovide<br />
in-depth discussion. The objectives of the document are:<br />
●<br />
Review, assess anddocument all fatigue parametersthat maybe<br />
grouped into a set of parameters (i.e., strength models,<br />
stress history models, analysis methods, etc.)<br />
●<br />
Review, assess and document strengths and weaknesses of<br />
current fatigue analysis and design procedures in conjunction<br />
with existing codes and standards.<br />
●<br />
Documentresearchgaps andrecomnend additionalresearchbased<br />
innumerous analyticalandexperimentalwork resultspublished<br />
every year.<br />
●<br />
Recommend a guideline on fatigue avoidance strategy based on<br />
numerous variablescontributingto the uncertainty of fatigue<br />
life, on recent research results and on current practices.<br />
9<br />
Assess and discussthe accuracyof fatigue life estimationand<br />
the complexity of computation based on the implication of<br />
uncertainties associatedwith the fatigue parameters and the<br />
time and effort necessary to carry out fatigue analysis and<br />
design to various levels of complexity.<br />
1.3<br />
SCOPE<br />
The following tasks were key elements in preparation of this<br />
document.<br />
●<br />
Review and assess global fatigue analysis, including fatigue<br />
as an integral part of design effort, current industry<br />
practices, codes and standards, and the implications of<br />
fatigue damage.<br />
●<br />
Review and assess all parameters within the stress model<br />
umbrella for their relative accuracy as well as application,<br />
1“3
including environmental conditions, structural response,<br />
generationofloa&, developmentof stress response amplitude<br />
operators (RAOS) and hot-spot stresses.<br />
●<br />
Review and assess all parameters within the stress history<br />
model umbrella, including scatter diagram, hindcasting, wave<br />
spectra and application ranges.<br />
Review fatiguedamage assessmentmethodologies, includingthe<br />
effects of numerous analysis and design uncertainties, and<br />
prepare a guideline to both improve fatigue performance of<br />
marine structures and simplify fatigue analysis.<br />
●<br />
Report the findingsin aclearand concise document, including<br />
directly applicable unpublished and published data.<br />
1-4
2.<br />
OVERVIEW OF FATIGUE<br />
2.1<br />
FATIGUE PHENOMENA<br />
Metal structures subjected to variable or repeated loads can fail<br />
without ever reachingtheir staticstrengthdesign loads. This type<br />
of failure,which consistsof the formationand growth of a crack or<br />
cracks, has come to be known as “fatigue”.<br />
Failures observed due to the growth of defects subjected to cyclic<br />
loadings is due to a very complex phenomena, affected by many<br />
parameters. Any environment or condition that results in cyclic<br />
Ioading and reversalof componentstressesmay cause fatiguedamage.<br />
Cyclic stresses are typically caused by machinery vibrations,<br />
temperature changes and wind and wave actions. But although<br />
vibrations and temperaturechanges may be important to fatigue in a<br />
local component, these loadings are not a major concern in the<br />
global behavior of typical marine structures. Thus, the overview<br />
presented in this section addresses wave and wind environments,<br />
excitation forces on mobile and stationary structures and the<br />
response of these structures to excitation forces.<br />
A defect subjected to a large number of cyclic stresses undergoes<br />
three phases of stable crack growth:<br />
●<br />
●<br />
●<br />
Crack initiation, or development of a defect into a<br />
macroscopic crack.<br />
Crack propagation, or development of a crack into a critical<br />
size.<br />
Cracked weldment residual strength exceedence.<br />
The relative durations of these three phases depend on many<br />
variables,includingmaterialproperties,defectgeometry, structure<br />
stiffness, stress cycle magnitudes, distribution and sequence,<br />
operating environment and maintenance. The objective is to prevent<br />
fatigue failure by designing to ensure that the time required to<br />
2-1
complete the three-phasestable crack growth is always greater than<br />
the design fatigue life.<br />
The basic characteristicsof defects and the fatigue phenomena may<br />
be summarized as:<br />
Eventhe most thorough inspectionsat the fabricationfacility<br />
will not reveal very small defects (less than 0.5 mm).<br />
These defects will grow when subjectedto cyclic stresses due<br />
to environmentalloads, structure dynamics (vortex shedding,<br />
machinery vibrations, etc.), temperature changes, etc.<br />
Repeated cyclic stresses and defect growth are additive,<br />
making the fatigue damage cumulative.<br />
In most cases, fatigue is insensitive to the presence of<br />
constant loads. Consequently, stress ranges (i.e., peak-topeak<br />
values) are used to characterize fatigue stresses.<br />
Although a small number of extreme stress ranges may<br />
contribute to fatigue damage, most fatigue damage is due to<br />
the occurance of a large number of small stress ranges.<br />
Poor structuraldesign details will amplify peak stresses.<br />
Distortions and residual stresses introduced during original<br />
fabrication (as well as extensive repair efforts) often<br />
adversely affect material resistance to crack growth.<br />
Corrosion and ocean environment adversely affect material<br />
resistance to crack growth.<br />
Asimplifiecl summary of fatigue phenomena is presented on Figure 2-<br />
.<br />
1.<br />
2-2
2.2 FATIGUE ANALYSIS<br />
2.2.1 Anal.vsisSequence<br />
The basic fatigue analysis sequence is shown as a block diagram on<br />
Figure 2-2 and further discussed in this overview and in Sections 3<br />
through 7.<br />
Fatique Environment<br />
Wave and wind environments are both site- and time-dependent.A<br />
brief observationofwind and the waves it generates shows that they<br />
are random phenomena,where wind speed, direction and duration and<br />
wave height, period and breadth continually change.<br />
Although the real sea is random, the wave environment can be<br />
described by two methods. In the deterministic method, the sea is<br />
described as composed of identical, regular, individualwaves. In<br />
the spectral method, the sea is described as a function of sea<br />
surface elevation due to regular waves combining to form an<br />
irregular sea.<br />
The service life of a vessel/structure may be 20 to 40 years.<br />
During the service life more than 500 millionwaves arelikely to be<br />
applied on the vessel/structure. The fatigue environment is often<br />
defined basedon a series of 15020 minues records taken every3 or<br />
4 hours. The environment is summarizd in a wave scatter diagram.<br />
The wave scatter diagram is a grid of boxes with rows of equal Hs<br />
(significant wave height) and columns for characteristic period,<br />
often Tz (zero up-crossing period) or Ts (significantperiod).<br />
For example: Wave records taken by a weather buoy can be sampled<br />
every four hours. The sample records are reduced by Fast Fourier<br />
Transform (FFT) and integratedto derive the statistical parameters<br />
of Hs and Tz. The whole of the sample parameters are sorted by Hs<br />
and Tz. The number of samples of each Hs-Tz combination are placed<br />
in the correspondingbox in the scatter diagram. Often the scatter<br />
2-3<br />
-J .;–<br />
., --,
diagram boxes are normalized so that the sum of all of the numbers<br />
is 1000. The shapes of the reduced spectra can be compared and a<br />
representative spectrum formula can be fit to the typical shape.<br />
The JONSWAP spectrum is often used to fit sampled spectra shapes,<br />
because of the flexibility offered by the Gamma and Sigma<br />
parameters; see AppendixA, Section3.3. Similar seastates are then<br />
combined into a scatter diagram.<br />
The wind loading on a structure is composed of mean and cyclic<br />
components. To carry out a fatigue analysis of a structure<br />
subjected to cyclic wind loading the magnitude of loading and<br />
associated frequencies must be quantified. Individual component<br />
members of a structuresubjectedto continuousmean wind loadingmay<br />
be susceptible to vortex shedding vibrations. A comprehensive<br />
coverage of wind-inducedfatigue phenomena is presented in Appendix<br />
D.<br />
Acomprehensive reviewof ocean environment,covering both waves and<br />
wind, is presented in Appendices A and B.<br />
FatictueStress Model<br />
The term fatigue stressmodel is often used to define a combination<br />
of analysis steps, covering:<br />
●<br />
●<br />
●<br />
Generation of loads<br />
Structural analysis to determine nominal stresses<br />
Estimation of hot spot stresses<br />
These analysis steps are identified as fatigue analysis blocks and<br />
combined into a single stress model block on Figure 2-2.<br />
The analysis steps undertaken to determine the local hot spot<br />
stresses are sequentialand an inaccuracyat any step contributesto<br />
compounding of the overall inaccuracy. Although many variables<br />
directly influencethe accuracyof estimatedhot spot stresses, some<br />
of the more importantvariables are listed below:<br />
2-4
●<br />
Loads generated as affected by the definition of environment,<br />
selection of wave theories, response characteristics of the<br />
vessel/structuresubjectedtoexcitationalenvironmentalloads<br />
and computer modeling.<br />
●<br />
Structural analysis as affected by the computer model,<br />
software package and engineering decision/selection of<br />
locations for determinatingof nominal stress.<br />
● ✍<br />
Hot spot stresses as affected by determination of stress<br />
concentrationfactors (SCFsdeterminedfromempiricalformulas<br />
based on databasesof numericaland experimentalwork) and the<br />
engineering decision on multiple recomputation of SCFS to<br />
account for variations in stress distribution (i.e.,<br />
reclassificationofdetail/joint for each transfer function).<br />
Another vary importantvariable,fatigue analysismethod, is briefly<br />
discussed in Section 2.2.2.<br />
Fatique Stress Historv Model<br />
The stresses computed may be either stress states (defined by wave<br />
height and wave period and representing a single cycle of loading)<br />
or peak values associatedwith discretewaves. A generalized stress<br />
history model combines this data with long-term wind and wave<br />
distributions (scatter diagram, spectra, directionality, etc.) to<br />
develop a long-term distribution of stresses.<br />
Material Resistance to Fatique Failure (StrenqthModel)<br />
The material resistance to fatigue failure will primarily depend on<br />
the characteristics of detail/joint geometry, material chemical<br />
compositionand mechanical properties,and the service environment.<br />
The material resistance is typically determined in a laboratory<br />
environmentby the applicationof constantamplitude stress cycle on<br />
various detail/joint geometries until fatigue failure occurs. By<br />
2-5
carrying out similar tests for different stress amplitudes a<br />
relationship between the stress amplitude (S) and the number of<br />
cycles (N) is established. The S-N curves developed for simple<br />
details (i.e., stiffener, cutout, etc., applicable for most ship<br />
details) account for the peak (hot spot) stresses and can be<br />
directly used with the member nominal stresses.<br />
The tubular joint details (i.e., T, K, Y, etc., joints applicable<br />
for an offshore platform) exhibit a wide variety of joint<br />
configurations and details. The S-N curves for tubular joint<br />
details do not account for hot spot stresses, requiring the<br />
application of stress concentration factors (SCFS) on computed<br />
nominal stresses.<br />
Cumulative Fatique Damaqe<br />
A relatively simple approach used to obtain fatigue damage requires<br />
dividingof stressrangedistributionintoconstant amplitudestress<br />
range blocks, assumingthat the damage per load cycle is the same at<br />
a given stress range. The damage for each constant stress block is<br />
defined as a ratio of the number of cycles of the stress block<br />
required to reach failure. The most often used Palmgren-Miner<br />
linear damage rule defines the cumulative damage as the sum of<br />
fatigue damage incurred at every stress block.<br />
2.2.2<br />
Analysis Methods<br />
A suitable fatigue analysis method depends on many parameters,<br />
includingstructureconfiguration,fatigueenvironment,operational<br />
characteristics and the design requirements. A fatigue analysis<br />
method may be deterministicor probabilistic. A fully probabilistic<br />
method accounting for uncertainties in defining stresses due to<br />
random loads, scatter in S-N data and randomness of failure is<br />
suited to marine structures. However, less complex deterministic<br />
methods are primarily used to analyze the fatigue lives of marine<br />
structures.<br />
2-6
A deterministic method is sometimes identified as probabilistic<br />
analysis as the randomnessof the ocean environment is accounted for<br />
by incorporatingthe wave spectra. Thus, dependingon how the loads<br />
are generated, the fatigue analyses method may be identified as:<br />
●<br />
Deterministic - Single Wave<br />
● Spectral - Regular Waves in Time-Domain<br />
● Spectral - Regular Waves in Frequency-Domain<br />
● Spectral - IrregularWaves in Time-Domain<br />
● Spectral - Wind Gust<br />
Further discussion on fatigue analyses parameters and analysis<br />
sequence is presented in Sections 3 and 4, respectively.<br />
2.3<br />
SIGNIFICANCE OF FATIGUE FAILURE<br />
An improperdesign may lead to an unacceptablecatastrophic fatigue<br />
failure, resulting in loss of life and damage to the environment.<br />
Non-catastrophic fatigue failures are also unacceptable due to<br />
difficulty and cost of repairs as well as the need to increase<br />
costly inspection and maintenance intervals.<br />
Numerous marine structures of different configurations are in<br />
operation. As illustrated on Figure 2-3, these structures may be<br />
grouped as “mobile”or “stationary”,depending on their functional<br />
requirement. Although mobile vessels/structurecan be moved to a<br />
shipyard for repairs, the total cost of the repair includes<br />
downtime. Stationary offshore vessel/structure inspections and<br />
repairs are extremely costly due to on-location work and their<br />
operating environment, yet the effectiveness of repairs is often<br />
uncertain. Thus, for bothmobile and stationarymarine structures,<br />
it is essential to consider avoidance of fatigue failure at every<br />
phase of design and fabrication.<br />
2-7
2.4 FATIGUE FAILURE AVOIDANCE<br />
Fatigue failure avoidance is not just a motto, but a goal that can<br />
be achieved with relativeease if the fatigue design is an integral<br />
part of the original design program.<br />
Despite their diversity,most marine structuresare designed tomeet<br />
established functional requirements, environmental criteria and<br />
rules and regulations. The design process is executed through<br />
several stages to optimize structure configuration and operational<br />
performance. Since the objectives identified to achieve<br />
optimization are not necessarily compatible, various trade-offs<br />
become necessary. To ensurethat fatigue failureavoidancestrategy<br />
is compatible with the overall design objectives an interactive<br />
design sequence is essential.<br />
2-8
1<br />
APPLICATION OF NUMEROUS<br />
CYCLIC STRESSES<br />
MATERIAL<br />
RESISTANCE<br />
AFFECTED BY<br />
FABRICATION<br />
EFFECTS<br />
MATERIAL<br />
RESISTANCE<br />
‘AFFECTED BY<br />
IN-SERVICE<br />
EFFECTS<br />
9STABLE<br />
CRACK GROWTH<br />
I FATIGUE FAILURE I<br />
Figure 2-1 Fatigue Phenomena Block Diagram Summary<br />
--l<br />
/:, J<br />
-.
ENVIRONMENTAL CRITERIA<br />
(DEFINITIONOF ENVIRONMENT<br />
HIND, WAVE ETC.)<br />
.—. —.— -<br />
r<br />
GENERATION OF LOADS “1<br />
I<br />
I<br />
I<br />
I STRUCTURAL ANALYSIS TO LI FATIGUE STRESS<br />
MODEL<br />
1’ OBTAIN NOMINAL STRESSES I<br />
1 I<br />
ESTIMATION OF HOT SPOT STRESSES<br />
EACH STRUCTURAL DETAIL<br />
L- -—- —. -— -—. .1<br />
rI -—-—--—.—.<br />
TIME HISTORY OF STRESSES +<br />
‘<br />
L-’<br />
-—-— . .—. —.<br />
J<br />
FATIGUE STRESS<br />
HISTORY MODEL<br />
RESISTANCE TO FATIGUE FAILURE<br />
~<br />
I<br />
ESTIMATION OF CUMULATIVE<br />
FATIGUE DAMAGE<br />
I<br />
Figure 2-2 Fatigue Analysis Block Diagram Summary
I<br />
MOBiLEANDSTATIONARY<br />
MARWE STRUCTURES I<br />
&<br />
Ifltil<br />
Rm4Mmm5<br />
n——<br />
Q ?LOAIWO<br />
MAW<br />
L<br />
Hmm<br />
. m Jails<br />
El<br />
●xa5 m<br />
-imr<br />
El 11-<br />
m-.<br />
WM!zs<br />
mm<br />
flwimm<br />
m@<br />
Figure2-3<br />
MOBILE AND STATIONARY<br />
MARINE STRUCTURES<br />
121<br />
-mm<br />
● —m<br />
8-wn<br />
El<br />
JAtm-v?<br />
u R41S1<br />
la<br />
mwm/<br />
Wur Vm2s
-.<br />
(THIS PAGE INTENTIONALLY LEFT BLANK)
3.<br />
FATIGUE DESIGN AND ANALYSIS PARAMETERS<br />
One approach to assess the variables,parameters and assumptionson<br />
fatigue is to separate the design from analysis. Fatigue design<br />
parametersdo affectthe fatigueperformanceand they can be revised<br />
during the design process to optimize the structure.<br />
Fatigue analysis parameters and assumptions affect the computed<br />
fatigue life of the structure. The analyses approach selected<br />
should be compatible with the structure configuration and its<br />
fatigue sensitivity. Both fatigue design and analyses parameters<br />
are summarized on Figure 3-1 and discussed in the following<br />
sections.<br />
3.1<br />
REVIEW OF FATIGUE DESIGN PARAMETERS<br />
All of the parameters affecting fatigue performance of a marine<br />
structure and its components can be grouped into three categories<br />
based on both function and chronological order. The three groups<br />
are:<br />
●<br />
●<br />
●<br />
Design parameters<br />
Fabrication and post-fabricationparameters<br />
In-Serviceparameters<br />
The parameters in these three groups actually dictate crack<br />
initiation,crack propagation to a critical size and exceedance of<br />
cracked weldment residual strength. While these parameters are<br />
assessed and incorporated into a design program to qualitatively<br />
enhance fatigue performance,quantitativeanalyses are necessaryto<br />
verify that the structure’s components have satisfactory fatigue<br />
lives. Fatigue analysis parameters and analysis sequence are<br />
discussed in Sections 3.2 and 3.3, respectively.<br />
3-1<br />
.,
3*1*1 Desiqn Parameters<br />
There are numerousparametersthat can be incorporatedinto a design<br />
to enhance fatigue performance, These parameters are grouped into<br />
four general categories:<br />
●<br />
●<br />
●<br />
●<br />
Global configuration<br />
Component characteristicsand structural details<br />
Material selection<br />
Fabrication procedures and specifications<br />
The effect of these parameters<br />
discussed as follows.<br />
are summarized on Figure 3-2 and<br />
Global Configuration<br />
The overall configuration of every marine structure, mobile or<br />
stationary, should be reviewed to ensure that the applied<br />
environmental forces will be minimized. Trade-offs are often<br />
necessary to ensure that the extreme environment and operating<br />
environmentloadingsare both as low as possible (althoughit may be<br />
that neither is minimized) to ensure overall optimum performance.<br />
Planned redundancy is extremely beneficial to fatigue performance<br />
because alternativeload paths are providedto accommodatea fatigue<br />
failure. Such redundanciesprevent catastrophicfailures, and also<br />
provide ample time for repair of local failures.<br />
Component Characteristicand Structural Details<br />
Wherever possible,acomponent’s arrangementand stiffnessshouldbe<br />
similar to that of adjacent components to ensure a relatively<br />
uniform load distribution. Nominal stresses at a given detail will<br />
be amplified because of the geometry of the detail. The ratio of<br />
the peak or hot spot stress to the nominal stress, known as the<br />
stress concentration factor (SCF), is affected by many variables,<br />
including component member load paths, interface plate thicknesses<br />
3-2
and in-plane/out-of-planeangles,and stub-to-chorddiameter ratios<br />
(for tubular members).<br />
The arrangement of structural details is very important from a<br />
standpoint of their configuration (affecting SCFS) and access<br />
(affectingqualityof work). <strong>Ship</strong>hullstiffenersare often arranged<br />
with these considerationsinmind. Similarly,tubular interfacesof<br />
less than 30 degrees are not desirable in order to ensure reasonable<br />
access for assembly and inspection.<br />
Material Selection<br />
Steel material is selected not only for strength but also for its<br />
other characteristics,includingweldabilityand durability. Thus,<br />
the material selected should have both the chemical composition and<br />
the mechanical properties to optimize its performance. The use of<br />
higher strength steel requires specification of higher material<br />
toughness requirements to meet the limits on fabrication flaws.<br />
Since the material with higher toughness can tolerate larger loads<br />
for a given flaw without brittle fracture during its service life,<br />
such a material is preferred.<br />
Impuritiesinsteel (includingCarbon,Nitrogen,Phosphorus,Sulphur<br />
and Silicon) can cause temper embrittlement, thereby decreasing<br />
notch toughness during the cooling of quenched and tempered steel.<br />
Desirable notch toughness (Charpy) and Crack Tip Opening<br />
Displacement (CTOD) test results are not always achieved at the<br />
fabrication yard. Inspection of the welded joint root, weld<br />
material and the heat-affected zone (HAZ) may show degradation of<br />
root toughness, sometimesextending into the parent material beyond<br />
the HAZ.<br />
Studies carried out by Soyak et al (Reference 3.1) to assess<br />
fracture behavior in alowtoughness HAZ indicatedthat a small lowtoughness<br />
area in the HAZcan be masked by the higher-toughnessarea<br />
surrounding it. Thus, Soyak et al recommend requiring testing of<br />
3-3<br />
..— .——.
not three but five Charpy specimens from the low-toughness HAZ<br />
region to more accurately predict brittle fracture.<br />
On the other hand, crack-toughness levels implied in the impact<br />
tests required in design guidelines may be overly conservative.<br />
Pense’s work (Reference 3.2) indicates that the ship hull strain<br />
rates during crack initiation,propagationand arrest are lower than<br />
those estimated, confirming higher levels of crack-toughness.<br />
Fabrication Specificationsand Procedures<br />
Degradation of root toughness extending into the parent material<br />
beyond the HAZ can be caused by procedures used in the fabrication<br />
yard. Loosely specified fabrication tolerances often result in<br />
fabrication and assembly distortions and may cause strain aging<br />
embrittlement. Unnecessarily tight tolerances could result in<br />
repair work that might contribute to degradation of material.<br />
Fabrication procedures contribute to the pattern of local weldment<br />
defect distribution, residual stress pattern in the HAZ, and<br />
material properties. Since these factors in turn directly affect<br />
crack growth, fabrication procedures should be carefully developed<br />
for each design.<br />
3.1.2<br />
Fabrication and Post-FabricationParameters<br />
Activities in the shipyard or fabricationyard directly impact the<br />
fabricatedmarine structure. These activitiescan be categorizedas<br />
either fabrication or post-fabricationparameters (Figure3-3).<br />
Fabrication Parameters<br />
The primary fabrication parameters can be defined by the questions<br />
who, what, when and how. Each of these parameters affects the<br />
fabricationquality, in terms of residualstresses,defects,repairs<br />
and post fabrication processes. These variables, which determine<br />
3-4
the general quality of fabrication, also affect specifics such as<br />
the rate of crack growth and corrosion.<br />
The four primary fabrication parameters are:<br />
●<br />
Who is Doinq the Work? (i.e. personnel qualificat<br />
on)<br />
●<br />
What are the Work Requirements? (i.e.,defining th(<br />
program<br />
●<br />
When is the Work Done? (i.e., sequence/timingof activity)<br />
●<br />
How is the Work Done? (i.e., following the specifications)<br />
Post-FabricationParameters<br />
Both the design parameters and fabrication parameters directly<br />
affect fatigue performance of a fabricated component, thereby<br />
influencing the post-fabrication processes. The post-fabrication<br />
processes discussed here are activities that enhance the fatigue<br />
performance of the structure component.<br />
The toe of the weld and the weld root often contain geometric<br />
imperfections and high localized stresses and therefore they are<br />
often the site of fatigue crack propagation. To enhance fatigue<br />
performance,modificationof both the weld geometry and the residual<br />
stress is recommended. The weld geometry can be improved by weld<br />
toe grinding,which is often specifiedto obtain a smooth transition<br />
from weld to the parent material. This process should improve<br />
fatigue life locally both by removing small defects left at the toe<br />
during welding and by reducing the stress concentrationat the weld<br />
toe due to elimination of any notches. Weld toe remelting (by TIG<br />
or plasma dressing) and the use of special electrodes for the final<br />
pass at the toe can also improve fatigue performance.<br />
Post-weld heat treatment (PWHT) is recommended to relieve residual<br />
stresses introducedin welding thick sections,typically defined as<br />
having a wall thickness in excess of 2.5 in (63 mm) in U.S. (less<br />
3-5
elsewhere). Both thermal stress relief and weld material straining<br />
to set up desirable compressive stresses at the weld toe are used.<br />
Typically, a node subjected to PWHT experiences both stress and<br />
strain relief and should exhibit improved fatigue performance.<br />
However, the efficiency of PWHT needs further verification. Some<br />
experts in the field consider it difficult to justify any<br />
improvement of fatigue performance as a result of stress relief.<br />
Corrosionprotection is necessaryto ensure as-designedperformance<br />
of the structure, including achieving the desired fatigue life.<br />
Post fabrication work on corrosion protection systems varies from<br />
installation of anodes for cathodic protection to coating and<br />
painting.<br />
3.1.3 In-Service Parameters<br />
The environment in which fatigue cracks initiate and grow<br />
substantially affects fatigue life. The environment affects<br />
corrosionand crack growth due to both the nature of the environment<br />
(i.e., sea water properties, including conductivity, salinity,<br />
dissolved oxygen, pH and temperature) and the magnitude and<br />
frequency of the applied loading (i.e., wind, wave and current<br />
characteristics).<br />
Environmental loads that cause reversal of stress on a marine<br />
structure component are primarily caused by wave and wind<br />
action. While the loading directionality and distribution is often<br />
carefully accounted for, the sequence of loading usually is not.<br />
The other in-serviceparametersreflect inspection,maintenance and<br />
repair philosophy and have a major influence on corrosion and the<br />
rate of crack growth. The in-serviceparameters are summarized on<br />
Figure 3-4.<br />
3-6
3.2 REVIEW OF FATIGUE ANALYSIS PARAMETERS<br />
3.2.1 Fatique Analysis Criteria<br />
Fatigue analysis criteria for marine structures are developed in<br />
conjunctionwith the overall design criteria. The structure type,<br />
environmentalconditionsand the scope of the overall design effort<br />
all affect the fatigue analysis criteria. A fatigue life that is<br />
twice as long as the structure’sdesign life is routinely specified<br />
to ensure satisfactory fatigue performance. Larger safety factors<br />
are often used for critical components where inspection and/or<br />
repairs are difficult.<br />
For many marine structures the use of a probabilistic fatigue<br />
analysis, based on a probabilistic simulation of applied forces,<br />
residual stresses, defects and imperfections, crack growth and<br />
failure, appears to be desirable. This true probabilistic method<br />
may be considered an emerging technology and the time and cost<br />
constraints often require alternative methods to develop a design<br />
that meets the fatigue criteria.<br />
Although the following sections refer to both “deterministic”and<br />
“probabilistic”fatigue methods, essentially the discussions cover<br />
deterministicmethods. The probabilisticmethodsdefined only refer<br />
to probabilistictreatment of the ocean environment.<br />
3.2.2 Interacting Parameters<br />
Fatiguedesign and analysis is carriedout in conjunctionwith other<br />
activities that ensure proper design of the structure to meet all<br />
pre-service and in-service loading conditions. The structure and<br />
its component members must have sufficient strength to resist the<br />
extreme loads for a range of conditions, and these conditions are<br />
often interdependent.<br />
The design is an iterative process in which the general<br />
configuration gradually evolves. Thus, the fatigue design and<br />
3-7
--<br />
analysis process is often initiated after the initial structure<br />
configuration has been defined, but while its components are still<br />
being designed and modified.<br />
3.2.3<br />
Stress Model Parameters<br />
A generalized stress model representsall of the steps necessary to<br />
define the local stress ranges throughout the structure due to the<br />
structure’s global response to excitation loads. These parameters<br />
are as follows:<br />
Motions [Hydrodynamics)Model<br />
Amotions (hydrodynamics)model includesvariousmodels necessaryto<br />
determine the applied excitation forces, response of the structure<br />
to these forces, and the resultant loads on the system. The choice<br />
of a model primarilydepends on the structureconfiguration. While<br />
a continuous finite element model may be used for ship-shaped<br />
structuresor semisubmersibleswith orthotropicallystiffenedplate<br />
system (i.e. continuoussystems),a discrete space frame consisting<br />
of strut members are typically used for the analyses of an offshore<br />
platform.<br />
Floating structures,whether ship-shaped,twin-hulledor of another<br />
configuration,may requirethe use of diffractionanalysesto define<br />
the hydrodynamiccoefficients. Diffractionpressuresgenerated are<br />
transformed intomember wave loadswhile the radiationpressuresare<br />
transformedintoaddedmass and dampingcoefficients. This approach<br />
is valid to obtain hydrodynamic coefficients for non-conventional<br />
geometries,the motion analysisutilizinghydrodynamiccoefficients<br />
does account for the effects of member interaction and radiation<br />
damping components.<br />
Bottom-supportedstructuresare generallymade up of small-diameter<br />
tubulars, and their drag and inertia coefficients can be defined<br />
based on previous analytical and model basin work on tubulars.<br />
However, some componentsare frequency-dependentfor arange ofwave<br />
3-8
frequencies of interest, requiring definition of frequency<br />
dependency.<br />
Thus, some of the more importantparameters to be considered in the<br />
development of a hydrodynamicsmodel are:<br />
●<br />
<strong>Structure</strong>configuration(continuousversus discrete systems).<br />
●<br />
<strong>Structure</strong> size and irregularityof shape.<br />
●<br />
<strong>Structure</strong> component member dimensions (with respect to both<br />
the structure and the wave length).<br />
●<br />
Component member arrangement (distancefrom each other).<br />
●<br />
Component<br />
coefficients.<br />
member shape, affecting its hydrodynamic<br />
Analysis Techniques<br />
Analysis techniques, or the approaches used to generate and apply<br />
environmental loads, fall into two categories: deterministic<br />
analysis and spectral analysis. Deterministicanalysis is based on<br />
the use of wave exceedance curves to define the wave occurrences.<br />
Spectral analysis (alsoreferredtoas probabilisticanalysis of the<br />
ocean environment only) is based on the use of wave spectra to<br />
properly account for the actual distribution of energy over the<br />
entire frequency range.<br />
The five approaches can be defined in these two categories:<br />
●<br />
Selected Wave[s) - Determinist-it<br />
Aclosed-form deterministicanalysisprocedure recommendedby<br />
Williams and Rinne (Reference 3.3) is often used as a<br />
screening process. This approach may be considered a<br />
marginally acceptable first step in carrying out a fatigue<br />
3-9
analysis of a fixed platform. As discussed in Section 3.2.4<br />
under Stress History Parameters, wave scatter diagrams are<br />
used to develop wave height exceedance curves in each wave<br />
direction and used to obtain the stress exceedance curves.<br />
Consideringboth the effort needed and the questionablelevel<br />
of accuracy of selecting wave heights to represent a wide<br />
range of wave heights and periods, it may be better to<br />
initiate a spectral fatigue analysis directly.<br />
●<br />
Reqular Waves in Time Domain - Spectral<br />
Because a spectralfatigueanalysis is carriedout to properly<br />
account for the actual distribution of wave energy over the<br />
entire frequency range, a sufficient number of time domain<br />
solutions is required to define the stress ranges for<br />
sufficientpairs of wave heightsand frequencies. A resultof<br />
this procedure is development of another characteristic<br />
element of spectral fatigue analysis, namely, the stress<br />
transfer functions, or response amplitude operators (RAOS).<br />
For each wave period in the transfer function, a sinusoidal<br />
wave is propagated past the structure and a wave load time<br />
history is generated. The equations of motion (structure<br />
response) are solved to obtain a steady state response. A<br />
point on the transfer functionat the wave period is the ratio<br />
of the responseamplitudeto the wave ampl” tude. A sufficient<br />
number of frequencies is required to incorporate the<br />
characteristicpeaks and valleys.<br />
●<br />
Random Waves in Time Domain - Spectral<br />
The use of randomwaves avoidsthe necessityof selectingwave<br />
heights and frequencies associated with the regular wave<br />
analysis.<br />
3-1o
●<br />
Reqular Waves in FrecjuencvDomain - Spectral<br />
This method, based on the use of regular waves in the<br />
frequency domain, requires linearization of wave loading.<br />
Approximatingthe wave loadingby sinusoidallyvaryingforces,<br />
and assuming a constant sea surface elevationdoes contribute<br />
to some inaccuracies. However, these approximations also<br />
allow equationsof motion to be solvedwithout having to carry<br />
out direct time integration, thereby greatly facilitating<br />
fatigue analysis work.<br />
The approach chosen should depend on the structure type and<br />
the environment. For most “rigid body” inertially driven<br />
floating structures, frequency-domain spectral fatigue<br />
analysis is recommended. However, for tethered structures<br />
such as a TLP, and for structures in areas where large waves<br />
contribute substantially to cumulative fatigue damage, the<br />
effects of linearizationand inundation are substantial. In<br />
these cases the preferredapproachmaybe time-domainspectral<br />
fatigue analysis. Even time-domain solutions at several<br />
frequencies may be sufficient to compare the RAOS obtained<br />
from a frequency-domain solution and to calibrate them as<br />
necessary.<br />
●<br />
Wind Gust - Spectral<br />
Most marine structures are designed to resist extreme wind<br />
loadings,but they are rarely susceptibletocyclic wind gusts<br />
that cause fatigue damage. Some structures, such as flare<br />
towers or radio towers, support negligible equipment and<br />
weights; as a result, they are often made up of light and<br />
slender members, making them susceptible to wind-caused<br />
fatigue damage.<br />
As with analysisof the wave environment,structuressubjected<br />
to wind turbulencecan be analyzed by quantifyingcyclic wind<br />
forces and their associated frequencies. The total applied<br />
3-11
wind loading on a structure is due to mean and cyclic<br />
components. The loads are computed and statically applied on<br />
the structureand then convertedto harmonicloads for dynamic<br />
analysis.<br />
The stresses obtained at each frequency are unitized by<br />
dividing them by the corresponding cyclic wind speeds.<br />
Application of wind spectra to define the occurrence of wind<br />
speeds and gust spectra to define the energy content of the<br />
gust on unitized stress ranges yields the stress spectrum.<br />
Further discussionwind loading is provided in Sections 6 and<br />
Appendix D.<br />
Structural Analysis Model<br />
A floating structure is by definition in equilibrium. The applied<br />
loads and inertial response from the motions analysis provide a<br />
balance of forces and moments for the six degree of freedom system.<br />
To obtain a stiffness solution,the structuremodel may be provided<br />
with hypothetical supports. A typical solution should yield close<br />
to zero loads at those hypothetical supports. The deformations<br />
obtained from stiffness analysis at member joints are transformed<br />
into stresses.<br />
A single- or a dual-hulled structure is a continuous system with<br />
large stiffenedmembers/components. Applied loads on the structure<br />
necessitatedeterminationof hullgirder bendingmoments in vertical<br />
and horizontal axes and local internal and external pressure<br />
effects. The use of beam elements may be appropriate when local<br />
pressureeffects are small and stressdistributionpatterns arewell<br />
understood. Since the local pressure effects are substantial for<br />
ship structuresand the local stressdistributionsrapidlychange as<br />
a function of several parameters, a finite element analysis is the<br />
generally recommended approach to determine the local stress<br />
distributions.<br />
3-12
The finite element models of increasing mesh refinement are often<br />
used to obtain accurate stress range data locally in fatigue<br />
sensitive areas. Thus, an overall coarse mesh model of the<br />
structure used in the first stage of analyses is modified by<br />
increasingmesh refinement in various fatigue sensitive areas. The<br />
finite element models are typically built from membrane plate<br />
elements,bendingplate elements,bar elementsand beam elementsand<br />
further discussed in Section 5.<br />
Because the individual joints and members define the global<br />
structure, the boundary conditions should also reflect the true<br />
response of the structure when subjected to the excitation<br />
loads. For a bottom-supported structure, individual piles can be<br />
simulated by individual springs. Whatever the support<br />
characteristics,a foundationmatrix can be developed to represent<br />
the foundation-structureinterface at the seafloor. It should be<br />
noted that the foundation matrix developed for an extreme<br />
environmentwould be too flexible for a milder fatigue environment.<br />
Thus, the foundationmatrix developed should be compatiblewith the<br />
applicable load range.<br />
Stress Response Amplitude Operators (RAOS)<br />
The stress RAOS or stress transfer functions are obtained by<br />
unitizing the stress ranges. If the wave height specified is other<br />
than the unit wave height (doubleamplitude of 2 feet or 2 meters),<br />
stress ranges at each frequency are divided by the wave heights<br />
input to generate the loads. Similarly, wind loads computed based<br />
on cyclic wind velocities at each frequency are divided by the<br />
respective velocities to obtain the unitized stress ranges.<br />
Stress Concentration Factors [SCF] and Hot Spot Stresses<br />
The stresses obtained from a stiffness analysis, and the RAOS<br />
generated,representnominalor averagestresses. However,the load<br />
path and the detailing of orthotropically stiffened plate or an<br />
intersection of tubular members will exhibit hot-spot or peak<br />
3’13<br />
/ Ii7<br />
-/
stresses several times greater than the nominal stresses. The<br />
fatigue test results for a wide variety of shiphull stiffener<br />
geometries can be used directly with the nominal stresses.<br />
At an intersectionof a tubular brace and chord, depending on the<br />
interfacegeometry,the maximum hot-spotstressesoften occur either<br />
on the weld toe of the incoming brace member or on the main chord.<br />
The ratio of the hot-spot stress to the nominal stress is defined as<br />
the stress concentrationfactor (SCF).<br />
SCF =U.aX/Un<br />
The SCF value is probably the most important single variable that<br />
affects the fatigue life of a detail/joint,necessitating accurate<br />
determination of SCFS.<br />
There are several practical approaches for determining SCF values.<br />
The first approach is to develop an analytical model of the<br />
detail/joint and carry out a finite element analysis (FEA). When<br />
modeled correctly, determination of SCFS by FEA is a very reliable<br />
approach. The second approach is to test a physical model and<br />
obtain the hot-spot stresses from measurements. Whether a straingauged<br />
acrylicmodel or other alternativesare used, the accuracyof<br />
hot-spotstresseslargelydependson the abilityto predict hot-spot<br />
stress locations and obtain measurements in those areas.<br />
Although reliable and recommended for obtaining SCFS, these two<br />
methods are time consuming and expensive. Thus, a third approach,<br />
based on applying empirical formulations to determine SCFS, has<br />
been extensively accepted for fatigue analysis of marine<br />
structures. A set of empirical formulae developed by Kuang<br />
(Reference 3.4) were derived by evaluating extensive thin-shell<br />
finite element analyses results. The formulae proposed by Smedley<br />
(Reference 3.5) and Wordsworth (Reference3.6) of Lloyds Register<br />
were derived from evaluating the results of strain-gauged acrylic<br />
models.<br />
3-14
The stressmodel parametersdiscussedabove are summarizedon Figure<br />
3-5. A summary of empirical equations, parametric study results<br />
obtained by using applicable empirical equations for T, K and X<br />
joints, and an illustrative finite element analyses results for a<br />
complex joint are presented in Appendix C.<br />
3.2.4 Stress History Model Parameters<br />
The wave scatterdiagram and wave directionalitydata are necessary<br />
whether a deterministic or a spectral analysis technique is used.<br />
In a deterministicanalysiswave exceedance curves are generated in<br />
each wave direction and used with the hot-spot stresses to obtain<br />
the stress exceedance curves.<br />
For a spectral fatigue analysis, a scatter diagram and the<br />
directional probability is used with wave or wind spectrato obtain<br />
the stress spectrum from hot-spot stresses. These parameters are<br />
summarized on Figure 3-6. Stress History Models are discussed<br />
further in Section 6.<br />
3.2.5 FaticiueDamaqe Comtmtation Parameters<br />
Many parameters affect the fatigue life computation. Some, such as<br />
stress sequence, maintenance and repairs, lapses in corrosion<br />
protection, etc., are not accounted for in fatigue damage<br />
computation. Fatiguedamage is characterizedby an accumulationof<br />
damage due to cyclic loading, with fatigue failure occurring when<br />
the accumulateddamage reaches the critical level. To evaluate the<br />
damage, the stress-time history is broken into cycles from which a<br />
distribution of stress ranges is obtained. The variable-amplitude<br />
stress range distribution is divided into constant-amplitudestress<br />
range blocks, Sri, to allow the use of constant-amplitude S-N<br />
curves.<br />
3-15
Selection of S-N Curve<br />
The S-N curve defines the relationship between a constant-stress<br />
amplitude block and the number of cycles necessary to cause the<br />
failure of a given detail/joint. Such S-N curves are largely<br />
derived by testingmodels ofsimplifieddetail/joint componentswith<br />
subjecting constant amplitude stress reversals in a laboratory<br />
environment. The laboratoryenvironmentis substantiallydifferent<br />
from the typical marine environment.Similarly, the laboratory<br />
models are idealized while actual marine structure details/joints<br />
incorporate fabrication residual stresses and substantial welding<br />
defects.<br />
The S-N curve defining a particular type of detail/joint and<br />
material properties is derived by obtaining the mean of the test<br />
data and then defining the mean minus two standarddeviations. S-N<br />
curveswere firstdevelopedfor fillet-weldedplate details and some<br />
small scale-tests on tubular joints. Later tests provided data on<br />
more complexdetails and thickerplate sections. The S-N curves for<br />
continuous system details (i.e., ship hull stiffening) are<br />
typically reduced by the ratio of hot spot-to-nominalstresses and<br />
can be used directly with shiphull nominal stresses to determine<br />
fatiguedamage. The S-N curves for discrete systemjoints represent<br />
the failure stresses and necessitate multiplication of nominal<br />
stresses by SCFS to obtain hot spot stresses.<br />
The choice of an applicable S-N curve depends not only on the<br />
material, configuration of the detail/joint and the fabrication<br />
effects (residualstresses,weld profile,defects,etc.) but also on<br />
the service condition of the structure. The original<br />
U.K. Departmentof Energy (DEn)recommendedQ-curve, based on simple<br />
thin plate details, has been replaced by a T-curve (Reference1.6).<br />
The American Petroleum Institute (API) recommended X-curve<br />
(Reference 1.5) is applicable to a welded profile that merges with<br />
the adjoining base material smoothly. If the weld profile is not<br />
smooth, then a lower X’-curve is applicable.<br />
3-16
While API S-N curves are applicableto stationarymarine structures,<br />
other S-N curves by DEn and Det norske Veritas (DnV - Ref. 1.7) may<br />
be equally applicableto stationaryand mobile vessels with tubular<br />
and orthogonally stiffened plate construction. The preferred S-N<br />
curve should be defined in the design criteria. Typical S-N curves<br />
applicableformarine structuresare illustratedon Figure3-7. S-N<br />
curves are discussed further in Section 5.<br />
Cumulative Damacle<br />
The calculation of cumulative damage is typically performed using<br />
the Palmgren-Minerdamage rule. In this approach fatigue damage is<br />
calculated by dividing stress range distribution into constant<br />
amplitude stress range blocks, assuming that the damage per load<br />
cycle is constant at a given stress range and equal to:<br />
Dti = l/N<br />
where,<br />
Dtiis the damage, and<br />
N is the constant-amplitudenumber of cycles to<br />
given stress range.<br />
failure at a<br />
Another key assumption of the Palmgren-Miner damage rule is that<br />
damage is independent of order in which loads are applied.<br />
Accordingly, for the case of a stress history with multiple stress<br />
blocks, Sti,each block having n cycles, the cumulative damage is<br />
defined by:<br />
This is the Miner-Palmgren formula, where:<br />
3-17
D is the cumulative damage,<br />
k is the number of stress blocks,<br />
n is the number of stress cycles in stress block i with<br />
constant stress range, and<br />
N is the number of cycles to failure at constant stress range.<br />
Although the linear Palmgren-Minerdamage rule is extensively used,<br />
the significance of constant-amplitudeloading and the sequence of<br />
loading (i.e.,large stress blocks during the beginning rather than<br />
toward the end of design life) may be important to correct<br />
assessmentof fatigue damage. This subject is discussed further in<br />
Section 7.<br />
Fatique Life Evaluation<br />
Fatiguedamage and fatigue life should redetermined at all critical<br />
hot-spot stress areas. While one or two areas may be targeted on a<br />
plate and stiffenerinterface,at least eight points are recommended<br />
on a tubularmember. If eight points, spaced at45 degree intervals<br />
around the circumference, are chosen, relatively accurate hot-spot<br />
stresses and fatigue damage data will be obtained. Typically,<br />
fatigue damage (D) is calculated on an annual basis. The fatigue<br />
life (L) is then determinedby taking the inverseof the accumulated<br />
damage ratio (D).<br />
3-18
DESIGN<br />
Osn40H rAnNsirsR8 FAMcA~ M POST-F~ATW MRA~<br />
PRIMARY AFFECT<br />
SHsRusX<br />
PRIMARY AFFECT uA-FuRtA PRIMAR7 EFFECT<br />
AFFECT<br />
PARA~<br />
~=~t~v:.Y<br />
-==’’”<br />
WEICSR cwuncmm<br />
ROBWOI<br />
Stroamca<br />
ENMRWMENT<br />
● *<br />
FAEIRiCATtW TCUfSANIXS<br />
7<br />
. Wa$h Z.me<br />
DActn<br />
FAMHXSW4 I%roarm ● k WOtw<br />
and<br />
FABRICAllON SEWENCE WALIN Repski Mate of<br />
AND TIME {1.mpwaturo)<br />
&OcA Growth<br />
ENMRWA4ENT#L LWCW3<br />
PaJt-<br />
RATE ff HEAT iNPtJl # T**, A7$Otudo CrJrc.dm<br />
Fobrkatlm —<br />
AND CK!JJNG Moon LOW+ of Str.as and<br />
PrOc*u*m<br />
● OkOcUmd F7c4cbtnty Ratad<br />
and Oll$rktkm<br />
CmKA Growth<br />
● Str+uhg Squa~<br />
===J-E=i=2:a*<br />
lNsPEcnot4<br />
MAINIIk+ANCE<br />
-:=%s: -<br />
WO<br />
REPA!R<br />
WALYSIS<br />
Os’tsss PAmMiTsRo<br />
~–––––______T ‘<br />
I<br />
I<br />
t<br />
/<br />
![<br />
I<br />
I<br />
I<br />
I<br />
I<br />
FAllWE AN#lYSAS<br />
~lERIA<br />
IN-SSRMCE<br />
SPREADING<br />
STRSNGIH FtEOUIREWNIS<br />
~ _ _ ‘__“?_!! ‘_- WRSMldSTW4Y MWSL P~ FAT@3LR OAMMM C0WUTA7SON ?~<br />
; ~.––. –_____T ~_––____7<br />
MOnWS -l<br />
II -----<br />
[LOAb uwn] II I<br />
I<br />
--—._<br />
[ P:=-’=E]<br />
SIRUS’SURA 1[<br />
I I<br />
ANMWAS<br />
It<br />
Mwm<br />
I I ANAll?4S lEC%tNICUES tJ_- SCASIER<br />
● Oefrxma!kma<br />
WAGRAM<br />
I [ ●sh~&.3wa:i<br />
● Str-scc I i<br />
lNllAl<br />
●<br />
I I<br />
Rqutor ~, h<br />
I I FAAK<br />
smucnmAL lim*-Oanah, +=cIroI<br />
OIRECTIONAUTY .<br />
● Rqulor Wwn h<br />
i<br />
I<br />
cm FlaJRAnm rmqum@lw161,~d 1 -r WND<br />
I “:::%% ‘&m,<br />
SI?3ESS HOT SPOT<br />
I<br />
.Mnd 13Joi, %cclrd<br />
WAVE<br />
RAOm<br />
STRESSES<br />
II<br />
I<br />
IT<br />
J<br />
II<br />
I<br />
11<br />
I<br />
II<br />
~ I EMP=~<br />
IJ<br />
Smss<br />
cmcwmmwm<br />
ANAL71tCAL F.E,A. r~<br />
(w,)<br />
II<br />
II<br />
II<br />
II<br />
II<br />
EKaEIANCE<br />
II<br />
II<br />
‘1 I<br />
II<br />
CURWE<br />
II<br />
{Oyrmruc<br />
I<br />
FAllOUE<br />
STRESS II S-H<br />
I<br />
OWA~<br />
WAW m WC MsTmr<br />
Cunuz<br />
I<br />
CC#APUTATtSW<br />
SPECTffA 11— I<br />
(Spmtrd<br />
Wl@a) II I<br />
II<br />
TrPIC#lLY<br />
LINACCCAINTEOPARAMETERS<br />
I I ;!$i%!l!ii%pr.t=llm I<br />
II II I<br />
~___________J &:-:=________J L___________I l_–___–. __J<br />
II<br />
II<br />
I<br />
I<br />
I<br />
I<br />
I<br />
I<br />
Figure<br />
3-1 FATIGUE DESIGN AND ANALYSIS PARAMETERS
Primarily Affect ---------+<br />
r<br />
I<br />
GLOBAL<br />
CONFIGURATION<br />
,[<br />
● Applied Forces<br />
* <strong>Structure</strong> Response<br />
● <strong>Structure</strong> Redundancy<br />
#<br />
* $tre$s Levels<br />
COMPONENT CHARACTERISTIC . * Stress Concentration<br />
AND STRUCTURAL DETAILS<br />
● Access, Workmanship and and Details<br />
MATERIAL SELECTION<br />
J<br />
* Chemical Composition<br />
and Weldability<br />
* Mechanical Properties<br />
and HAZ<br />
1* Corrosion Fatigue Behavior<br />
r<br />
4<br />
* Local Deformations and<br />
FABRICATION PROCEDURE 4 Residual Stress Pattern<br />
AND SPECIFICATIONS<br />
* Defect Distribution and<br />
< <<br />
Initial Rate of Growth<br />
Figure 3-2 Design Parameters
Primarily Affect .--------+’ & In-turn Affect --------~<br />
WELDER QUALIFICATION<br />
t<br />
IFABRICATION TOLERANCES~<br />
IFABRICATION SEQUENCE ~<br />
AND TIME (Temperature)<br />
FABRICATION<br />
QUALITY<br />
Residual Stresse:<br />
Defects<br />
Repairs<br />
t 4<br />
Post-Fabrication<br />
IRATE OF HEAT INPUT ~<br />
Processes<br />
IAND COOLING I b 1<br />
c<br />
o<br />
R<br />
R<br />
o<br />
s<br />
I<br />
o<br />
N<br />
&<br />
c<br />
R<br />
A<br />
c<br />
K<br />
G<br />
R<br />
o<br />
w<br />
T<br />
H<br />
WELD TOE GRINDING<br />
POST-<br />
,..,<br />
IPWHT FOR STRESS RELIEF<br />
FABRICATION<br />
4<br />
CORROSION PROTECTION<br />
PROCESSES<br />
Figure 3-3 Fabricationand Post-FabricationParameters
Primarily Affect .~:--------+.<br />
ENVIRONMENT<br />
* Air<br />
* Splash Zone<br />
* Sea Water<br />
ENVIRONMENTAL LOADING<br />
* Type, Amplitude and<br />
Mean Level of Stress<br />
* Directional Probability<br />
and Distribution<br />
* Stressing Sequence<br />
i<br />
Corrosion<br />
and<br />
Rate of<br />
Crack Growth<br />
INSPECTION<br />
MAINTENANCE<br />
—<br />
REPAIR<br />
Figure 3-4 In-Service Parameters
MOTIONS MODEL<br />
— .—. ● ;LOADS MODEL ~<br />
I<br />
iMASS MODEL~<br />
ANALYSIS TECHNIQUES<br />
●<br />
Single Wave-Deterministic<br />
* Regular Waves In Time-Domain<br />
Spectral<br />
I<br />
*<br />
*<br />
STRUCTURAL<br />
ANALYSIS<br />
MODEL<br />
Deformations<br />
Stresses<br />
* Random Waves In Time-Domain<br />
Spectral<br />
●<br />
Regular Waves In Frequency-Oomairn<br />
Spectral<br />
* Wind Gust-Spectral<br />
I<br />
STRESS<br />
RAos<br />
I * Finite Element Analysis (FEA)I STRESS<br />
t I I<br />
* Testing-Strain Gauge I<br />
m<br />
i<br />
HOT<br />
SPOT<br />
STRESSES<br />
t<br />
CONCENTRATION 1<br />
.—<br />
* Empirical Formulations FACTORS<br />
(SCF)<br />
Figure 3-5<br />
Stress Model Parameters<br />
5’-7<br />
- ,“
SCAITER<br />
DIAGRAM<br />
Y<br />
WAVE<br />
DIRECTIONALIIV<br />
EXCEEOANCE<br />
- -<br />
CURVE<br />
(Oetenninistic<br />
Analysis)<br />
STRESS<br />
HISTORY<br />
J<br />
WIND<br />
WAVE OR WIND<br />
WAVE 4 SPECTRA<br />
(Spectral<br />
SPREADING<br />
4<br />
Analysis)<br />
Figure 3-6 Time History Model Parameters
loa<br />
1<br />
10<br />
-—<br />
—<br />
—<br />
-t-i-nT<br />
1[<br />
-t<br />
_~ _<br />
—_<br />
t<br />
— .- . J . . . -,-<br />
. . — — . . .<br />
—<br />
4<br />
1 j~l I@<br />
Endurume(cycles)<br />
Figure 3-7 Typical S-N Curves
,,- ... , . . . . . . . . . . . :.:- .-, .....==—<br />
.:<br />
(THIS PAGE INTENTIONALLY LEFT BLANK)
4. GLOBAL REVIEW OF FATIGUE<br />
4.1 APPLICABLE ANALYSIS METHODS<br />
4.1.1 Background<br />
Analysis and design of marine structures in the past often did not<br />
include explicit treatment of fatigue. With the installation of<br />
offshore platforms in deeper water increasedemphasis was placed in<br />
fatigue design. An experience-based allowable stress methods<br />
developed were soon complementedwith detailed analyses methods.<br />
<strong>Ship</strong> structure design often did not incorporate explicit treatment<br />
of fatigue through analysis. However, with the increasing use of<br />
higher strength steels, the cyclic stress ranges also increased,<br />
necessitating fatigue analysis of more structures. Although the<br />
allowable stress methods developed are used in the design of<br />
majority of ship structures, more and more of the new designs<br />
incorporatedetailed analysis methods.<br />
Several methods may be applicable and acceptable for the fatigue<br />
analysis and design of a marine structure. The most suitablemethod<br />
depends on many parameters, including structure configuration<br />
(shape, redundancy, details/joints, etc.), fatigue envir[ nment,<br />
operational characteristics/constraints, and the design<br />
requirements. The complexity and cost of this analysis and design<br />
effort should be compatible with available design informat” on and<br />
the desired degree of accuracy of the analysis and design.<br />
The design and analysisparametersdiscussed in Sections 3.1 and3.2<br />
are summarizedon Figure3-1. The four dotted-line boxes around the<br />
analysis parameters illustrate a typical analysis sequence.<br />
Although the methods used in obtaining the hot-spot stress (stress<br />
model), stress spectrum (stresshistorymodel), and the fatigue life<br />
may differ, the general sequence shown is usually followed. A<br />
different sequence is applicable for a simplified analysis and<br />
design method. An allowable stress approach is one such example.<br />
4-1
The different methods and their applicationsequences are discussed<br />
in the following sections.<br />
4.1.2<br />
Simplified Analysis and DesictnMethods<br />
The simplified analysis and design methods applicable to ship<br />
structures and offshore structures are based largely on both<br />
theoretical knowledge and past experience and account for the<br />
environmentlikely to be encountered. Typically, ship hull girders<br />
are designed to resist maximum bending moments due to still water<br />
plus awave-induced conditionderivedfrom harsh North Atlanticwave<br />
data (Reference 4.1). The basic hull girder, designed for the<br />
extreme environment loading, is intended to have ample crosssectional<br />
area and moment of inertiato keep the magnitude of stress<br />
reversalslow and exhibit low susceptibilityto fatiguedamage. The<br />
minimum plate and scantling sizes specified and the detailing<br />
developed are intended to keep the nominal and peak stress ranges<br />
low to prevent fatigue failures in the secondary members. In<br />
addition,steel is specifiedto ensurethat its chemical composition<br />
and mechanical propertieswill make it less susceptible to fatigue<br />
failure.<br />
Similarly, offshore platform joints are designed to resist maximum<br />
punching shear and crushing stresses. The joint details are<br />
developed to minimize the SCFS and cyclic stress ranges to make them<br />
less susceptible to fatigue failure. Such an indirect approach to<br />
fatiguedesign shouldbe supplementedby an empirical approach based<br />
on constant stress range cycle fatigue life test data.<br />
ShitI<strong>Structure</strong>s<br />
An allowable stress method for ship structuredesign should be used<br />
to assess applied stresses against allowable stresses. The<br />
objective of applying the method is to identify those conditions<br />
that requireno furtherfatigue assessmentand those conditionsthat<br />
require more comprehensive fatigue analyses.<br />
4-2
An allowable stress method, also considered a screening process,<br />
relies on both theory and experience. The procedure developed<br />
should be calibrated against available fatigue failure data and<br />
typically incorporatesthe following steps:<br />
1. Computation of wave-induced loads<br />
2. Determinationof applied stress levels<br />
3. Determinationof allowable stress levels<br />
4. Adjustment of allowable stress levels<br />
5. Assessmentofvariouscomponents/detailsforsusceptibi lityto<br />
fatigue failure.<br />
The wave-inducedloads are computedusing simplifiedformulae,where<br />
the long-storm distribution of fatigue loading is represented by a<br />
single characteristic value. The vertical bending moment is<br />
computed as a function of the vessel length, breadth and block<br />
coefficient along the longitudinal axis. The applied (nominal)<br />
cyclic stress amplitude is determined by using beam theory and<br />
dividing the vertical bending moment at any point along the<br />
longitudinal axis with hull girder section modulus.<br />
The allowable stresses depend on many variables. For a simplified<br />
method an allowablestress may be defined as a function of location<br />
(deck, side shell, etc.) and detai1 geometry (1ocal stress factor).<br />
Typically, such a method is based on a 20-year service life,<br />
standard corrosioneffects and a nominal geographic area. Thus if<br />
specific service life or routing information is available, the<br />
allowable stress levels are adjusted. Two of the of the simplified<br />
analysis methods are:<br />
1. ABS’ Al1owable Stress Method<br />
This allowablestressmethod by Thayamballi (Reference4.2) is<br />
primarily intended for use in fatigue screening of tankers.<br />
The simplifiedformulae presentedallow calculationof several<br />
types of loadingon atanker due to wave-inducedmotions. The<br />
loading types and their relevancy are:<br />
4“3
●<br />
Vertical bending moment - needed to determine stresses<br />
along the longitudinal axis<br />
●<br />
Internal tank load - needed to determile stresses at<br />
tank boundaries<br />
●<br />
Externalpressureload- needed todeterm” ne stressesat<br />
outer hull<br />
Each of these component loads are applied to the structure<br />
independentof one another. The method implementsbeam theory<br />
to obtain nominalstresses,except for special caseswhere ABS<br />
Steel Vessel Rules require special consideration. ABS Rules<br />
requiring structural analysis also provide substantial<br />
flexibilityfor engineeringjudgement. The fatigue sensitive<br />
areas of the deck, tanks and the hull shell, where the<br />
stresses are to be determined, are illustratedon Figure 4-1.<br />
Although the method is intended to provide allowable stress<br />
levels for normal operating routes, the allowable stress<br />
levels can be adjusted. Thus, a vessel operating in harsh<br />
geographic regions can still be screened for fatigue by<br />
reducing the allowable stress levels as function of the<br />
severity of the environment. The structural components of a<br />
vessel having stress levels meeting the reduced allowable<br />
stress levels may not require a detailed fatigue analysis.<br />
2. Munse’s Method<br />
This allowable stress method for determining ship hull<br />
performance by Munse et al (Reference 4.3) is a practical<br />
method of designing ship hull structural details for fatigue<br />
loading.<br />
The method is considered reliable, as it is based on a study<br />
of measured fatigue failure (S-N curves) data for 69<br />
structuraldetails. The design method also incorporatesthe<br />
4-4
esults of work covering assessment of 634 structural<br />
configurations (fromReferences4.4 and 4.5). It establishes<br />
the basis for selecting and evaluating ship details and<br />
developing a ship details design procedure. This method<br />
accounts for three of the most important parameters that<br />
affect fatigue life of a ship detail:<br />
●<br />
Mean fatigue resistance of local fatigue details (S-N<br />
curve)<br />
●<br />
Applicationof a ’’reliability”factorto accountforS-N<br />
data scatter and slope<br />
●<br />
Applicationofa “randomload” factorto account for the<br />
projected stress history<br />
Munse’s design method can also be used to estimate fatigue<br />
life based on actual or assumed stress history and a<br />
reliability factor. A study carried out at the American<br />
Bureau of <strong>Ship</strong>ping (ABS) (Reference4.6) to evaluate fatigue<br />
life predictionsutilizedseveralmethods, includingMunse’s.<br />
The study, based on stress histories derived from strain<br />
measurements of containership hatch-corners, provided good<br />
comparative results. Although Munse’s method neglects the<br />
effect of mean stress, the fatigue lives computed compared<br />
well with lives that are computed using other methods.<br />
Munse’s design method is an acceptable fatigue design<br />
procedure for all vessels. This design method allows proper<br />
selectionof design details and providesfor design of a costeffective<br />
vessel appropriate for the long term environmental<br />
loadings. Vessels that are considered non-standard due to<br />
their configuration and/or function (such as a tanker with<br />
internal turret mooring or a drillship) should be further<br />
analyzed, including a thorough spectral fatigue analysis.<br />
Munse’s design procedureis suimnarizedin the block diagramon<br />
Figure 4-2.<br />
4-5
Offshore structures such as a semisubmersibledrilling vessel is a<br />
continuous system,typicallyhaving orthogonallystiffenedmembers.<br />
While a simplified method, such as Munse’s, may be an applicable<br />
screening method, such structures have very specialized<br />
configurations, response characteristics and structural details.<br />
Thus, each structure should be considered unique, requiring a<br />
detailed fatigue analyses.<br />
An offshore platform is made up discrete members and joints. Since<br />
each structure is unique, a detailed fatigue analysis is<br />
recommended. However, asimplified method may beappl icable if such<br />
a method can be developed based on a large number of similar<br />
structures in a given geographic region. Such a method was<br />
developed for the Gulf of Mexico by American Petroleum Institute<br />
(Reference 1.5) and discussed further.<br />
The simplifiedAPI method (Section5.1.1 of Reference 1.5) is based<br />
on defining the allowable peak stresses as a function of water<br />
depth, design fatigue life, member location and the applicable S-N<br />
curve. Although the approach can be modified to apply to other<br />
geographic areas, it was developed by calibrating previously<br />
completed fatigue analyses of fixed offshore platforms. The<br />
maximum allowable stress method is applicable to typical Gulf of<br />
Mexico platforms with structural redundancy, natural periods less<br />
than three seconds, and the water depths of 400 feet or less.<br />
ThisAPI allowablestressmethod i$ intendedfor use as a simplified<br />
fatigue assessmentprocedurefor Gulf of Mexico platforms subjected<br />
to long-term cyclic stresses considered small relative to the<br />
extreme environment stresses. The method attempts to predict<br />
fatigue behavior as a function of the design wave event for a<br />
generalized platform. It should be noted that the applied force<br />
levels can vary substantiallywith platform geometry. The relative<br />
importanceof extremedesignwaves and operatingenvironment fatigue<br />
waves changes with both the water depth and the actual member/joint<br />
4-6
location. Thus, the method should be used with caution. Detailed<br />
discussionon this method and thecal ibrationeffort is presentedby<br />
Luyties and Geyer (Reference4.7).<br />
4.1.3<br />
Detailed Analyses and Desicm Methods<br />
The detailed analyses and design methods applicable to ship<br />
structures and offshoremarine structuresgenerally follow the same<br />
analyses sequence and incorporate the variables associated with<br />
strength model, time history model and damage computation. The<br />
differencesamongthe various types of detailed analyses are largely<br />
in the methodology implemented to obtain hot-spot stresses, to<br />
develop the stress spectrum and to compute the fatigue life.<br />
Adetailed fatigueanalysis is reconmnendedfor all marine structures<br />
susceptibleto fatiguefailure. While simplifieddesignmethods are<br />
valid in determining the viability of structural details/joints of<br />
typical ships/tankersbuilt from mild steelor offshore platforms in<br />
shallow waters of Gulf of Mexico, a detailed fatigue analysis is<br />
often necessary for other structures. Projected fatigue lives of<br />
a marine structure subjected to cyclic stresses should then be<br />
determined at all critical areas. The uncertainties in fatigue<br />
design and analysisparameters require that more emphasis be placed<br />
on the relative fatigue lives computed than on the absolute lives<br />
obtained. As a result, fatigue analysis is considered to be a<br />
systematic process to identify details/joints susceptible to<br />
failure, and to modify those susceptible areas to yield fatigue<br />
lives substantiallyin excess of the design life. The following are<br />
some detailed analysesoptions that apply to ship structures and to<br />
fixed and mobile marine structures.<br />
<strong>Ship</strong> <strong>Structure</strong>s<br />
A ship that fails to meet simplified fatigue analysis requirements<br />
will not necessarilyhave fatigue failures. It only implies that a<br />
more detailed fatigue analysis is required. Typically, detailed<br />
4-7
analysis is likely to be requiredwhen one or more<br />
are applicable:<br />
of the following<br />
●<br />
The ship structure configuration has unique<br />
characteristics.<br />
●<br />
The structure is built from high strength steel.<br />
●<br />
The use of high strength steel allowed reduction of scantling<br />
sizes basedon strengthrequirementsand due considerationfor<br />
fatigue phenomenawas not given.<br />
●<br />
The operational routes for the vessel are more severe than<br />
typical,making the structuralcomponentsmore susceptibleto<br />
fatigue failure.<br />
The detailed fatigue analysis sequence for ship structures is<br />
similar to fatigue analysesof other marine structures and includes<br />
all of the analyses parameters shown on Figure 3-1. However, the<br />
ship geometry,appreciableforwardspeed and the varying operational<br />
routes require a special effort to determine the ship motions,<br />
applied loads, stress distribution of loads and the long term<br />
distribution of fatigue stresses. Typically, a detailed fatigue<br />
analysis is a spectralfatigue,requiringdeterminationof long term<br />
fatigue stress distribution for each case, accounting for each<br />
seastate and the applicableduration for that seastate.<br />
Although very different from simplified fatigue analyses described<br />
in Section 4.1.2, when the spectral fatigue analysis approach is<br />
modified to representthe long term fatiguestress distributionwith<br />
a shape factor (i.e.Weibull approach), it is sometimes identified<br />
as a simplified fatigue analyses.<br />
Some of the characteristicsof a spectral fatigue analysis and an<br />
alternate Weibull approach are as follows:<br />
4-8
1. SDectral FaticlueAnalvsis<br />
Although spectral fatigue analyses for ship structures and<br />
other often stationary.offshore structures are similar, the<br />
methods used to determine loads and stresses are different.<br />
Ashipstructure requiresdeterminationof hull girder bending<br />
moments in vertical ,and horizontal axes along the entire<br />
longitudinal axis (i.e., hull length). In addition, local<br />
internaland external pressure effects need to be determined.<br />
Most often the appliedwave loads are computed with the useof<br />
linear ship motion theory for wave crestline positions at 90<br />
degree phase angle separation (i.e. in-phase and out-of-phase<br />
componentsof wave). Since the fatigue damage occurs largely<br />
due to normal operating sea states the use of linear ship<br />
motion theory is considered appropriatefor large majority of<br />
spectral fatigue analyses. However, some vessels may have<br />
unique configurations,move at high speeds or be susceptible<br />
to extreme loading fatigue damage. For such vessels the<br />
ability to predict wave nonlinearities and vessel hogging,<br />
sagging and racking effects accuratelymay become important.<br />
In such instances a non-linear ship motion theory may be<br />
preferred over linear ship motion theory. Further discussion<br />
on the specifics of global and local load determination is<br />
presented in Section 5.<br />
The structural analyses needed to convert the in-phase and<br />
out-of-phasecomponents of the load transfer function varies<br />
1argely with the character sties of the structure<br />
configuration. The beam elements used in the structural<br />
stiffness analyses of a discrete system, such as an offshore<br />
platform, may be appropriate for standard ship structures<br />
where other detailed analyses and experience allow reasonably<br />
accurate estimation of local stress distribution. This<br />
approachmay be appropriate if loading is largely due to hull<br />
girder bending moments in vertical and horizontal axis.<br />
However, secondary girder bending moments due to external<br />
4-9
dynamic loads on vessel bottom may be appreciable. In<br />
addition,vesselscontainingcargo such as oilt iron ore etc.,<br />
will have inertial loads on internal tank walls/transverse<br />
bulkheads.<br />
The secondary bending, when appreciable, does affect the<br />
magnitude of local stress distribution. The geometric<br />
complexities also contribute to the difficulty in estimating<br />
local stressdistribution. Since the fatigue life estimate is<br />
function of stress range cubed, the accuracy of fatigue life<br />
estimate is very much a function of the accuracy of local<br />
stress distribution. Thus, a finite element analysis is the<br />
generally reconmnendedapproach to determine the local stress<br />
distributionsfor continuoussystemsuch as ships and tankers.<br />
The stress range transfer functions are obtained to define<br />
response of the ship structure for all sea states covering a<br />
range of frequencies. Thus, in-phase and out-of-phase loads<br />
at each frequency and for each wave direction must be<br />
determined to define the stress range transfer function. In<br />
practice, the effort can be curtailed. A careful review of<br />
load transfer functions should allow selection of several<br />
importantfrequenciesand determinationof stresses for those<br />
frequencies.<br />
The number of constant amplitude stress range cycles to reach<br />
failure is empiricallydefinedas an S-N curve that mayor may<br />
not includethe effect of localizedstress peaking. Thus, in<br />
addition to selecting an S-N curve appropriate for the<br />
structuraldetail and operatingenvironment,the S-N curve and<br />
the structural analyses should be consistent. The stress<br />
range histogram developed and the S-N curve selected for the<br />
location allows determination of fatigue damage per year and<br />
fatigue life by using Miner’s linear cumulative damage rule.<br />
4-1o
2. Weibull AtIDroach<br />
The Weibull shape factor is a stress range distribution<br />
parameter. The Weibull shape factor used with the<br />
characteristicstress range allows carrying out of a fatigue<br />
analyses with a relatively<br />
Since the Weibull approach<br />
few structural analysis cases.<br />
differs from detailed spectral<br />
fatigueanalysisonly in how the stressrange is obtained,the<br />
accuracyof fatigue lives obtainedwith this approach largely<br />
depends on the validity of Weibull shape factor.<br />
The Weibull shape factor may vary between 0.8 and 1.2. If<br />
information on structure and route characteristics are not<br />
available, a shape factor of 1.0 may be used. Shape factors<br />
obtained by calibrating the characteristic stress range<br />
against a spectral fatigue approach indicate that single most<br />
important variable affecting the shape factor is the<br />
environment. In severe North Atlantic and Pacific wave<br />
loadings,the shape factoris higher;the shape factoris also<br />
generally lower for those ship structures with longer hulls.<br />
Although the shape factor may be somewhat different for<br />
differentparts of the structure (i.e. bulkheads, bottom) and<br />
itmay also depend on the number of cycles to failure, further<br />
work is necessary to document those effects.<br />
Fixed and Mobile Marine <strong>Structure</strong>s<br />
The structures referred to in this section are both floating and<br />
bottom-supported steel structures. Most organizations that issue<br />
recommendations, rules, regulations and codes distinguish between<br />
floating and fixed structures because of the differences in their<br />
configurations and the resulting differences in applied loads,<br />
structureresponse,redundancyand accessibilityfor inspectionand<br />
repairs. The requirements vary substantially in scope and detail<br />
from one document to another, but efforts to provide consistent yet<br />
flexible fatigue analysis requirements have been successful.<br />
4-11
In general, the minimumrequirement for fatigue analysis is defined<br />
as the need to ensure the integrityof the structure against cyclic<br />
loading for aperiod greater than the design life. Some documents,<br />
such as the ABS MODU rules (Reference4.8) state that the type and<br />
extent of the fatigueanalysisshould depend on the intendedmode of<br />
operation and the operating environments. Thus, the designer,with<br />
the Owner’s input and concurrence is responsible for developingthe<br />
design criteria, methodology and analysis documentation for<br />
certification of a design that meets the fatigue requirements.<br />
Further discussion on fatigue rules and standards is presented in<br />
Section 4.2<br />
Fixed <strong>Structure</strong>s<br />
As illustrated on Figure 3-5, there are several alternative<br />
approaches to determining the hot-spot stress, stress history and<br />
fatigue life. A flowchart shown on Figure 4-3 illustrates a<br />
deterministicanalysisapplicablefor a fixed platform in amoderate<br />
water depth site subjectedto relativelymild fatigue environment.<br />
The method relies on obtaining hot-spot stresses for one or two<br />
selectedregularwaves and generationof wave exceedancecurves from<br />
the scatter diagram to obtain the stress history. Although this<br />
method requires substantialcomputer use and is considered to be a<br />
detailed analysis, it is also considered to be a screening method<br />
and useful in initial sizing of the structure components.<br />
Amoredesirable alternativeapproachto a deterministicanalysis is<br />
to carry out a spectralfatigue analysis. The appliedwave loads on<br />
a structurecan be generated in the time domain and in the frequency<br />
domain. A structure,such as a flare boom,maybe subjectedto wind<br />
loading only. For such structureswind gust loads can be similarly<br />
generated to evaluate wind-induced fatigue loading. The stress<br />
spectrum is then generated from hot spot stresses, scatter diagram<br />
and specific wave or wind spectra.<br />
One variable in defining the stress spectrum is whether or not to<br />
account for wave spreading. The purpose for distributing the wave<br />
4-12
energy about the central direction by using a ‘spreading function”<br />
is to represent the nature more realistically. Considering the<br />
uncertainties and complexity of implementation,wave spreading is<br />
not generally incorporated into design. While it is a valid<br />
parameter that can be used to more accuratelydetermine the fatigue<br />
lives ofan as-designedor as-builtstructure (seeSection 6.1.4 for<br />
definition of spreading function), it is often unconservative to<br />
neglect it when dynamics are significant.<br />
It is also necessary to assess the significance of short-term<br />
density functionsdevelopedfrom statisticalparameters. The joint<br />
probability of significant wave height and characteristic period<br />
(i.e., each sea state) is used to develop short-term probability<br />
density function of the stress range. This function is often<br />
idealized by a Rayleigh distribution and can be further improved.<br />
This improvement,incorporationof a rainflow correction factor, is<br />
discussed by Wirsching (Reference 4.9). Fatigue damage is then<br />
typically computed for each sea state by using the S-N curve and the<br />
Miner-Palmgren cumulative damage formulation. An alternative to<br />
this approach is based on weighting and sunrningthe probability<br />
density functions to obtain a long-term probability density<br />
function. Total damage can then be computed based on either<br />
numerical integrationor the use of Weibull shape parameter and a<br />
closed form solution. Chen (Reference 4.10) offers a short-term<br />
closed form method that facilitates spectral fatigue analysis.<br />
Spectral fatigue analysis is discussed further in Sections 5, 6 and<br />
7.<br />
Mobile and Stationary Vessels<br />
Both conventional single-hull and twin-hull mobile and stationary<br />
vessels differ from fixed structures in the characteristics of<br />
applied environmental forces and the response of the structure to<br />
these forces. Thus, fatigue analysisof these vessels differs from<br />
that of fixed structures primarily in generation of applied forces<br />
and determination of stresses. Those vessels going from port-to-<br />
4-13<br />
-7 ‘-)<br />
‘><br />
I .-/,
port are also subjectedto differentenvironments,necessitatingthe<br />
use of scatter diagrams applicable for each route.<br />
While a diffraction analysis method may be used to develop the<br />
excitationalforcesdirectly, it is often usedto compute equivalent<br />
hydrodynamic coefficients. These coefficients are then used in<br />
Morison’s formulationto generate wave forces. A typical spectral<br />
fatigue analysis sequence,includinggeneration of dynamic inertial<br />
response loads compatiblewith excitational forces, is illustrated<br />
on Figure 4-4.<br />
Inthe past conventionalsingle-hullvesselswere generallydesigned<br />
conservatively to meet both strength and fatigue requirements.<br />
Following initiationof monitoring programs to obtain wave loading<br />
and stress histories of selected cargo ships and tankers, fatigue<br />
design criteria were further improved. One reason for the<br />
preference of this design approach over the analysis approach is<br />
that most vessels are mobile and subjectedto multitude of site and<br />
time specific environment over their design lives, necessitating<br />
certain conservatism in their design. The use of vessels for<br />
specializedfunctions,such as bow-mooredstoragetanker or a drillship<br />
with a large opening (moonpool) to facilitate drilling,<br />
necessitateddetailedfatigueanalysesto evaluate the other fatigue<br />
sensitive areas throughout the structure.<br />
The detailed fatigue analysis, carried out on increasing number of<br />
floating structures, follow the basic steps shown on Figure 4-4.<br />
While both space frame models with beam elements and finite element<br />
models are used to analyze twin-hull structures, finite element<br />
models are almost exclusively used for single-hull vessels.<br />
4.1.4 Other Methods<br />
Complete ProbabilisticMethods<br />
A reliability-basedfatigue analysis is ideally suited to account<br />
for various uncertainties associated with fatigue parameters.<br />
4-14
Although considered to be an emerging technology and necessitate<br />
time consuming effort, probabilisticmethods have been effectively<br />
utilized in some fatigue analyses. Typically, such a method<br />
accounts for:<br />
● .<br />
Inaccuraciesin defining stresses due to random loadings<br />
●<br />
Uncertainties and observed scatter in S-N data<br />
●<br />
Randomness of failure in the use of simplified models<br />
A probabilistic method recommended by Wirsching (Reference 4.11)<br />
utilizesa fulldistributionalprocedureand the variablesdiscussed<br />
above are assumed to have a log-normal distribution.<br />
Adetailed analysis and design method, based on the use of a finite<br />
element model, to determine environmental loading, vessel response<br />
and load and stress distribution does not need to be a complete<br />
probabilisticmethod. Daidola and Basar(Reference 4.12) d“ scussing<br />
lack of statistical data on ship strengths and stresses recommend<br />
development of a semiprobabilisticanalysis method which does not<br />
require a distribution shape.<br />
Fracture Mechanics Methods<br />
A fracture mechanics method addresses the relationship between<br />
defect geometry, material, and the stress history. The defect<br />
geometry can be accurately modeled with finite elements. Stress<br />
intensityfactorscharacterizingthe defect behavior and the fatigue<br />
crack growth laws allow determination of defect growth<br />
characteristics. Thus, a hypothetical or an actual defect is used<br />
as the basis for determining the fatigue life and identifying the<br />
necessary corrective measure.<br />
The initial defect size and location and the stress intensity are<br />
very important parameters in determining crack growth period to<br />
failure. The fracturemechanics approach is a useful tool to assess<br />
4-15
the sensitivity of fabricationdefects in determining the fitnessfor-purpose<br />
of the component. This concept, first described by<br />
Wells (Reference 4.13), allows engineering assessment of weld<br />
defects to determine those defects that require repair as well as<br />
those that are considered fit-for-purposewithout a repair.<br />
4.2<br />
FATIGUE RULES AND REGULATIONS<br />
The primary objective of the various recommendations, rules,<br />
regulations and codes applicable to marine structures is to ensure<br />
that the design and analysis process results in construction of<br />
marine structures.that can resist both extreme loads and cyclic<br />
operating loads and have adequate fatigue lives.<br />
Rules and recommendations issued by classification societies and<br />
certifying agenciesmay representthe minimum requirements based on<br />
research and development. The hull girder design criteria given by<br />
each of the four leading classificationsocieties (American Bureau<br />
of <strong>Ship</strong>ping, Lloyd’s Register of <strong>Ship</strong>ping, Bureau Veritas and Det<br />
Norske Veritas) is very similar and differs only in some of the<br />
details. While the design basis primarily addresses stillwater and<br />
wave-induced bending moments, some discussion on dynamic stress<br />
increments and fatigue file assessment is often provided. Recent<br />
research and development efforts have produced several recommended<br />
fatigue design guidelines. Rules and reconunendationson offshore<br />
structures are very specific on fatigue design. Guidelines are<br />
provided to carry out both simplified and detailed analyses.<br />
Commentary to<br />
development of<br />
such guidelines also provide background for the<br />
fatigue design methods.<br />
Fatigue design methods chosen vary depending on several factors,<br />
including the owner’s design philosophy. Most fatigue design<br />
methods are variations of a method based on application of S-N<br />
curves representingthe fatigue strengthof similardetails/joints.<br />
A basic S-N curve applicablefor a given detail/joint also requires<br />
adjustments to incorporate the influence of variables. Although<br />
4-16
many design rules implement this approach, the recommended S-N<br />
curves are often different from each other.<br />
Assessment of defects detected during fabrication, or cracks<br />
discovered while the structure is in service, is best accomplished<br />
using fracturemechanicsand crackgrowth laws. Fitness-for-purpose<br />
considerations will then directly affect repair programs and<br />
inspection schedule.<br />
The reconunendations,rules, regulations and codes that apply to<br />
fatiguedesign have evolvedover the past 20 years, and severalhave<br />
been revised or reissued in the last five years. These documents<br />
are discussed briefly below as they apply to vessels and other<br />
marine structures.<br />
The American Welding Society (AWS) and American Institutefor Steel<br />
Construction (AISC) fatigue design specifications (Reference4.15)<br />
provide the basis for approximate fatigue design based on S-N<br />
curves. However, unless the method developed accounts for the most<br />
likely loads and other uncertainties, various critical and noncritical<br />
fatigue cracks are.likely to occur.<br />
Most documents on fatigue provide substantial flexibility in<br />
carrying out comprehensivefatigue design and analysis, while also<br />
incorporating extensive guidelines. Various DnV documents on<br />
specific types of structures such as Steel <strong>Ship</strong>s (Reference4.16),<br />
Tension Leg Platforms (Reference4.17, Part 3, Chapter 6) and Fixed<br />
Steel Platforms (Reference4.17, Part3, Chapter 4) provide general<br />
guidelines and referto acomprehensive documenton fatigue’analysis<br />
(Reference 1.7). The UEG Recommendations (Reference 1.8) are<br />
similar to U.K. DEn Guidance Notes (Reference 1.6), differences<br />
largely limited to the revisions introduced in the latest (fourth)<br />
edition of Guidance Notes.<br />
4-17<br />
-7 “7
4.2.1 Applicable Methods<br />
Simplified Analvsis Methods:<br />
ABS provides a simplified allowable stress method, suitable for<br />
fatigue screening of tankers. As discussed in Section 4.1.2, the<br />
method allows substantial flexibility for engineering judgement.<br />
Both DnV (Reference 1.7) and API (Reference 1.5) provide for<br />
simplified fatigue assessmentof fixed offshore platforms. The API<br />
approach requires that the peak hot-spot stresses for the fatigue<br />
design wave do not exceed the allowable peak hot-spot stresses.<br />
This simplifiedapproach is based on detailed fatigue evaluation of<br />
typical Gulf of Mexico jackets in less than 400 feet water depth,<br />
with natural periods less than 3 seconds. Variations in structure<br />
geometry, and in the approximationsintroduced,make the simplified<br />
analysis best suited for screening of similar structures for<br />
sensitivity to fatigue loadings.<br />
The simplified DnV fatigue analysis is useful if the long-term<br />
stress distribution for a given area is not known. This simplified<br />
method provides an empirical relationshipto determine the maximum<br />
allowable stress range during a 20-year life as a function of S-N<br />
curve parameters, long-term stress distribution as function of a<br />
Weibull parameter and the complete ganunafunction. This method is<br />
quite useful as a design parametric tool because it allows<br />
assessment of joint configurations for weld type and plate<br />
thicknesses and facilitates selection of details least susceptible<br />
to fatigue failure. However, since it is difficult to define<br />
accurately and/or conservativelythe long-term stress distribution<br />
as a function of a Weibull parameter, the computed fatigue lives<br />
should be used cautiously.<br />
4-18
Detailed Anal.vsisMethods<br />
The detailed fatigue analysis sequence for ship structures is<br />
similar to fatigue analyses of other marine structures. While<br />
appreciableforward speed and ship motions complicate determination<br />
of cyclic stress distributions, finite element based spectral<br />
fatigue analyses approachesrecormnendedby classification societies<br />
are similar to those recotmnendationsapplicable to offshore<br />
structures.<br />
The recommendationsand rules applicableto fixed offshore platforms<br />
are generally quite flexible in the use of applicable analysis<br />
methods. To ensurestructural integrity,al1 cyclic 1oads that wil1<br />
cause appreciablefatiguedamage must be considered, includingthose<br />
due to transportation and all in-service loading for stationary<br />
structures. Several methods of determining the applied loads are<br />
acceptable to DnV (Reference 1.7), API (Reference 1.5) and the DEn<br />
(Reference 1.6). For fixed platforms, both deterministic and<br />
spectral methods can be used to generate the applied loads and<br />
determine the hot-spot stresses. However, a spectral analysis<br />
approach is often recommended to properly account for the wave<br />
energy distribution over the entire frequency range.<br />
Comparative studies carried out on a benchmark API platform,<br />
utilizing four separate approaches (one deterministic and three<br />
spectral), yielded large scatter of fatigue lives due to inherent<br />
differences from one analysis approach to another. Such results<br />
justify the philosophy conveyed in most recommendations and rules,<br />
including API (Reference 1.5) and DEn Guidance Notes (Reference<br />
1.6), that the fatigue analysis be treated as a systematic<br />
parametric analysis, requiring determination of the sensitivity of<br />
various parameters that affect fatigue lives.<br />
.<br />
4-19
4.2.2 SCFS, S-N Curves and Cumulative Damaqe<br />
Stress Concentration Factors (SCFsl<br />
It is desirable that the discontinuitiesthat result in high stress<br />
concentrationsbe evaluated by laboratorytesting or finite element<br />
analysis. But because these methods of obtaining stress<br />
concentration factors (SCFS) are often not practical, empirical<br />
formulations are widely used to determine the SCFS. Most<br />
reconunendationsand rules provide general guidelines on the use of<br />
SCFS and refer other reference documents. Lloyd’s Register was<br />
responsible for carrying out extensive strain-gaged acrylic model<br />
tests and developing SCF formulas. These empirical formulas are<br />
incorporated into Lloyd’s Register Rules (Reference 4.18).<br />
Assessment of various SCF formulas is discussed further in Section<br />
5.4 and Appendix C.<br />
S-N Curves<br />
For the purposes of defining fatigue strength as a function of<br />
constant amplitudestress and the number of cycles to reach failure,<br />
welded joints are divided into several classes. DnV (Reference1.7)<br />
provides an S-N curve identifiedas “T-curve”for all tubular joints<br />
and eight other classes to define other joints, depending upon:<br />
●<br />
●<br />
●<br />
The geometrical arrangement of the detail<br />
The directionof the fluctuatingstressrelative to the detail<br />
The method of fabrication and inspection of the detail<br />
API provides two S-N curves to define the tubular joints. The X-<br />
curve presumes welds that merge with the adjoining base metal<br />
smoothly (i.e., profile control), while the X’-curve is applicable<br />
for welds that do not exhibit a profile control. The API X-curve<br />
was originallybased on the 1972 AWS test data and has been upgraded<br />
based on later editions ofAWS D1.1 (Reference4.14).<br />
4-20
The DnV X-curve and the DEn Guidance Note Q-curve of 1977 were also<br />
based on the original AWS test data and the recommended S-N curve.<br />
Recent experimental work carried out in Europe has provided<br />
additional data on fatigue strengthof tubular joints. Statistical<br />
evaluation of these test results provided the basis for revision of<br />
both the DnV (Reference 4.17) X-curve and the DEn Guidance Notes<br />
(Reference1.6) Q-curve. As illustratedon Figure 4-5, the slopeof<br />
the new T-curve is steeper and typically results in lower lives,<br />
often necessitatingan increaseinwall thickness. The DEn Guidance<br />
Notes reconunendedT-curve is identical to the DnV T-curve up to 10<br />
million cycles for catholicallyprotected areas.<br />
The basis for the revision of the S-N curves by both DnV and DEn is<br />
primarily due to evaluation and assessmentof test data. While the<br />
AWS data are based on some plate and some small-diameter thin-wall<br />
sections,the Europeandata are obtainedmostly from largerdiameter<br />
tubulars with 5/8 inch and 1-1/4 inch (16 mm and 32 mm) wall<br />
thicknesses. It appears that an inverse log-log slope of 3.0<br />
(versus4.38 for the API X-curve)was chosen for the T-curve because<br />
of the scatter of data and to ensure consistency with the British<br />
Standards BS5400. Basedon statisticalevaluationoftest data and<br />
Gurney’s (Reference4.19) analyticalstudieson plate thickness,the<br />
T-curve is adjusted due to changes in plate thickness.<br />
Although the DnV (Reference4.17) document states that all tubular<br />
joints are assumed to be of Class T, an X-curve is also considered<br />
acceptable,providedweld profiling is carried out. The comparison<br />
of the API X-curve and the T-curve (Figure 4-5) shows that the two<br />
curves intersect at about 500,000 cycles and would yield similar<br />
1ives for a plate thickness of 1-1/4 inch (32 mm). However, for<br />
plate thicknesses greater than 1-1/4 inches the use of a T-curve in<br />
the computation of fatigue lives will result in shorter lives.<br />
4-21
Cumulative Damaae<br />
The use of the Palmgren-Miner 1inear damage rule is considered<br />
appropriate by all of the recommendations, regulations and rules.<br />
A cumulative sum of the number of cycles at each constant stress<br />
divided by the number of cycles to failure should always be less<br />
than l.Oforthe desired service (design)life. While this value is<br />
directly tied to the S-N curve selected,the desirable ratio (i.e.,<br />
safety factor) of fatigue to service life is not always specified.<br />
The API reconunendedfatigue life is at least twice the service<br />
life. For criticalmembers that may affect structure redundancy and<br />
integrity,API recotmnendsthe use of higher fatigue to design life<br />
ratios.<br />
The DEn Guidance Notes reconunendadditional safety factors to<br />
account for structural redundancy and the implications of fatigue<br />
failure on the structure. However, no specific safety factor is<br />
recommended.<br />
4.2.3 Fatiaue Analvsis Based on Fracture Mechanics<br />
The fatigue crack propagation analysis is typically used to assess<br />
crack growth and fitness-for-purposeof defects discovered at the<br />
fabrication yard. Test data on crack growth can also be used to<br />
determine fatigue lives. The DnV CN 30.2 document (Reference 1.7)<br />
provides a crack growth rate data and fracture mechanics-based<br />
procedure for fatigue analysis and design.<br />
Whether the welded joint details have surface or root defects, the<br />
growth of such defects into fatigue cracks depends on several<br />
factors, including joint connection geometry, cyclic stress range<br />
history, weld profile and defect size. The equations provided to<br />
solve for the number of cycles to reach fatigue failure containmany<br />
parameters and allow evaluation of various joint and defect<br />
geometries. As an example, butt weld toe defects in a connecting<br />
plate whether in air or seawater, can be assessed with and without<br />
4-22
ending restrictions. Cruciform and tubular joint defects can be<br />
similarly assessed. The DnV CN 30.2 document provides standard<br />
crack growth parameters to facilitate a fatigue analysis based on<br />
fracturemechanics. LotsbergandAndersson (Reference4.20) further<br />
discuss fracturemechanics-basedfatigueanalysis and i11ustratethe<br />
approach with several examples of crack growth calculation.<br />
4*3<br />
CURRENT INDUSTRY PRACTICES<br />
Current industry design practices for marine structures are<br />
significantly more advanced than the design practices of only 20<br />
years ago. The extensive use of ever more powerful computers and<br />
the deve~opmentofa wide range of softwarepackages has facilitated<br />
the design and analysis of marine structures. Research work on<br />
long-term ocean environment, model basin studies on structure<br />
motions, structurecomponentmember testingfor stressdistribution,<br />
buckling,yielding and fatigue failureall have been instrumentalin<br />
developing better and more effective means of designing marine<br />
structures. Structural reliability research has also provided the<br />
means to incorporate the large number of uncertainties into the<br />
analysis and design effort.<br />
Fatigue analysis and design is perhaps the part of the overall<br />
analysis and design effort that benefits the most from these<br />
developments. Since the hot spot stress is a primary variable<br />
influencingfatigue life, analyticaland experimentalprograms have<br />
been carried out to helpdevelopdetails/jointswith lower hot spot<br />
stresses. Good design detailingwithout fabricationquality is not<br />
adequate. Thus, parameters affecting fabrication quality are<br />
incorporated into current design practices and fabrication<br />
specifications. It is feasible to analyze each joint of a discrete<br />
system such as a fixed platform. However, a continuous system, such<br />
as a ship, has thousands of details/joints and lends itself to a<br />
selective analysis. Current industry practice is to select number<br />
of cross-sections along the hull and analyze a dozen or more<br />
details/joints at each cross-section.<br />
4-23
Although additionalresearch is needed to expandthe availabledata,<br />
the industry has the ability to incorporate most sophisticated<br />
analysis procedures into fatigue design. The degree of<br />
sophisticationneeded to design a marine structure that has fatigue<br />
life in excess of its design life depends both on the structure and<br />
its operating environment. Thus, the effort necessary may be<br />
grouped into ordinary and special designs.<br />
4.3.1 Ordinarv Desiqns<br />
All marine structurescan be designed effectivelyby ordinary means<br />
if those structuresare not going to be subjectedto any appreciable<br />
fatigue environment. For example, offshore platforms in relatively<br />
shallow waters may be susceptibleto typhoon/hurricaneloading but<br />
less susceptibleto cyclic loadings that cause fatigue, eliminating<br />
the need for comprehensivefatigue analyses. Such structures can be<br />
designed for other loadingconditionsand checked against fatigueby<br />
approximate allowable stress procedures.<br />
The design of ships still is largely based on design rules (such as<br />
ABS, Reference4.1)developed by combiningtheoreticalknowledgeand<br />
design experience. Most ships in-serviceare designed to meet these<br />
rules and other fatigue design procedures (References 1.2 and 4.3)<br />
to ensure that the component details meet fatigue requirements.<br />
This approach has been quite satisfactoryfor most ships. Recently<br />
built vessels, especially large tankers built in the last several<br />
years have exhibited substantial fatigue problems. These problems<br />
may be largely attributed to the use of high strength steel,<br />
resulting in the use of lower plate thicknesses and yielding higher<br />
stress levels. As a result, detailed fatigue analysis and design<br />
procedures are implementedon more and more vessels.<br />
4.3.2 Sr)ecializedDesicms<br />
Those vessels with specialized functions and/or configurations, or<br />
which are likely to be moored in a specific area for an extended<br />
4-24
period, are also designed tomeet the rules and other fatigue design<br />
procedures. However, such vessels also require spectral fatigue<br />
analysisto define the loadings,response and stress distributions.<br />
Often, model basin tests are also carried out to validate the<br />
applied loadings and motions.<br />
Stationary marine structures are generally unique and have<br />
specialized functions. Since the design criteria and functional<br />
requirements dictate the general configuration of such structures,<br />
each structuremust be analyzedthoroughlyto generate the loads,to<br />
determine the response to these loads, and to determine its<br />
susceptibility to.fatigue. Most specialized structures require<br />
spectral fatigue analysis.<br />
4.4<br />
SENSITIVITY OF FATIGUE PARAMETERS<br />
Fatiguedesign and analysis parametersdiscussed in Sections 3.1 and<br />
3.2 i11ustrate the general interaction of these parameters. The<br />
specific interactions and the actual sensitivities of these<br />
parameters depend largely on the structure’s global configuration,<br />
joint configuration and details, material characteristics,<br />
fabrication quality and the design requirements other than fatigue.<br />
Therefore, fatigue analysis and design efforts often incorporate<br />
flexibilityto carry out parallel studies to assess the sensitivity<br />
of major parameters that affect fatigue life.<br />
Although the parametersillustratedon Figure3-l are all important,<br />
some of the more important parameters for fatigue life improvement<br />
are:<br />
●<br />
Enhance fabrication quality and minimize defects<br />
●<br />
Minimize applied loads and motions to minimize nominal cyclic<br />
stress ranges<br />
●<br />
Optimize the design for uniform load distributions<br />
4-25
●<br />
Optimize the design details to minimize SCFS<br />
Another parameter that is not important to the actual fatigue life<br />
but very important to the computed fatigue life is the analysis<br />
method and the assumptionsused in the analysis. Although there is<br />
no substitute for experience, comparative studies carried out by<br />
others should be utilized and the analysis method selected and the<br />
assumptionsmade should be applicableto the marine structure being<br />
designed.<br />
4.5<br />
FATIGUE DESIGN AND ANALYSIS CRITERIA<br />
Fatigue design and analysis criteria are generally covered in one<br />
chapter of the structural design basis document. Fatigue criteria<br />
may also be jointly prepared by the engineer and the owner as a<br />
separate design brief to document the fatigue design and analysis<br />
basis.<br />
4.5.1 Basis for the Preparation of Criteria<br />
The design and analysis criteria serve the purpose of clearly<br />
defining the work to be undertaken. Three primary variables that<br />
affect the fatigue design and analysis criteria are:<br />
●<br />
The owner’s requirements for work scope<br />
and schedule<br />
●<br />
The engineer’s assessment of the<br />
sensitivity to fatigue and the required<br />
marine structure’s<br />
level of analysis<br />
●<br />
The role of classification societies<br />
The owner, engineer and classification society all agree that the<br />
design and analysis should lead to quality fabrication and ensure<br />
the structural and operational integrity of the marine structure<br />
throughout its design life. To accomplish these goals, a design<br />
should provide a balance between efficiency and redundancy and also<br />
4-26
incorporates inspection-strategy(References4.21, 4.22 and 4.23).<br />
As a result, the design effort must incorporate consideration of<br />
global response, alternate load paths, local stress distributions,<br />
structural detailing, material selection, fabrication procedures,<br />
etc., to ensure that the structure’s fatigue sensitivity is<br />
minimized. However, the extent of the fatigue analysis is a<br />
function of cost as well as technical considerations. A marine<br />
structure costing $1 million and another costing $50 million will<br />
not be analyzed to the same extent. In lieu of extensive analysis,<br />
approximate analysis combined with greater safety factors is<br />
appropriate for less costly structures.<br />
A fatigue criteriadocument maybe very general, stating the design<br />
and analysis objectivesand the classificationand/or certification<br />
requirements. It can also list every method to be implemented and<br />
every assumption to be made in the execution of fatigue analysis.<br />
Most often the document will specify the scope of work, define<br />
overall methodology, and provide the data necessary for fatigue<br />
analysis.<br />
A typical fatigue design and analysis criteria table of contents<br />
contains the following elements:<br />
1. INTRODUCTION<br />
1.1 Objectives<br />
1.2 Scope<br />
1.3 Third Party Inputs<br />
2. MODELLING<br />
2.1 Loads Model<br />
2.2 Mass Model<br />
2.3 Stiffness Model<br />
3. OCEAN ENVIRONMENT<br />
3.1 Applicable Sea States<br />
3.2 RecommendedWave Theories<br />
3.3 Wave Directionality and Distribution<br />
3.4 Wave Scatter Diagrams and Recorded Data<br />
3.5 Wave Spectra<br />
4. PRELIMINARYANALYSIS<br />
4.1 Applicable Method<br />
4.2 Accuracy of Results<br />
4-27
5.<br />
6.<br />
7.<br />
8.<br />
DETAILED ANALYSIS<br />
5.1 <strong>Structure</strong> Motions and Loading<br />
5.2 Calibration of Loading<br />
5.3 Nominal Stresses<br />
5.4 Applicable-SCF Formulations<br />
5.5 S-N Curve and Fatigue Damage Calculation<br />
FATIGUE SENSITIVITY STUDIES<br />
6.1 Study Parameters<br />
6.2 Areas Selected and Extent of Study<br />
REFERENCES<br />
APPENDIXES<br />
4.5.2 Almlicable Software<br />
The analysis method chosen has to be compatible with the computer<br />
softwares available. Since a wide range of computer software is<br />
available, the analyses method and the software should be chosen<br />
based on structure configuration, applicable environmental loads,<br />
structuralresponseto appliedloading,stressdistributionpatterns<br />
and susceptibilityto fatigue failure.<br />
,.-.<br />
The softwarepackagesnecessaryto carry out the analysis and design<br />
functions should facilitate determination of:<br />
●<br />
●<br />
●<br />
●<br />
Ocean environment loads<br />
<strong>Structure</strong> motions<br />
Structural analyses and stress distributions<br />
Stress history and fatigue damage evaluation<br />
While there are special-purpose software programs such as SEALOAD<br />
(Reference 4.24) to generate wave loads and SHIPMOTION (Reference<br />
4.25) to determine motions, these and other software programs are<br />
often a component of larger generalized systems. A large system<br />
will facilitateexecutionof all functionsfromwave load generation<br />
to fatigueclamageassessmentwithin the system,eliminatingthe need<br />
for external data transfers.<br />
4-28
‘Thereare numerous finite element programswell-suited for detailed<br />
analyses and design of continuous structures such as ships,<br />
semisubmersibles and TLPs. The best known of these programs in<br />
public-domain are ANSYS, NASTRAN, SAPIV, DAISY and SESAM. Mansour<br />
and Thayamballi (Reference 4.26) provide a survey of computer<br />
software and they discuss programs specifically developed for the<br />
marine industry.<br />
4.5.3<br />
Fatique Versus Other Desiqn and Scheduling Requirements<br />
Fatigue analysis and design is only one of many aspects of the<br />
overall analysis and design effort. Because the final as-designed<br />
structure must meet many varied pre-service and in-service<br />
requirements, the fatigue design effort reflects the necessary<br />
interactions among various activities. The design criteria<br />
typically includespecific assumptionsand procedures to coordinate<br />
such activities. As an example, some of these interactions for a<br />
fixed offshore platform design project are as follows:<br />
●<br />
A computermodel used to generateextreme environment loads is<br />
also used for fatigue analysis,with changes in hydrodynamics<br />
coefficients and foundationmatrix as necessary.<br />
●<br />
A computer analysis model used for stiffness analysis should<br />
not account for the effect of thickened brace stubs, but the<br />
stress ranges used for fatigueanalysis should account for the<br />
increased thickness.<br />
The overall design schedule often dictates that fatigue design and<br />
analysis be carried out inunediatelyafter the structure’s general<br />
configurationis finalized. But the fatiguedesign must incorporate<br />
flexibility,to allowfor significantconfigurationrevisionsduring<br />
the detailed design, which will affect both the applied loads and<br />
the stress ranges. The desired flexibility is often obtained by<br />
carrying out parametric studies to identifythe effects of changes,<br />
4-29
and by providing sufficientmargin when determining the desirable<br />
fatigue lives.<br />
4-30
LOCATIONs oF FATIGUE CMCKS<br />
Figure 4-1 Typical Fatigue Sensitive <strong>Ship</strong> <strong>Structure</strong> Details<br />
DesignProcedure<br />
‘“m<br />
Choose a loadingshape paratmter k,<br />
‘ftheweibull ‘distribution<br />
2.<br />
*<br />
<strong>Ship</strong> Detail Identify the number designation of<br />
Catalog the critical details. (Fi s. A.1<br />
through A.12 of Appendix A 7<br />
3“v ‘ind<br />
4“w’<br />
‘“v<br />
‘“- oflo-,.<br />
: I) 5-N curve slope, m, of detail<br />
2)‘:z’’’:’tressrange<br />
(See Table B.1 and Fig. B.1 of Appendix B)<br />
Find random load factor,
ICOMPUTER MODEL I MASS MODEL I<br />
I<br />
I<br />
WAVE LOADING<br />
SPECIFICATIONOF WAVE GRID<br />
WAVE THEORY, MARINE GROWIH, DRAG<br />
& INERTIACOEFFS,WAVE DIRECTION,<br />
WAVE PERIOD AND WAVE HEIGHT<br />
TRANSFER OF WAVE LOADS<br />
FROM NON-STRUCTURAL TO<br />
STRUCTURAL MEMBERS<br />
t<br />
DEFINITIONOF ENVIRONMENTAL DATA<br />
EXCEEDANCE CURVE, DIRECTIONAL<br />
PROBABILITIES,ANNUAL ANO<br />
FAIGUE WAVES<br />
L===l<br />
DEFINllIONOF FATIGUEPARAMETERS<br />
FATIGUE LIFE<br />
EVALUATION<br />
Figure<br />
4-3 A Typical Detenai.nistic Fatigue Analysis Flow Chati
TW=’T<br />
——— ———___ __<br />
r<br />
I<br />
I<br />
I<br />
I<br />
I<br />
I<br />
I<br />
I<br />
I<br />
I<br />
I<br />
MOTIONS MODEL<br />
● W/ m W/O DiffmcUrn Andpls<br />
● LOd -muon<br />
STRUCTURES MODEL<br />
1<br />
I<br />
I<br />
I<br />
I<br />
I<br />
I<br />
I<br />
———— ————<br />
I r 1<br />
TESllNG<br />
I I STRAIN GAUGE j<br />
I L ——— ——— ——<br />
1 I<br />
L ———— —.—— ———— .J<br />
I<br />
I<br />
I<br />
I<br />
EMPIRICAL<br />
ANALYTICAL I<br />
A<br />
STRESS RAO’S<br />
u FORMULATIONS F.E.A.<br />
n<br />
I<br />
: I I<br />
% ———— ———— ———— ——— J<br />
~<br />
4<br />
——— ——— —.. ——_ ___ __<br />
I<br />
HOT SPOT<br />
STRESSES<br />
I<br />
___<br />
__<br />
...<br />
I<br />
/<br />
> DEFINE DEFINE DEFINE<br />
u<br />
p WAVE SPECTRUM SCATTER DIAGRAM DIRECTIONALPROB.<br />
~J<br />
Xgo<br />
%2<br />
&<br />
VI<br />
STRESS SPECTRUM<br />
——— ———— ———— ———— _. —— _<br />
+ \<br />
I DEFINE I<br />
1 S-N CURVE 1<br />
FATIGUEDAMAGE<br />
ANALYSIS<br />
Figure 4-4 A ~lcal<br />
Spectral Fatigue Analysis Flow Chart
tl<br />
MASSMODEL<br />
1 I r<br />
*<br />
RIGID 600Y<br />
MASS MATRIX<br />
HYTIR&D#MtC<br />
+ E m<br />
OYNAtilC<br />
WSSEL ACCELERATIONS MOTION ANALYSIS<br />
LOADINGS:<br />
AND<br />
ANO WAkE LOAD<br />
lNERTtA + WAVE WAX LOAOINGS - GElN&R~TlC&4S<br />
t<br />
I<br />
nSTIFFNESS<br />
ANALYSIS<br />
FOR ALL LOADINGS<br />
I<br />
WAVE HEl&T<br />
I<br />
t<br />
GENERATESTRESS RESPONSE<br />
AMPLITUDEOPERATORS(I?AO’S)<br />
ANO SELECTCRIICAL MEMBERS<br />
FOR FATIGUE ANALYSIS<br />
v<br />
DEFINE ENMRONMENTALOATA<br />
SCATTER DIAGRAM. YfAW<br />
SPECIRLIM AND<br />
DIRECTIONAL PROBABiU~ES ‘<br />
t<br />
I<br />
DEflNE S-N CURVE ANO<br />
STRESS CONCENTRATIONFACTORS<br />
I<br />
PERFORMSPECTRAL<br />
FATIWE ANALYSIS<br />
I<br />
F@re 4-4 A Typical Spectral Fatigue Analysis Flow Chart<br />
(<br />
.–k.)<br />
---F’
1<br />
—.<br />
—<br />
—<br />
—<br />
—<br />
-.. -<br />
.. -<br />
. -<br />
I<br />
—<br />
7- +<br />
x<br />
—<br />
\<br />
—<br />
--<br />
—<br />
+<br />
—<br />
—<br />
..<br />
.<br />
——— - -<br />
—<br />
—— —.<br />
—- . —.<br />
—.<br />
.——+..<br />
... . —-<br />
—<br />
—<br />
—<br />
-.<br />
,..<br />
-. .<br />
— .<br />
. —<br />
—-<br />
. ——<br />
. —.. .— -<br />
—, .,<br />
..-. -<br />
. . ,——. . — — .<br />
.-—. . . —.— -— .— --<br />
... . . -. --- -<br />
,..<br />
“Iok 105<br />
6<br />
Enduran#(cycles)<br />
UuxK,<br />
-dtism~<br />
K, s-<br />
m ~<br />
log,, Io& *,@ IQ&<br />
B 2343x101S 15.3697 35.3~ 4.o 0.1S21 0.414M l.olxlo~ 102<br />
C L012x1014 14.0342 32.3153 3.5 0.2Ml 0.47m<br />
D 3.9::x10U IZ.6M7 29.0144 3.0 02095 0.4824<br />
E 3_23Wx10U 12.3169 2S.S216 3.o 0.250J 0.5~<br />
F 1.2s9x101z 1223X3 2g.1~0 3.o 0.2133 0.5~<br />
K*<br />
Nlmmi<br />
4.23X1(+3 78<br />
1.5M012 53<br />
l.wxlo~ 47<br />
M3X1012<br />
F21231z10U 12.~ 27.S38? 3.o 0.22~ 0.52U 0.43x101Z 35<br />
G 0.56&c10i1 11.732S 27.0S14 3.0 0.1793 0.4129 0.23x101J 29<br />
WO~lOU 11JS42 26.6324 3.o 0.1346 0.42S1 0.16x101z 25<br />
T 4=s1012 12.- 29.1520 3.0 0.- OS’20 1.46xl@ 53*<br />
w<br />
●<br />
l~ti~w<br />
Fm~*eTc~~k_ti&lok<br />
IwAN) - t 2.MM . 0.Z43M . 31~1@~<br />
Figure 4-5 me<br />
DnV X- and the New T-Cumes
“k- N<br />
++’<br />
b.<br />
--<br />
0<br />
N<br />
0<br />
NOTE -<br />
PERMISSIBLE CYCLES OF LOAD N<br />
These curves may be represented mathemahcall y as<br />
N= 2x109 30 -m<br />
() JZ<br />
where N is the permmslble number of cycles for apphed cyclic stress range Jd. with A aref and m as listed below.<br />
1 IIref m<br />
sTRESS RANGE AT<br />
INvERSE<br />
ENDURANCE LIMIT AT<br />
CURVE<br />
2 MILLION CYCLES LOG-LOG SLOPE 2W MILL1ON CYCLES<br />
x 14.5 ksi [100 MPa) 4.3s 5.o7 ksi (35 MPa)<br />
x’ 11.4 ksi (79 MPa) 3.74 3.33 kai (23 MPa)<br />
Figure 4-6 API X- and X ~-roes and DnV T-Cume
TOP[C<br />
U.K.DEPARTMENTOF ENERGY(OEn)<br />
AMERICANPETROLEUMINSTITUTE<br />
GENERALCONSIDERATIONS<br />
o Fatiguelife Life*~Servlce Life Life> 2x Serv~ceLife<br />
o FatigueLoading<br />
o FatigueAnalyslsjDes~gn<br />
All cyclicloads(Ref.21.2.1OC) (Ref.5.2.5)<br />
- SimplifiedMethod No Yes,allowablestressmethod<br />
appl~cableto Gulfof~exlco (GOM)<br />
(Ref.5.1.1)<br />
- DetailedAnalysis Recommended Recommendedfor:<br />
waterdepth~400 ~t (122m),or<br />
* lifeshouldnotbe 620years - platformperiod> 3 see,or<br />
andan additionalfactoron - environmentharsherthanGOM<br />
Ilfeis recommendedwhen (Ref.5.1.2)<br />
structuralredundancyis<br />
ln~dequate(Ref.21.2.10f)<br />
DETERMINATIONOF STRESSES<br />
o Ob.ject~ve To determinecycl~cstressranges To determineCYCIICstresses<br />
(j.e.,meanstressesare neglected- properlyaccountingactualj<br />
Ref.21.2.11)<br />
distributionof waveenergyover<br />
entirefrequencyrange,spectral<br />
analysistechniquesare recommended.<br />
o Modeling No spectficreference Spaceframeanalysistoobtaln<br />
structuralresponseandstress<br />
distribution(Includlngdynnmlc<br />
effects)<br />
o Analysis A detailedfatigueanalyslsallowln9 Typically,spectralanalysisto<br />
eachcrltlcalareato be considered. determinestressresponsefor each<br />
sea state.<br />
J<br />
Page1 of 5<br />
.<br />
Figure4-7 Comparisonof Recommendations<br />
U.K.GuidanceNotesandAPI RP 2A
~ AMERICAN ETPOLEUMlNSTITUTE<br />
)tiotSpotStress RangeIs theproductof the nom~na~ RangeIsohtalnedby multiplyingthe<br />
stressrangeIn the braceandthe nominalstressrangeat’tubular<br />
SCF. ItIncorporatesthe effectsof jointby SCF.<br />
overalljointgeometrybut.omitsthe<br />
stre<strong>ssc</strong>oncentratingInfluenceof<br />
theweld itself.<br />
iTRESSCONCENTRATION<br />
‘ACTORS(SCFs~<br />
} Scf ho referencesgiven SCFSdefinedarebaseduponmodl~led<br />
Kelloggformulas forchord and<br />
Marshallformula[ forbrace 1 (Ref.<br />
C5.1,Table5.1.1-1)<br />
ISimpleJoints- Nodal<br />
JEmpiricalEquations<br />
OtherJoints<br />
iTRESSHISTORY<br />
Nonedefined<br />
Noned~flned<br />
SCFSdefinehot spotstresses<br />
Immediatelyadjacentto the olnt<br />
intersection(0.25”toO.1d from<br />
weldtoe,Ref.C 5.4)<br />
K, T, Y andX jointsdefinedfor<br />
axial,In-planeandout~ofplane<br />
loading<br />
Recommendsa braceSCF~6 (Ref.5.5)<br />
)WaveClimate<br />
Page2 of 5<br />
i<br />
Figure4-7 Comparisonof Recommendations<br />
U.K.GuidanceNotesandAP1fW 2A<br />
Waveclimatesmay be derivedfrom<br />
bothrecordeddataandhlndcasts.<br />
Aggregateof allsea statesto be<br />
expectedoverthe longterm<br />
condensedIntorepresentativesea<br />
states.<br />
A sea state,characterized by wave<br />
energyspectrumandprobabilityof<br />
occurrence,may be definedby:<br />
o Twoparameterscatterdiagrams<br />
o Directionalscatterdiagrams<br />
o i)lrectlonalscatterdiagramswith<br />
spreading
:1<br />
U.K.DEPARTMENTOF ENERGY(DEn)<br />
AMERICANPETROLEUMINSTITUTE<br />
The long-termwaveheight<br />
distributionmay be representedby<br />
the sumof twoWelbulldlstrlbutlons<br />
one fornormalandanotherfor<br />
hurricaneconditions(Ref.Ftg.<br />
C5.2.1)<br />
FATIGUESTRENGTH<br />
o Definedby S-ticurvesbasedon Mean-minus-two-standard devlatlon X curveIs suffclentlydevaluedto<br />
experimentaldata curves[Ref.21.2.10.f) accountforthickness/sfzeffect.<br />
log(tq= log (K1]-do -m.log(S8)<br />
A slopeof m=-3adoptedbasedon<br />
data<br />
o TubularJoints<br />
- Recommended Fullpenetrationwelds- T curve Smoothweldmetalmergingwith<br />
(Ref.21.2- 12a)<br />
parentmetal- X curve,otherwiseX’<br />
curve(Ref.C5,4)<br />
- Alternate Partialpenetrationwelds- W curve NotCovered<br />
o OtherJojnts Oneof 8 classes: 0, C, D, f, ~2,G<br />
& W,dependingon geometry,stress<br />
ReferstoAWS 01.1<br />
directionandmethodof manufacture<br />
and InspectIon<br />
Q OtherParametersAffectingS-N<br />
Curves<br />
- Environment CatholicallyprotectedjointsInSea S-N curves(X’andX) presume<br />
waterequivalentto jointsIn air. effectivecathodicprotection.<br />
Unprotectedjointsin seawater Fatigueprovisionsof AWS D1.1apply<br />
requireS-Ncurveto be reducedby a to membersand jotntsIn atmospheric<br />
factorof 2 on llfe(Ref.A21.2.13a) service.<br />
P<br />
. ,<br />
Figure4-7 Comparisonof Recommendations<br />
U.K.GuidanceNotesandAPIRP 2A<br />
\<br />
f
TOPIC<br />
U.K.DEPARTH~NT(lFENERGY(lJEn)<br />
AMERICANPETROLEUMINSTITUTE<br />
- PlateThickness<br />
Ooesnotrecommendfurtherreduction<br />
of S-ll%rve for freecorrosion(fC)<br />
basedon testdataon bothFC and<br />
cathodicprotection(CP).(Ref.C<br />
5.5)<br />
nodaljoints BasicS-N curvefortB=32mn (T Not covered<br />
curve)<br />
CorrectIonS = s~ (32/t)$<br />
non-nodaljoints BasicS-NcurvefortBf22nsm (B-G ~ot covered<br />
curve)<br />
CorrectionS = S It./t]*<br />
(Ref.FigureA.2!.2.!f3b)<br />
- Weld Improvement 30% in strength(2.2factoron life) Profilingallowstheuse of X-curve<br />
by controlledmachiningorgrlnding ratherthanXi-curve<br />
of weldtoe (Ref.FigureA,21.8)<br />
‘ATIGUEDAMAGECOMPUTATION<br />
Note: Requireda smoothconcave<br />
profileat weld toewlthmln. 0.5~<br />
penetrationIntotheplate.<br />
ieconmnended Method<br />
Cunrnulative damageby ~fner’sRule<br />
(Ref.21.2.14)<br />
Cwnulativedamage byt41ner’srule<br />
wherestressresponses’foreachsea<br />
statearecombinedtntothe long<br />
termstre<strong>ssc</strong>flstrlbution, which<br />
shouldthenbe usedtocalculatethe<br />
cumulativedamageratio. Alter-‘<br />
natively,thedamageratiomaybe<br />
computedfor eachseastateand<br />
combinedto obtainthecumulative<br />
damageratio,(Ref.5.2.4)<br />
Page4 of 5<br />
Figure4-7 Comparisonof Recommendations<br />
---- U.K.GuidanceNotesandAP1 RF 2A<br />
(:;;
TOPIC<br />
U.K.DEPARTMENJOFEIWRGV(oEn)<br />
OTtiERCOMPONENTS<br />
Castor ForgedSteel<br />
Covered(Ref.21.2.15)<br />
flotcovered<br />
OTHERCONSIDERATICINS<br />
Fatiguesertsitlvlty and<br />
of failurestudies.<br />
WI<br />
consequence<br />
RecommendsIdentlflcationof<br />
crltlcaljolntslmembersand<br />
developinga selectlveinspection<br />
programcompatiblewithbothfatigue<br />
sensitivityand failureconsequence<br />
(Ref.21.2.10e)<br />
Beneficialreduction~p toe ~eve~of<br />
tensileresidualstress.”However,<br />
no benefitsassumedon fatfguellfe.<br />
(Ref.21.2.11)<br />
Conslc&eredbeneflc~a]as ~esldua”<br />
stressesinfluehtiecrack<br />
Inltlatlon.However,no benefits<br />
assumedon fatiguellfe.<br />
Treatmentof low stress<br />
cycles<br />
Non-propagatingstressat Y= 107<br />
(Ref.A21.2.13c)<br />
Non-propagatingstressa? 1!= 200 x 106<br />
Treatmentof highstre<strong>ssc</strong>ycles<br />
T curveextrapolatedbackto stress<br />
rangeSB = 292~ (Ref.A! 21.2.13d)<br />
Endurancellmlts=<br />
5.07ksl (35?4Pa forX-curve<br />
3.33kst (23MPa1<br />
forX’-curve<br />
Page5 of 5<br />
Figure4-7 Comparisonof Recommendations<br />
U.K.GuidanceNotesandAP1 RP 2A
(THIS PAGE INTENTIONALLY LEFT BLANK)<br />
-,-.
5.<br />
FATIGUE STRESS MODELS<br />
5.1<br />
REVIEW OF APPLICABLE MODELING STRATEGIES<br />
The structure configuration essentially dictates the modeling<br />
strategies and the analysis methodologies. Various strip methods<br />
are used to determinethe wave loadingson long, slender bodies such<br />
as ships. The strip theory can account for the effect of diffracted<br />
and radiatedwaves. The hydrodynamicloadings on ships, aswell as<br />
semisubmersibles,can beobtainedfromthree-dimensionaldiffraction<br />
analysis.<br />
Discrete systems, such as bottom-supported fixed platforms, are<br />
substantiallydifferent from continuous systems, such as ships and<br />
semisubmersibles, in the characteristicsof the applied loadings,<br />
their response to these loads, and the resulting stress<br />
distribution. Although the components of the strength model are<br />
similar for both systems, the specifics and the related<br />
uncertainties are different. Thus, fatigue stress models for<br />
bottom-supported and floating marine structures are discussed<br />
separately in Sections 5.2 and 5.3, respectively.<br />
5.1.1 Modelinq Strategies<br />
Analytical models are developed to determine excitational loads,<br />
motions/response,anddeformations/stresses. The levelofdesirable<br />
model complexity depends on many variables, including:<br />
●<br />
The desired level of accuracy of results.<br />
●<br />
Theaccuracyofvariables/assumptions input intothe analysis.<br />
●<br />
The effect of modeling complex<br />
the interpretationof results.<br />
ty on modeling errors and on<br />
5-1
●<br />
The effect of modeling complexity on analysis schedule and<br />
cost.<br />
The current state-of-knowledgeprovides us with the tools necessary<br />
to develop and analyze models. The desirable level of modeling<br />
sophistication, different for each structure, is thus determined<br />
based on tradeoffs among some of the variables given above.<br />
The goal of a modeling strategy should always be to achieve<br />
realistically accurate results consistently and without excessive<br />
complexity. The analysis assumptions and the modeling strategy is<br />
very important in minimizing modeling accuracy/error. Most<br />
engineers rely on previouswork and engineeringjudgement to reduce<br />
the level of modeling errors, typically defined as the ratio of<br />
actual-to-predicted results. Such a subjective approach can be<br />
supplemented by statistical methods to define the modeling<br />
uncertainty. The mean value of the modeling error, Xme, is defined<br />
as the “bias.”While the modeling uncertainty is referred to as the<br />
random component of the modeling error. The modeling uncertainty,<br />
given by its coefficient of variation, (C.o.v.)x is meaningful<br />
me<br />
only if sufficient data is available.<br />
5.1.2<br />
Comparison of <strong>Structure</strong>s<br />
A discrete system composed of numerous members and joints (such as<br />
an offshore platform) is modeled as a 3-D space frame. Individual<br />
members of the system are modeled as stick elements, with correct<br />
dimensions (diameter, net length) and hydrodynamic coefficients.<br />
The two basic premises affectingthe accuracy of wave loadings are:<br />
●<br />
The hydrodynamicforces are typically computed based on water<br />
particle kinematics along each member centerline. When the<br />
wave length-to-cylinderdiameter ratio is less than about 10,<br />
the wave force computed based on a stick model centerline is<br />
too conservative.<br />
5-2
●<br />
The water particle kinematicsare assumedto be unaffectedly<br />
the presence of such members. When the cylinders are spaced<br />
so that they are at least 3 or 4 diameters apart, the wave<br />
inertia forces on one cylinder are relatively unaffected by<br />
the presence of the other cylinders as the radiation effects<br />
are small.<br />
Since platform member diameters are typica” ly less than 3 feet<br />
(2.Om) for braces and less than 6 feet (2.0 m) for legs, the two<br />
basic premises are valid. Even if a 10 foot (3.Om) diameter leg is<br />
utilized, for a wave period of 6 seconds the wave length-to-leg<br />
diameter ratio is in excess of 18. Thus, diffraction effects are<br />
smal1.<br />
However, a45 foot (14m) diameter column of a tension leg platform<br />
will have a wave length-to-columndiameter ratio of only about 4 for<br />
a wave period of 6 seconds. The columns are likely to be only3 to<br />
4 diameters apart. The column spacing is even less for a<br />
semisubmersiblehavingthree columns on each pontoon. Thus, the two<br />
premises are not applicable for structures made up of large<br />
members. The water particlekinematicsat member centerlinesareno<br />
longer valid for smallwave periods and the presence of such members<br />
in the proximity of others affects the water particle kinematics.<br />
Although the stickmodel ofa platformcan be modeled from one joint<br />
node to another, the applied loads could be in error by 2% to 3%<br />
because the loadings on member ends within the chord are computed<br />
more than once due to member overlaps, Most software packages<br />
include an option to define the member ends within the chord,<br />
preventing multiple computation of the applied loads, buoyancy and<br />
weight at each joint.<br />
Accurate definitionofa ship’sdeck strength is importantto define<br />
the box-girder-likeresponse of the entire hull. If a strip method<br />
is not used, the plate elements of the model used in a diffraction<br />
analysis (for loads) and the finite element analysis (for stresses)<br />
5-3
shall have sufficiently fine mesh and member properties to ensure<br />
accuracy of the results. On other floating structures,such as the<br />
TLP and a semisubmersible,the diaphragm action of the deck plating<br />
can be represented either by shear plates or by equivalent beams.<br />
5.2<br />
FLOATING MARINE STRUCTURES<br />
Both mobile and stationarymarine structures are discussed in this<br />
section. The overall discussion is applicable to configurations<br />
ranging from ships and barges to semisubmersiblesand tension leg<br />
platforms (TLPs).<br />
The floating marine structure configuration and the mode of<br />
operation (mobile versus stationary) are the primary variables<br />
affecting the development of an appropriate “loads” or<br />
“hydrodynamics”model. The problems encountered and the technique<br />
applied to determine the wave loads are different for ships and<br />
other stationary marine structures for several reasons:<br />
●<br />
While ships are treated as slender bodies, most offshore<br />
structures other than FPSOS, FOSS and drillships can not be<br />
treated as slender bodies.<br />
●<br />
The three-dimensional flow calculation technique can be<br />
appliedto typicalstationarystructuresbut cannot be applied<br />
to ships that have a constant forward speed.<br />
●<br />
Steady-state response of a stationary structure to<br />
excitationalwave loads allowsdeterminationof relativewater<br />
particle velocities and accelerations and assessment of<br />
structurecompliancy (netloading). These excitationalloads<br />
have less influence on ships in-motion (i.e., near-complete<br />
compliancy).<br />
●<br />
Stationaryfloatingmarine structuresaremoored/tethered and<br />
are subjectedto low-frequencydrift forces,which, due to the<br />
5-4
“radiation pressure” of waves, significantly affect the<br />
mooring/tetheringsystem design.<br />
5.2.1 shill<strong>Structure</strong>s<br />
Determination of Loads<br />
Seakeeping and wave loads on ship structuresare determined largely<br />
based on two-dimensional solutions of flow problems for plane<br />
sections. Combinationsof various plane section solutions provide<br />
an approximate loading for the entire hull. Approaches based on<br />
utilizing the plane sections of slender hulls are identified as<br />
“strip methods.” Typically, a strip method utilizes a linear<br />
relationship between wave amplitude and response in a frequencydomain<br />
solution. However,non-linearresponses in a time domain can<br />
be also solved.<br />
A two-dimensional flow problem is often analyzed for a range of<br />
variables. Typically,solutionsareobtainedfor one wave direction<br />
and a number of frequencies. Then other wave directions, defining<br />
an angle of encounter between the wave and the ship, are chosen and<br />
solutionsobtained. The studyresults are interpolated to determine<br />
the ship responseamplitudes. Althougheightwave directionsshould<br />
be considered for stress analysis (head and following seas, beam<br />
seas, bow quartering and stern quartering), several directions can<br />
be disregarded (globaleffectsof port and starboardquarteringseas<br />
are similar) for motions analysis.<br />
Typically, strip methods disregard the longitudinal forces due to<br />
surge motions of the ship. Longitudinal forces are small and the<br />
use of Froude-Krilof forces and hydrostatic head appears to be<br />
satisfactoryto determinethe hull longitudinalstresses. However,<br />
work carried out by Fukusawa et al (Reference 5.1) indicates that<br />
the deck longitudinal stresses of a fully loaded tanker may be<br />
increased appreciablydue to longitudinalwave forces.<br />
5-5
The ship motion and wave action result in truly complicated<br />
interaction of variables affecting the loading on the hull<br />
structure. Theloads dueto incident,diffractedand radiatedwaves<br />
and due to ship forward motions may be approximated for various<br />
sections of the hull by the use of strip theory. Loads due to<br />
diffraction and radiation can be also directly obtained from a<br />
three-dimensional flow solution. Work carried out by Liapis and<br />
Beck (Reference5.2) provides a very good comparison of various 3-D<br />
flow solutions, strip theorysolutionand experimentalresults. The<br />
added mass and damping coefficients plotted against frequency on<br />
Figure 5-1 indqcate that the coefficients obtained by Liapis and<br />
Beck are quite close to those obtained based on both strip theory<br />
and experimental work. Actually, over the range of applicable<br />
frequencies,the three sets of coefficients based on 3-D solutions<br />
show larger scatter.<br />
Considering the difficulties of applying 3-D solutions and the<br />
proven reliability of good strip methods, a strip method is 1ikely<br />
to remain the preferred approach to determine the applied loads in<br />
most ships. <strong>Ship</strong>s with special characteristicCS, including<br />
supertankers, navy vessels, drillships, etc., are the likely<br />
candidates for application of 3-D flow solutions. It should be<br />
emphasized that whichever solution method is chosen, substantially<br />
greater inaccuraciesare introduced into the hull loading due to:<br />
●<br />
Uncertaintieson wave height and period (wave statistics)<br />
●<br />
Uncertaintiesregarding ship routing and the correlationwith<br />
wave environment<br />
●<br />
●<br />
The variable nature of ship cargo and ballasting<br />
Inaccuraciesin hull response to applied loads<br />
The precedingdiscussioncoverswave loadingon ships susceptibleto<br />
cumulative fatigue damage. A linear theory is applicable to<br />
determine the applied loading for fatigue analyses and design. In<br />
an extremelyharshenvironment,wave nonlinearitieshave substantial<br />
5-6
influence on the applied loading. However, a linear theory can<br />
still be used in a harsh environmentto produce approximateloadings<br />
as harsh environment generally contributes very little to the<br />
cumulative fatigue damage.<br />
If an appreciable portion of fatigue damage is due to harsh<br />
environment loading, some of the important variables not accounted<br />
for in linear theory should be evaluated:<br />
●<br />
●<br />
●<br />
●<br />
●<br />
Wave steepness<br />
Wave slamming<br />
Viscous effects<br />
Hydrostatic effects (due to flaring ship sections)<br />
Hydrodynamic effects (due to flaring ship sections)<br />
These primary and other secondary nonlinearity effects on ship<br />
loading can be accounted for by perturbation and simulation<br />
methods. Second-order perturbation methods are relatively simple<br />
and they are used to solve the wave action/ship motion problem in<br />
the frequency domain. A detailed discussion of second order<br />
perturbationmethods is presented in References 5.3 and 5.4.<br />
Another approach to determine the non-linear effects is the<br />
integration over time of the applied forces on the structure. A<br />
detaileddiscussionofsuchsimul ationmethods, includingprinciples<br />
of effective computer simulations, is presentedby Hooft (Reference<br />
5.5).<br />
Motions Model and Anal.vsisTechniques<br />
Since the linear ship motion theory is considered appropriate for<br />
large majority of spectral fatigue analyses, the modeling and<br />
analysis technique is further discussed.<br />
Typically, a standard ship or a tanker has two distinct drafts, one<br />
for laden and another for ballast condition. The pre-analyses<br />
effort usually covers the following:<br />
5-7
●<br />
Preparationof a table of offsets for the vessel,defining the<br />
geometry with stations (20 or more)along the longitudinal<br />
axis and points (15 or more) at each station (i.e. describing<br />
the transverse section).<br />
●<br />
Preparationof weightdistributionto define structure (steel)<br />
and variables (ballast,cargo, fuel, etc.).<br />
●<br />
Utilization of table of offsets and weight distribution to<br />
compute bending moments and shear forces at each station.<br />
The<br />
shear force and bending moment diagrams developed along the<br />
length of the vessel facilitate equilibrium checks.<br />
The vessel motion anlalysis requires definition of vessel<br />
hydrodynamic properties: For a linear strip theory based ship<br />
motion computerprogram,the hydrodynamicpropertiesdefiningvessel<br />
added mass and damping coefficientsmay be input based on available<br />
data onsimilarvessels. Conforrnalmapping approach is also used to<br />
define the added mass and damping coefficients. However, if the<br />
vessel configuration is unique, a 2D or 3D diffraction analysis is<br />
recommended to define the hydrodynamicproperties.<br />
The 1inear strip theory based ship motion program, uti1izing the<br />
hydrodynamiccoefficients,is usedto generateequilibriumsolutions<br />
for vessel motions in six degrees of freedom. Then, the transfer<br />
functionscan be defined for verticaland lateral bending,torsional<br />
moments, vessel accelerations and hydrodynamic pressures at each<br />
station along the vessel longitudinal axis.<br />
Finite Element Stiffness Model<br />
The load transfer function, both in-phase and out-of-phase<br />
components, are used in the stress analyses to obtain corresponding<br />
stress range transfer functions. The computer model and the<br />
structural analysis used is very important to define local stress<br />
ranges. Fatigue is a local phenomena.and it isimportant to define<br />
5-8
their function, selecting appropriate element aspect ratios (less<br />
than 1:2) will contribute both to better accuracy and a better<br />
model.<br />
5.2.2<br />
Stationary Marine <strong>Structure</strong>s<br />
Determination of Loads<br />
Stationarymarine structureshavevariousconfigurationsand exhibit<br />
a wide range of compliancy. A substantial effort is desirable to<br />
minimize the fatigue loadings on stationary structures. For a<br />
moored tanker FPSO the smallestfunctionalsize exhibitinga minimum<br />
silhouette is desirable. For structures composed of columns and<br />
pontoons, the column spacing,column water plane area, displacement<br />
of pontoons affecting overall center of buoyancy and the total<br />
displacement are some of the interactingparameters that affect not<br />
only the magnitude and characterof the applied loading but also the<br />
response of the structureto applied loading (see Reference 5.6 for<br />
structure configurationoptimization).<br />
While the hydrodynamic forces on a slender stationary body can be<br />
determined based on strip method or diffraction theory, a structure<br />
made up of columns and pontoons can be determined either by<br />
Morison’s equation or by diffraction theory. As discussed in<br />
Section 5.1, large diameters disturb the flow, leading to<br />
diffraction which is highly frequency dependent. There are two<br />
benefits of using diffraction theory:<br />
●<br />
Diffraction usually causes a reduction in the wave loads,<br />
●<br />
Viscosity can be ignored and thus, treating the flow as<br />
irrotational,potential flow theory may be used.<br />
The hydrodynamicloads actingon astructure are typically generated<br />
using a combination of three-dimensionaldiffraction theory, i.e.,<br />
a source-sink distributed potential theory (Reference 5.7) and a<br />
conventional Morison’s equation. Although a two-dimensional<br />
5-1o<br />
/ 1A
analysis program can be used, a three-dimensional program<br />
facilitates overall analysis effectiveness.<br />
To analyze, the structure surface is divided into panels, much like<br />
a finite elementmodel and the potentialflow problem is solved over<br />
each panel and yields diffraction and radiation pressures on these<br />
panels. While the diffractionpressuresare transformedintomember<br />
wave loads, the radiation pressures are transformed into added mass<br />
coefficients. Hydrodynamic drag forces on these members and both<br />
the drag and potentialforces on smallermembers (simulatedby stick<br />
elements) are generated using Morison’s equation. Diffraction<br />
effects are strongly dependent on frequency, so a range of<br />
frequenciesmust be addressed.<br />
Mass Model<br />
Typically the deck structural members are modeled by using<br />
equivalent members to represent the deck structure mass and<br />
stiffness. All other members subjectedto hydrodynamicloading are<br />
modeled, with appropriate mass distribution. The accuracy of<br />
structuremass and its distributiondirectly affect the accuracy of<br />
structure motions.<br />
Motions Model and Analysis Techniques<br />
The mass model discussedabove allowsdeterminationofa structure’s<br />
inertialresponseto the appliedexcitationalenvironmentalloadsby<br />
obtaining solutions to the six-degree-of-freedom equilibrium<br />
equations. Considering the rigid-body motions, the dynamic force<br />
equilibrium on a structure can be expressed using the following<br />
system of six simultaneousequations:<br />
[ [W+[Mal1 {x}+[ [CRI+[CV] J {x}+[IfJ {X} = {FO}+{FI}+{FD,} 5-1<br />
This equation differs from that in Section 5.3.3 in that (1) primary<br />
damping is due to wave radiation and viscous effects, (2)<br />
5-11
hydrostatic stiffness is introduced and (3) the make-up of applied<br />
forces differs.<br />
The terms given represent:<br />
[M] = 6x6 structuremass matrix<br />
[Ma] = 6x6 added mass matrix<br />
[CR] = 6x6wave radiation damping matrix<br />
[Cv] = 6x6 linearized viscous damping matrix<br />
[KH] = 6x6 hydrostatic stiffness matrix<br />
(FD) = 6x1 linearizedwave drag force vector<br />
(F,) = 6x1 wave inertia force vector<br />
(F~~) = 6x1 diffracted wave inertia force vector<br />
{x}, {x}, {x} = 6x1 structure displacement, velocity and<br />
acceleration<br />
If a structure such as a TLP is tethered to the seafloor, the<br />
stiffness matrix is modified from:<br />
[&J {x} to [[&-J + [K~l1 {x}<br />
where, the [KT]represents 6x6 tether stiffness matrix.<br />
As discussed in previous sections,the structureconfigurationsand<br />
the motion characteristics (i.e., steady state harmonic motion)<br />
facilitate the 6 x 6 motions equations solution over the frequency<br />
domain.<br />
It is recommended that the significant wave height in the wave<br />
scatter diagram that is likely to contribute most to the fatigue<br />
damage be chosen to linearize the drag forces for all wave<br />
frequencies.<br />
The basic approach discussed here has been utilized frequently in<br />
recent years, and is discussed herein as it was implementedon the<br />
design and analysis of a TLP by $ircar et al (Reference5.8). The
approach, also called “consistent method” differs from the<br />
conventionalanalysesmethod only in the generation of hydrodynamic<br />
loads. The hydrodynamic loads for a conventional analysis are<br />
typicallygenerated based on a method by Hooft (Reference5.5) with<br />
a modified form of Morison’s equation. Although the conventional<br />
method also yields reliable results in most cases, it should be<br />
noted that the hydrodynamic interactionamong component members of<br />
the structure is neglected. Figure5-2 shows that the appliedheave<br />
and pitch loadingsbased on both consistentand conventionalmethods<br />
are very similar for wave periods (4 to 8 sec.) that contribute<br />
largely to fatigue damage. For larger wave periods (9 to 15 sec.)<br />
representing less frequent larger waves, the consistent method<br />
provides more reliable results.<br />
Stiffness Model<br />
Typicallythe hydrodynamicmodel,mass model and stiffnessmodel are<br />
all developed from the same structuralmodel. The stiffness model<br />
incorporates correct member cross-sectional areas and stiffness<br />
properties,joint releases and boundary conditionsto allow correct<br />
distribution of structuralmember loadings and stresses.<br />
The stiffness analysis is performed for each wave period and<br />
direction to obtain in-phase and out-of-phasemember stresses. It<br />
is necessarythat nominal stressescomputed are realistic. Thus, if<br />
stick members are used to represent large members with internal<br />
chords and bulkheads,additionalfinite element study of such areas<br />
may be necessary. By using the loads from stick model analyses as<br />
the applied loads on a detailed finite element model of a joint,<br />
accurate stress distribution can be obtained to define the nominal<br />
stresses in each sub-componentof such complex joints.<br />
5.2.3<br />
Overview and Recommendations<br />
Although allowable stressmethods may be used to size the component<br />
members of marine structures and to develop better details, a<br />
detailed fatigue analysis is recommended for each structure. Each<br />
5-13
structure is unique and an allowable stressmethod based on typical<br />
structures and a typical environmentwill only provide information<br />
on relative susceptibility of various joints/details to fatigue<br />
failure. In addition,newer vesselsare often constructedfrom high<br />
strength steel, allowingthe use of thinner plates. <strong>Ship</strong> structure<br />
scantlingsizes are basedon strengthrequirementsand any reduction<br />
in scantling sizes without due consideration for fatigue phenomena<br />
is likely to make the allowable stress method unconservative.<br />
Therefore, allowable stress methods can be used as a “screening<br />
process” and a detailed fatigue analysis is recommended to ensure<br />
integrity of the design.<br />
<strong>Ship</strong> <strong>Structure</strong>s<br />
The use of a linear ship motion theory is appropr”ate for fatigue<br />
analysis of most vessels. For most vessels structural dynamic<br />
amplifications,wave nonlinearities,and effects such as springing<br />
due to high forward speeds have negligibleeffect on overall fatigue<br />
lives. However, some vesselsoperating in harsh environmentsmaybe<br />
subjected to appreciable fatigue damage due to harsh environment<br />
loading. For such vessels the ability to predict wave<br />
nonlinearities and vessel hogging, sagging and racking effects<br />
accurately may become important. In such instances, a non-linear<br />
ship motion theory may be preferred over a linear ship motion<br />
theory.<br />
Fatigue is a local stress phenomena and it necessitates accurate<br />
definition of stresses for very complex geometries. In addition to<br />
primary hull girder bending in horizontal and vertical axis,<br />
substantial secondary girder bending moments will occur due to<br />
external dynamic loads on vessel bottom and internal inertialloads<br />
due to vessel contents. Thus, a beam theory based nominal stresses<br />
due to primary hul1 bending are inaccurateboth due to complexityof<br />
geometry and the local load effects. A finite element model should<br />
be developed to represent the behavior of the vessel and to<br />
determine the local stress distributions accurately.<br />
5-14<br />
HL
For each load component (in-phase and out-of-phase) at each<br />
frequencyof agiven wave directionthe finite elementmodel is used<br />
to generate local stress distributions. The stress range transfer<br />
functions are then generated for each wave direction. Although<br />
current computers are well suited to compute large problems, the<br />
number of frequenciesnecessary to define the transfer function may<br />
be small. Using the predominantload transfer function as guide a<br />
limited number of frequencies (say4 to 6) maybe adequateto define<br />
the other transfer functions. The use of a stress range<br />
distribution parameter allows carrying out of a fatigue analysis<br />
with relatively few structural analyses cases. The accuracy of<br />
fatigue lives obtained largely depends on the validity of the<br />
Weibull shape factor used.<br />
The shape factors obtained by calibratingthe characteristicstress<br />
range against spectral fatigue approach indicate that the single<br />
most important variable affecting the shape factor is the<br />
environment. While the shape factor may vary from 0.8 to 1.2,<br />
depending on the route characteristicsand on structuregeometry, a<br />
factor of 1.0 may be used when such information is not available.<br />
Stationary Marine <strong>Structure</strong>s<br />
The accuracy of stress transfer function for a joint/detail of a<br />
stationarymarine structuredependson many variables, includingthe<br />
accuracy of applied loads, motion response characteristics and the<br />
stress distribution. Hydrodynamic forces may be determined by<br />
either Morison’s equation or by diffractiontheory. Since the wave<br />
length-to-member sizes are small for most floating (i.e.<br />
semisubmersibles,TLPs) structures, diffraction effects should be<br />
accounted for.<br />
A 2D or 3D diffraction analysis can be used to generate the<br />
hydrodynamic coefficients. Then, utilizing these coefficients,<br />
Morison’s equation can be used to generate theappliedl oads. As an<br />
alternative,diffraction analysis can be used to generate the wave<br />
5-15
loads directly. Since the diffraction effects are strongly<br />
dependent on frequency, a wide range of frequenciesmust be used.<br />
The response of the floating structure to applied wave loadings<br />
depends on its own geometry, stiffness and mass properties. Water<br />
plane area and its distribution (i.e., hydrostatic-stiffness)and<br />
mass properties directly affect the natural periods and the heave,<br />
pitch and roll response amplitudes. For a tethered structure,such<br />
as a TLP, tether stiffnesswill predominate hydrostatic stiffness.<br />
Tether pretensionswill control surge and sway natural periods and<br />
response amplitudes. The primary damping is due to wave radiation<br />
and viscous effects.<br />
It is recommended that the “consistent approach” discussed in<br />
Section 5.2.2 is used to accuratelygenerate hydrodynamicloads. A<br />
finite element model of the structure can be used to obtain the<br />
solution to the motions analysis and determine the stress<br />
distributions. As an alternate,a stickmodel maybe used to obtain<br />
solutions to the equations of motion and to define global<br />
deformations and forces. Then, additionalfinite element models of<br />
various interfaces will be necessary to determine local stress<br />
distributions accurately. The boundary conditions for the finite<br />
element models will be the stick model deformations.<br />
5.3<br />
BOTTOM-SUPPORTEDMARINE STRUCTURES<br />
This section discusses bottom-supportedmarine structures that are<br />
represented by three-dimensional space frames and composed of<br />
cylindrical shells. The dynamics of a large gravity-type bottom<br />
supported structure dynamics are somewhat similar to those of a<br />
fixed platform. However, the characteristicsof excitationalloads<br />
on gravity structureshave more in common with floating structures.<br />
5.3.1 Load or HydrodynamicsModel<br />
A wave force acting on a single stationary element is due to both<br />
the accelerationof water particles (inertialforce) and the kinetic<br />
5-16
energy of the water particle (drag force). These forces are given<br />
by Morison’s equation as:<br />
5-2<br />
where:<br />
F =<br />
hydrodynamicforce vector per unit length acting normal<br />
to the axis of the member<br />
F,&FD=<br />
inertia and drag components of F<br />
P<br />
=<br />
density of water<br />
cm ‘<br />
inertia force coefficient<br />
cd =<br />
drag force coefficient<br />
D=<br />
diameter of a tubular<br />
u“ =<br />
duw<br />
component of the velocity vector of the water normal to<br />
the axis of the member<br />
= component of the accelerationvector of the water<br />
normal to the axis of the member<br />
liw=— dt<br />
II<br />
= denotes absolute value<br />
An appropriateapproachto estimatethe wave forces with reasonable<br />
accuracy is to assess the load model in its entirety, and for its<br />
component elements.<br />
The element diameter should reflect any geometric variations,<br />
including marine growth. The Cd and Cm values applied may range<br />
5-17
typically from 0.6 to 0.8 and 1.5 to 2.0, respectively. Very<br />
comprehensive experimental data obtained from full-scale<br />
measurements of the second Christchurch Bay Tower (References5.9,<br />
5.10 and 5.11) validate the coefficients in use today. As<br />
illustrated on Figure 5-3, the Cd and the Cm values applicable for<br />
most cylindrical members near the water surface (Level 3) are 0.66<br />
and 1.8, respectively. Although these values are applicable for<br />
Keulegan-Carpenter (Ke) number in excess of 30, even when Ke is<br />
reduced to 5, the inertiacoefficient,Cm, value reaches 2.0, while<br />
the drag coefficient, Cd, gradually increases to unity at Ke equal<br />
to 10.<br />
These coefficients also decrease with the distance from the water<br />
surface. However,becausethe uncertaintiesinmarine growth (which<br />
directly affects the surface roughness and therefore the drag<br />
coefficient)and the additionaleffortnecessaryto input,check and<br />
justify different coefficients,it is advisableto use one set ofC~<br />
and Cm values.<br />
The use of conventionalMorison’s equation and the wave kinematics<br />
for regular two-dimensionalwaves has proven to be valid for jacket<br />
structures in moderate water depths. Assessment of measureclwave<br />
forcedata (Reference5.12) for extremewave loading associatedwith<br />
directionally spread seas in a hurricaneenvironment in the Gulfof<br />
Mexico compares quite well with those analytically computed.<br />
Morison’s equation is<br />
cylindrical members by<br />
also valid to compute forces on nonapplying<br />
appropriate Cd and Cm values and<br />
equivalent diameters. Suitable values of Cd and Cm for different<br />
cross-sections may be obtained from a Det norske Veritas (DnV)<br />
document (Reference4.16.)<br />
If the extreme loadings are to be computed, an applicable wave<br />
theory, compatiblewith the wave steepness,water depth, etc., must<br />
be used. The applied total load on a structure composed of many<br />
members is then the cumulative sum of loads computed on each member<br />
5-18
for a pre-definedwave height, wave period and crestline position.<br />
This conventionalregularwave method produces applied hydrodynamic<br />
loads that has been validated by an extensive performancerecord of<br />
structures in shallow-to-moderatewater depths. However, such a<br />
method is not advisable for structures in deeper water and<br />
exhibiting dynamic response. More rigorous approach to represent<br />
the true response characteristicsis necessary (References5.13 and<br />
5.14).<br />
5.3.2 Mass Model<br />
For a bottom-supportedstructurein relativelyshallowwater, amass<br />
model may or may not be necessary. Such a rigid structure has<br />
natural periods that are less than about 3 seconds and exhibits<br />
little dynamic response when subjected to long-period waves<br />
associated with a harsh environment. For such an environment the<br />
static forces obtained due to water particle kinematics can be<br />
increased slightly to account for the dynamic response predicted<br />
(i.e., computation based on estimated natural periods). However,<br />
most of the fatigue damage is likely to occur due to short-period<br />
waves, necessitatingdeterminationof platform dynamic response to<br />
a wide range of wave periods.<br />
Whether platform dynamic response is to be determined or not, the<br />
dynamic amplificationfactors (DAF)used in a deterministicfatigue<br />
analysisrequire an accurateestimateof naturalperiods and the use<br />
of a mass model is recommended to obtain an eigenvalue solution.<br />
For a spectralfatigue analysis,only the use of amass model allows<br />
determinationof platform dynamic response and direct generation of<br />
structure inertia loads that are compatible with the excitation<br />
loads due to waves.<br />
A mass model of a three-dimensionalspace frame should incorporate<br />
all structuralmembers. The mass will be accuratelydefined if the<br />
weight of all structuraland non-structuralmembers,deck equipment,<br />
ballast,hydrodynamicmass, etc., are accountedfor correctly and in<br />
their respective locations. Ideally, all member weights should<br />
5-19
therefore be defined uniformly along the member lengths. However,<br />
considering the cost of modal analysis, most structural member<br />
weights are input as lumped masses at member ends that attach to<br />
applicable joints.<br />
5.3.3 Motions Model and Analyses Techniques<br />
The mass model discussedabove allowsdeterminationofa structure’s<br />
initial responseto applied excitationalenvironmentalloads by the<br />
use of equilibrium equation solutions. The dynamic force<br />
equilibrium on a structure can be expressed using the following<br />
system of six simultaneousequations:<br />
[ [M + [Ma]} {x} + [cl {X} + [K] {x} = {FD}+ {F,} 5-3<br />
where:<br />
[M] =<br />
[Ma] =<br />
[c] =<br />
[K] =<br />
(FD) =<br />
(Fl) =<br />
{x},{x},{x} =<br />
6 x 6 structuremass matrix<br />
6 x 6 added mass matrix<br />
6 x 6 structure damping matrix<br />
6 x 6 structure stiffness matrix<br />
6x1 wave drag force vector<br />
6x1 wave inertia force vector<br />
6x1 structure displacement, velocity and<br />
acceleration<br />
The terms on the right hand side of the dynamic equilibrium<br />
equations represent external forces applied to the structure.<br />
Following solution of the equilibrium equations, the structure<br />
dynamic response can be moved to the right hand side of the equation<br />
to define the resultant loading.<br />
Thus, the net loading using Morison’s equation given in Section<br />
Eqn. 5-2 can be rewritten as:<br />
F net = : PD2[Cm~-<br />
(Cm-l) UJ + ~f)C~Dulul<br />
5-20
where:<br />
u<br />
= defined as the net velocity vector component s<br />
UW-UG<br />
Uw<br />
u=<br />
= the componentof the velocity vector of the water<br />
= the structure velocity<br />
Uw<br />
= the component<br />
water<br />
of the acceleration vector of the<br />
U*<br />
cm<br />
= the structure acceleration<br />
= added mass coefficient is often taken to be a<br />
variable ranging from 1.5 to 2.0. It is<br />
recommended that Cm be taken as 2.0, which is<br />
consistent with the potential flow solution for<br />
added mass.<br />
It is necessary to choose an appropriate method or analysis<br />
technique that is compatiblewith fatigue design parameters such as<br />
the structure configuration and its susceptibility to fatigue and<br />
the environment. If the structure dynamics are negligible, and a<br />
deterministicanalysis,based on the use of wave exceedence curves,<br />
may be appropriate for initial sizing of platform components.<br />
However, for most structures, the dynamic response should be<br />
incorporated into the fatigue analyses as illustrated in the above<br />
given equilibrium equations.<br />
A rigorous analysis using a time integration method to determine<br />
platform global and local dynamic responses at each wave height and<br />
period is time consuming and costly. Therefore, it is desirable to<br />
have an alternative analysis procedure. One such alternative<br />
proposed by Serrahn (Reference5.15) consists of a hybrid time and<br />
frequencydomain analysismethod. The analysisflow charton Figure<br />
5-4 summarizes this analysis methodology.<br />
Global spectral static and dynamic responses (e.g. base shear and<br />
overturningmoment) aredeterminedat selecteddiscretewave heights<br />
and periods. The static response is determined based on an applied<br />
5-21
load analysisof adetailed three-dimensionalmodel of the platform.<br />
An eigenvalue (modal) analysis is also performed on the same model<br />
to determine platformnaturalperiods and mode shapes. The platform<br />
global dynamic responses are determined by separating each applied<br />
static wave loading into its Fourier series components and solving<br />
directly for the dynamic response (This method of solution is<br />
detailed in Appendix E of Ref. 5.15). Spectral analyses for both<br />
the static and dynamic responsesare then performedand the spectral<br />
inertial load calculated. Inertial load sets are then developed<br />
from the modal results of the previous eigenvalue analysis which<br />
produce the calculated global spectral inertial response (This<br />
method of inertial load development is detailed in Appendix F of<br />
Ref. 5.15).<br />
Such analyses can be repeated for various wave spectra, structural<br />
damping, platform peri?d, etc. at a relatively nominal increase in<br />
analysis time and computer cost. Therefore,this method facilitate<br />
parametric studies to assess fatigue sensitivity of the platform.<br />
Of the three spectral analysisoptions availableto define the wave<br />
loading, the frequency-domainsolution, providing member and joint<br />
in-and out-of-phase wave loads is most frequently used due to its<br />
simplicity. For an iterativedesign process, an analysis approach<br />
utilizing random waves or regular waves in time domain is<br />
appropriate but not frequently used due to both time and cost<br />
constraints. Thus, a hybrid time and frequency domain method is<br />
well suited for spectral fatigue analysis of a bottom-supported<br />
structure. Figure 5-5 illustratesthe scatter of fatigue lives as<br />
a function of the analysis method chosen.<br />
Another appropriate procedureto define hydrodynamicsandwave-force<br />
model, proposed by Kint and Morrison (Reference5.16), is based on<br />
a short extract from a random simulation substituted for a design<br />
wave. The proposedprocedureoffers a valid and a relativelysimple<br />
alternative to the conventional regular wave analysis. Inertial<br />
loads due to structure response can be obtained and dynamic<br />
amplification factors (DAFs) determined by performing a number of<br />
5-22
simulations of random waves. The basic DAF approach, allowing<br />
combination of inertial loads compatible with static loads, is<br />
further discussed by Digre et al (Reference 5.17). Typically,<br />
simulation of the response is performed, the ratio of dynamic-tostatic<br />
loads determined (i.e. DAF)<br />
until the DAF stabilizes. Larrabee<br />
and the process is repeated<br />
(Reference5.18) also provides<br />
further discussion on the logic beh” nd DAF approach.<br />
5.3.4 Stiffness Model<br />
The load and the stiffness models are essentially the same.<br />
Typically, a three-dimensionalspace framemodel of the structureis<br />
made up of individual joints and members, each defining the joint<br />
and member incidence, coordinates, hydrodynamic coefficients,<br />
etc. that are necessary for generationof environmentalloads. The<br />
loads model, providedwith member cross-sectionareas and stiffness<br />
properties,joint releases,and boundaryconditions,transformsinto<br />
a stiffness model. The structure mass model incorporates the<br />
correctmember sizes,joint coordinatesand boundaryconditions,and<br />
can be considered a stiffness model. Static stiffness analysis<br />
solution follows standard structural analysis technique. Dynamic<br />
analysis is typically based on a modal (eigenvalue) analysis<br />
solution; two modal analysis solution techniques may be used:<br />
●<br />
The subspace iteration technique is a Ritz-type iteration<br />
model used on a lumped mass system that produces eigenvectors<br />
and eigenvalues for a reduced set of equations. This is the<br />
method of choice for most fixed offshorestructuressinceonly<br />
a relativelysmall number of modes are required to adequately<br />
model the total structure response.<br />
● The Householder tridiagonalization technique first<br />
tridiagonalized the dynamic matrix, then computes all<br />
eigenvectors and eigenvalues by inverse iteration. This<br />
technique is most appropriate for structures with a small<br />
number of degrees of freedom, for structureswhere all modal<br />
5-23
esponsesare required , or where consistentmass modeling has<br />
been used.<br />
Once eigenvectors and eigenvalues have been determined, specific<br />
dynamic analysesunder load (suchas wave loading)maybe performed.<br />
As previously mentioned, rigorous time integration analyses may be<br />
undertaken, evaluating the dynamic response of the platform over<br />
many cycles of wave loading until steady state response is achieved.<br />
However, the previously recommended approach of expressing the<br />
applied loading as a Fourier series and solving and superimposing<br />
the response of each platform mode to each Fourier sinusoidal<br />
component allows direct determinationof platform dynamic response<br />
without time consuming and costly time integrationanalyses.<br />
The global analysis carried out is often intended to analyze the<br />
three-dimensionalspace frame lateral deformations and ensure that<br />
all components of the structure meet fatigue requirements. An<br />
emphasis should also be placed on plan-level components near the<br />
water surfaceand subjectedto vertical (out-of-plane)deformations.<br />
5.3.5 Overview and Recommendations<br />
Small structures in shallow-to-moderate water depths and in<br />
relatively mild environments are typically not analyzed for<br />
fatigue. Often, stress levels are evaluated and API’s simplified<br />
allowable stress method is used to verify the integrity of design.<br />
Other structures are designed for a wide range of pre-service and<br />
in-service design conditions, including fatigue. Since a fatigue<br />
analysis is carried out to ensure that the design has adequate<br />
safety against damage due to fatigue during the planned life of the<br />
structure, it should address the variables affecting fatigue<br />
appropriately. Modeling and analysisvariables (stiffnessand mass<br />
models, loading coefficients, stress RAOS, SCFS, etc.), affecting<br />
the strength model, and the wave climate (scatter diagram,<br />
directionalprobability,wave spectrum),affectingthe time history<br />
model, incorporatesubstantial uncertainties.<br />
5-24
The analysis effort must be kept comparatively flexible and<br />
manageable and the level of effort should be compatiblewith design<br />
objectives and available information.<br />
It is recommended that a simplified allowable stress approach or a<br />
deterministic fatigue analyses be limited to initial sizing of<br />
members, if considered desirable. A thorough spectral fatigue<br />
analysis is recommended to identify fatigue sensitive<br />
components/detailsof a structure and to take corrective measures.<br />
Considering its relative ease of application a spectral frequencydomain<br />
method is well suited for design. A time-domain method is<br />
better suited to determine the response of a bottom-supported<br />
structure. Since it is time consuming and costly to determine<br />
global and local dynamic response of the platform for each wave<br />
height and period, an alternate less time consuming method is<br />
desirable. Several methods (References 5.15 and 5.16) are<br />
appropriate. A hybrid time- and frequency-domain analysis method<br />
(Reference 5.9), also facilitates carrying out of extensive<br />
parametric studies to assess fatigue sensitivity of structure<br />
components for a wide range of variables and is recommended for<br />
fatigue analyses and design.<br />
5.4<br />
DEVELOPMENT OF HOT-SPOT STRESSES<br />
5.4.1<br />
Nominal Stresses and Stress RAOS<br />
The stresses obtained from a stiffness analysis, and the response<br />
amplitude operators (RAOS)generated, represent nominal or average<br />
stresses. Ingeneral,correct inputof member cross-sectionalareas<br />
and sectionpropertiesallowdeterminationof nominal stressesquite<br />
accurately.<br />
More complex joints, incorporating bulkhead and diaphragm sub<br />
assemblies, require careful evaluation to determine the realistic<br />
load paths. To determine the nominal stresses at complex joints,<br />
5-25
either multiple stick elements (for each load path) or a finite<br />
element model should be utilized.<br />
5.4.2 Stress Concentration Factors and Hot-SPot Stresses<br />
Background<br />
The locations at which maximum stresses occur are called hot spots.<br />
Hot spots usually occur at discontinuities such as the stiffener<br />
edge or a cutout. On tubular member intersections, they usual1y<br />
occur on either the weld toe of the incoming tubular (brace) or of<br />
the main tubular (chord), depending on the geometry of the joint.<br />
The stress concentration factor (SCF) is evaluated by taking the<br />
ratio of the hot-spot principal stress to the nominal principal<br />
stress. The hot-spot stress used in fatigue life assessment is<br />
raised to a power of the inverse of the slope of the S-N curve<br />
used. Since the inverse of the slope of S-N curve is usually<br />
between 2.5 and 4.0, the choice of SCF can have approximately a<br />
cubic effect on damage. Thus the SCF value is probably the most<br />
important variable affecting the applicable stress ranges through<br />
the life of a structure and thus the fatigue life of joints.<br />
There are several practical approaches for determining SCF values:<br />
●<br />
Develop an analyticalmodel ofthedetail/joint and carry out<br />
a finite element analysis (FEA).<br />
●<br />
Test a physical model and obtain hot-spot stresses from<br />
measurements.<br />
●<br />
Use empirical formulations.<br />
The use of FEA is the most reliable and reasonably cost-effective<br />
approach for complex joints. When modeled correctly, the SCFS<br />
obtained by FEA are very reliable and depend largely on the mesh<br />
sizes used in the analysis. Whether the physical model used to<br />
determine the hot-spot stresses is an acrylic model or another<br />
5-26
alternative, the accuracy of hot-spot stresses depends largely on<br />
the ability to predict hot-spot stress locations in advance and<br />
obtain measurements in those areas.<br />
Since the use of both FEA and the physical model requires<br />
substantial investmentof time and cost, they can be used only on a<br />
selective basis. Thus, most structure hot-spot stresses must be<br />
defined based on the applicationof empirical formulations.<br />
Joint Geometry<br />
The primary variables affecting the magnitude of stress<br />
concentrationare weld profile and joint geometry. The weld profile<br />
is accounted for in the S-N curve. The joint geometric<br />
characteristics determine the magnitude of stress concentration.<br />
For most simple structuraldetails, typically awide range of plate<br />
and stiffener joints, the nominal stresses can be used directly to<br />
compute fatigue lives as the effect of SCFS are incorporatedin the<br />
S-N curves.<br />
The joint geometriesof tubularmembers are quite complex andthe S-<br />
N curves are used with the hot spot stresses, requiring definition<br />
of SCFS for each joint geometry and loading. The SCFS are<br />
determined for axial load, in-planemoment and out-of-planemoment.<br />
Typically, a peak SCF is determined and conservatively applied to<br />
eight points around the intersection. For the crown and saddle<br />
points shown on Figure 5-7 separate SCFS can be determined. At<br />
other locations, the SCFS are then interpolated between the crown<br />
and saddle positions.<br />
Joint Definition<br />
When tubular members frame into one another, they form a tubular<br />
joint, with the largest diameter or thickest member being the<br />
through member or chord and all other members being braces.<br />
5-27
Braces may have stubs or cones, which are the part of the brace<br />
member welded to the chord. Typically,both the stubs and the cones<br />
are thicker than the brace members.<br />
To facilitatethe developmentand use of empiricalequationsseveral<br />
parametersare used in definingthe characteristicsof a joint. The<br />
chord diameter and thickness are referred to as D and T<br />
respectively. The brace or stub diameterand thicknessare referred<br />
to as d andt. The angle from the chord to the brace is defined as<br />
theta 0. The ratio of the brace diameter to chord diameter is<br />
defined as beta, B. The ratio of chord radius to chord thickness is<br />
defined as gamma, y. The ratio of brace thickness to chord<br />
thickness is defined as tau, ~. The empirical equations used to<br />
determine SCFS utilize the parameters . The (3,~, y, T. The<br />
terminologyused in defining a simple joint is shown in Figure 5-7.<br />
Joint Type and Classification<br />
Joints are classified intotypes based on geometry and loading. The<br />
joint type usually looks like the letter formed from the brace and<br />
chord intersection. Four basic joint types exist in offshore<br />
structures:<br />
1) T or Y joint<br />
2) K joint<br />
3) KT joint<br />
4) X joint<br />
Figure 5-8 shows the four common joint types.<br />
Although the joint type usually looks like the letter formed from<br />
the brace and chord intersection,the joint is actually classified<br />
according to load distribution. If the axial load is transferred<br />
between the brace and chord by shear, then the joint is classified<br />
as a T or Y joint. If the load is transferredbetween the braces at<br />
a joint, without traveling through the joint, then the joint is<br />
5-28
classified as a K joint. If the load is transferred by some<br />
combinationof shear through the joint and brace-to-brace,then the<br />
joint is classified as a KT joint.<br />
If the load is transferredthrough one side of the chord to another,<br />
then the joint is classified as an Xjoint. Figure 5-9 shows joint<br />
classification by load distribution.<br />
5.4.3<br />
Empirical Eauations<br />
Prior to the discussion of empirical equations it is beneficial to<br />
briefly discuss the available data on SCFS. Review of various<br />
publisheddata (References1.8, 5.19, 5.20, 5.21 and 5.22) indicate<br />
that substantial scatter of SCFS is observed. Variations in SCFS<br />
occur in both nominally identical joints and in symmetrical<br />
locations of joints where one would expect little variations in<br />
SCFS. Material and fabrication imperfectionscontribute to the SCF<br />
variations. Lalani et al (Reference 5.23) point out that the<br />
parameterscontributingto these variationscan be grouped intotwo:<br />
●<br />
Experimental error, including modeling, gauge position and<br />
measurements and the loading.<br />
●<br />
Expected variations due to material and fabrication<br />
imperfections,includingvariations in weld profile, size and<br />
imperfections.<br />
The use of empirical formulationshas been extensively accepted for<br />
fatigue analysisof marine structures. A set of empirical formulae<br />
developed by Kuang (Reference 3.2) were derived by evaluating<br />
extensive thin-shellFEA results. The formulae proposed by Smedley<br />
(Reference3.3) and Wordsworth (Reference3.4) of Lloyd’s Registry<br />
were derived from evaluating the results of strain-gauged acrylic<br />
models. Other empirical equations published include those by<br />
Gibstein (References 5.21, 5.24), Efthymiou (5.19) and Wordsworth<br />
(5.25).<br />
5-29
Whatever the basis for an empirical formula, the formula has an<br />
applicable range of parametersand the level of conservatismvaries<br />
not only with the formulation but also within the applicable range<br />
of parameters. The use ofSCFs also requires judgement not only on<br />
the applicabilityof an empirical formula but also on assessment of<br />
implicationsof in-plane and out-of-plane loadings/stresses.<br />
The parametricequationsdevelopedby Kuang,Smedley-Wordsworth,and<br />
$medley consist of different relationships defined by the joint<br />
variables D, T, d, t, L, g, and 9.<br />
Different equations are applicable for d fferent joint types.<br />
Presently, the joint types and the applicable equations most often<br />
used are listed below:<br />
Joint T.YPe<br />
TorY<br />
K<br />
KT<br />
x<br />
Applicable Equations<br />
Kuang, Smedley-Wordsworth,& Efthymiou<br />
Kuang, & Smedley-Wordsworth<br />
Smedley-Wordsworth<br />
Smedley-Wordsworth,& $medley<br />
The empiricalequationsgiven byUEG (Reference1.8) are based on an<br />
extensive database and relate to Woodworth equations. Modification<br />
of Woodworth equations and the extension of the validity ranges<br />
allow the application of UEG equations to joints with extreme<br />
geometries. Comparisonof variousempiricalequationsshow thatUEG<br />
equations yield generally conservative values of SCFS and are<br />
considered to be most reliable. On the otherhand none of the<br />
equations appear to allow accurate determination of K-joint SCFS<br />
subjected to axial loading.<br />
An excellent overview and reliability assessment of SCF empirical<br />
equations are providedby Ma et al (Reference5.20),Tolloczko et al<br />
(Reference 5.22) and Lalani et al (Reference 5.23). Further<br />
discussion on SCFS and the predicted chord SCF for the different<br />
equations for T and K joints are presented in Appendix C.<br />
5-30
Details of Equations<br />
The details of some of the equations are given in Appendix C. The<br />
equations are given in simple terms of joint geometry: D, T, d, t,<br />
L, g, and r. The Kuang brace SCFS have been modified for the<br />
Marshall reduction. The Smedley-Wordsworth chord SCFS have been<br />
modified for the recommendedd/D limitation.<br />
The parametric equations should not be used outside of their<br />
assigned limits without justification. Near the assigned limits,<br />
the SCFS rapid decrease should be noted to determine if the<br />
calculated SCF is unconservative. The Smedley-Wordsworth effects<br />
revised for d/D limitation can dramatically increase SCFS for d/D<br />
ratios near 1.0.<br />
Minimum Stress Concentration Factor<br />
The minimum stress concentration factor for all modes of loading<br />
should be 2.0. This is generally accepted as an industry lower<br />
bound. However, acrylicmodel tests from the Tern project in United<br />
Kingdom showed a SCF of 1.6 could be used as a lower bound.<br />
5.4.4 Illustrationof a T-Joint SCFS<br />
A typical T-joint with an assumed applied axial load is used to<br />
illustrate the application of empirical equations.<br />
The joint shown on Figure 5-10 is classified by load path and the<br />
joint variablesare specifiedin orderto determine an SCF according<br />
to Kuang and Smedley-Wordsworth criteria. The Kuang brace SCF<br />
includes a Marshall reduction factor, Qr. The Smedley-Wordsworth<br />
chord SCF calculation uses the d/D limitation.<br />
5.4.5 Overview and Recommendations<br />
Uncertainties<br />
5-31
The SCF equations currently in use for simple tubular joint design<br />
are based on results of acrylicmodel tests and finite element (FE)<br />
analysis. Lloyd’s Register has recently studied these empirical<br />
questionsand assessedtheirreliabilitywhen comparedagainststeel<br />
specimentest data. Althoughthe empiricalequationsare considered<br />
reasonably reliable, substantial uncertainties exist as the SCF<br />
equations:<br />
●<br />
●<br />
●<br />
Sometimes do not properly account for relative braceloads<br />
Sometimes do not properly represent the stress at the brace<br />
and chord connection of interest<br />
Axial SCF value for crown and saddle is not constant<br />
The FE analysis of SCFS yield substantially different values<br />
depending on both the modeling techniques and the computer program<br />
used. The use of a thin or a thick element, modeling of the weld<br />
and the definition of chord length substantially influence the<br />
computed SCFS.<br />
SCF equations for a T or a Y joint typically contain a term for<br />
chord length. Since the appropriate length for a chord is not<br />
defined, most designers use the chord can length. While this is<br />
conservative,the use of the half of the bay length to representthe<br />
chord could be very unconservative.<br />
Substantial work carried out in Europe need further assessment and<br />
analyses. An API Task Group will be formed in 1991 to review the<br />
SCF equations in detail, to identifytheir validity and limitations<br />
and to recommend preferred SCF equations for specific joint types<br />
and load components.<br />
An API initiated joint industry project (JIP) is proposed to<br />
summarize the computer programs used and modeling strategies<br />
implemented to investigate variables affecting the SCF (including<br />
chord length) and to developguidelineson obtainingSCFS by the use<br />
of FE analysis.<br />
5-32
Screeninq Process<br />
For a preliminarydesign ofa structure it is common practice to use<br />
a blanket SCF of 5.0 or 6.0 for all joints, depending upon dynamic<br />
effects. If the structure is susceptible to dynamic amplification<br />
the higher blanket SCF should be used. Once the fatigue sensitive<br />
joints are identified during this screening process, the SCFS for<br />
these joints should be determined.<br />
In the determinationofSCFs a parametricstudyofvariablesd/Dand<br />
t/T should be considered. The joint fatigue life is a function of<br />
nominal brace stress and SCF. To increase joint fatigue life, the<br />
nominal brace stress or the SCF should be reduced. An increase in<br />
brace diameter can dramaticallyreduce nominal brace stress without<br />
a significant increase in SCF. This is particularlytrue for brace<br />
members intersecting large diameter legs. However, where members<br />
are more similar in size, an increase in brace diameter also<br />
requires an increase in chord diameter.<br />
By increasing the brace diameter rather than increasing the brace<br />
thickness, a more effective section can be used and prohibitively<br />
low diameter to thickness ratios can be avoided. Increasing the<br />
brace diametermay be the easiestwayto increasejoint fatiguelife<br />
during preliminary design. The chord diameter may also have to be<br />
increased to offset the SCF increasesif the brace area and section<br />
modulus are increased.<br />
Comprehensive Desiqn<br />
Once the member diameters are finalized a comprehensive fatigue<br />
analysis and design may be carried out. The parameter most easily<br />
modified during this stage is the member thickness. An increase in<br />
brace thickness increases brace axial and bending section<br />
properties, which will reduce brace nominal stress. However, as<br />
stated above, the chord thickness should be increased accordingly.<br />
Otherwise the brace nominal stress reduction will be offset by the<br />
joint SCF increase, resulting in marginal difference in fatigue<br />
5-33
life. During the comprehensive design the best parameter to<br />
increase is brace thickness while keeping t/T constant.<br />
Further improvementsin fatiguelivesmaybe obtained by determining<br />
the SCF through the use of finite elements analysis or models<br />
tests. Another alternative to lower the SCF is to stiffen the<br />
joints with rings and thus reduce the SCFS to the lower bound<br />
values. However, considering the increased fabrication costs of<br />
stiffened joints, the use of rings should be considered the least<br />
desirable option to lower the SCFS and improve the fatigue lives.<br />
The validity of SCF equations and their sensitivities to various<br />
geometric parameters are illustrated in Appendix C. It is<br />
recommended that the tables and figures provided are studied to<br />
determine an acceptable approach compatible with the specific<br />
problem on hand. A finite element study results are also included<br />
in Appendix C to illustratethe range of SCFS for a typical complex<br />
joint. Since empirical equations are applicable for only simple<br />
joints, a FEA is recommended for determination of complex joint<br />
SCFS.<br />
5-34
-“<br />
30 Calculation l.iaoas 6 Beck (1905)<br />
Exserlment, Gerritsma & Beukelman 11964)<br />
\ $trln Theory<br />
1 30 Calculation Inglis and Price (79B21<br />
I<br />
-, ,<br />
-, * JO calculation Chang [lg77]<br />
=zl=~<br />
\<br />
\<br />
“$ ‘\<br />
---- -. - . . - . .-<br />
‘-”-”””<br />
Figure 5-1 Comparison of Heave Added Mass and Oamping<br />
coefficients Based on Different Methods<br />
(From Reference 5.2)<br />
Im<br />
lm-<br />
140-<br />
120-<br />
lM -<br />
w-<br />
m-<br />
40-<br />
20-<br />
1<br />
0<br />
2<br />
18 a2 2s<br />
Figure 5-2 Comparison of Wave Loading Based Conventional<br />
and Consistent Methods<br />
(From Reference 5.8)
?.s<br />
I<br />
z -<br />
]LEvE~ ~<br />
1,5<br />
\<br />
i i+<br />
11 Is<br />
1CM<br />
I<br />
0.5<br />
5<br />
Icd<br />
0 L<br />
I 23 57\o 20 3Q 50 70<br />
Keuleqon - Carpenter- No.<br />
Figure 5-3 Comparison of Mean Cd and Cm Values for Christchurch Bay Tower<br />
(From Reference 5.9)<br />
/“7; “< , ._.
PlatlOrmDetailed<br />
3-D ModW<br />
1<br />
I Generate -C 1<br />
I WsMebads I<br />
Sekted H@%ioci I<br />
I<br />
s- AmJyses<br />
and DAFs<br />
I<br />
StaficForms I I Ineftial Forcss [<br />
Msmber Erid<br />
Total I=Orms<br />
Figure 5-4 Dynamic Wave mad Analysis Methodology
-.<br />
Figure 5-5 Comparison of Detailed Fatigue Analyses Techniques
I ‘-l<br />
!1r.“N%%’,<br />
QN<br />
SECTION<br />
A-A<br />
I ‘STUDOF HEA~ WALL OR<br />
SPECIAL STEEL IN BRACg<br />
[OPTIONAL)<br />
\<br />
THROUGH J ,\<br />
i<br />
BRACE<br />
‘A L‘\<br />
p<br />
DETAIL OF SIMPLE JOINT<br />
DETAIL OF OVERLAPPING<br />
JOINT<br />
Figure 5-6 Jo~t GeOIW~ (From -f ereme 1-5)<br />
Crown point “<br />
Figure 5-7 Stiple Joint Terminolo~ (From Reference 1.6)
( -t )<br />
(<br />
T–JOINT<br />
Y–JOINT<br />
( \ I /<br />
K–JOINT K–T JOINT X–JOINT<br />
Figure 5-8 Common Joint Types
JOINT IN=PLANE OUT–OF–PLANE<br />
CLASSIFICATION AXIAL BENDING BENDING<br />
T<br />
I<br />
+, (Q) ~+=<br />
P/2 P/2 M/2 M/2<br />
\<br />
/’<br />
P<br />
M<br />
M<br />
L<br />
\<br />
P<br />
x.P<br />
\<br />
x J<br />
“’ M<br />
M<br />
(“<br />
M<br />
/“<br />
x“ /’ M<br />
K<br />
‘c<br />
P sin Q<br />
-%<br />
‘o <<br />
P sin 9<br />
\<br />
P<br />
T<br />
Figure 5-9 Joint Classification by Mading
SMEDLEY-WORDSWORTH<br />
0<br />
d =— =<br />
D<br />
600<br />
m<br />
VALIDITY CHECK<br />
= 0.500 0.13 s a =0.500 s 10 /<br />
Y“<br />
D<br />
T=<br />
1200<br />
2(40)<br />
= 15.0 lzsy= 15.0 s 32 /<br />
T = +=<br />
20<br />
-m-<br />
= 0.500 0.25 s T = 0.500 s 1.0 /<br />
e = 90° 30° s e =90° s 90° 4<br />
AXIAL SCF CHORD<br />
SADDLE SCF = Y T 6 (6.78 - 6.42 0°”5) Sin(1”7~.7B3)e = 8.40<br />
CROWN SCF = [0.7i-l.37Y005~(l-d)][2Sin0*5@ - Sin30]<br />
.—<br />
KUANG<br />
a<br />
T(2Y6-T)(3-6 Sino)$ino<br />
+[<br />
l=S(l.2-B)(COS40+ 1.5)1 = z 16<br />
][1.05+ 309<br />
2Y-3<br />
.<br />
=— d = 600<br />
= 0.500 0.30 s B<br />
D<br />
= 0.500 s 80 /<br />
Tm<br />
Y<br />
Y-<br />
T<br />
..<br />
--D- ‘+%<br />
= 0.0333 0.015 s y = 0.0333 s 0.060 J<br />
t 20<br />
7 =_ T =<br />
40<br />
a =— =<br />
II 1200<br />
L m<br />
= 0.500 0.20 s T = 0.500 s 0.80 /<br />
= 0.0571 0.05 S a = 0.0571 s 0.30 /<br />
Q = 90° 30° so = 90° s 90° /<br />
AXIAL SCF CHORD<br />
SCF = 1.177y-0”808e-i ”283~l”333d-0.057sin1.6g40 = 7.40<br />
Figure 5-10 Sample Evaluationof a T-Joint
6. FATIGUE STRESS HISTORY MODELS<br />
Creation of the fatigue stress historymodel requires determination<br />
of the fatigueenvironmentand applicationof the environmentto the<br />
structure to produce stresses. The environment can be applied to<br />
the structure by either a spectral analysis or by a time-domain<br />
analysis. The spectral analysis derives the stress range and an<br />
average N number of cycles from the statistical properties of the<br />
stress response spectrum. A true time-domain analysis sorts the<br />
stress ranges and accumulatesthe stress range counts as the stress<br />
time history is being generated. For practical reasons a hybrid<br />
time-domain method is often used to generate stress history.<br />
6.1<br />
DETERMINATIONOF FATIGUE ENVIRONMENTS<br />
To evaluate the fatigue life of a fixed structure or a floating<br />
vessel a representativefatigue environmentmust be modeled. For a<br />
fixed structurethe fatigueenvironmentwill be the typicalwave and<br />
wind conditions for the surrounding area. For a ship the fatigue<br />
environment will be the typical environmental conditions along<br />
various routes.<br />
6.1.1 Data Sources<br />
The types of environmentaldata range from actual wave and/or wind<br />
records to recreated (hindcast)data. The wave and wind recordsmay<br />
be raw recordings (not generally available) or condensed summary<br />
reports produced by government agencies or environmental<br />
consultants. Hindcastdata are generatedby various computermodels<br />
using environmental information available for the area or nearby<br />
areas.<br />
6-1
Wave Records<br />
Older wave and wind informationhas come from voluntaryobservations<br />
by ship personneland frommeasurementsby weather ships and coastal<br />
weather stations. The most likely source of current wave records<br />
are from government agencies such as the National Oceanographic and<br />
Atmospheric Administration (NOAA),obtained through various means,<br />
including weather platforms and weather buoys. Newer techniques<br />
using measurements from satellites provide more comprehensivewave<br />
records. Hoffman and Walden (Reference6.1) discuss environmental<br />
wave data gathering in detail.<br />
While majority of the published wave data is from the North<br />
Atlantic, much of the data applicableto the Pacificwere published<br />
in Japanese and Chinese. Several recent publications (References<br />
6.2, 6.3 and 6.4) in English provide additionaldescription of wave<br />
environment in Asia - Pacific.<br />
The older wave and wind data has the advantage that it covers many<br />
years (decades), but the disadvantages are that the wave heights<br />
were visually estimated, the wave periods were crudely timed, and<br />
the wind measurements were likely biased by the vessel speed.<br />
Various data analysts have devised formulas to correct the<br />
“observed” data. For example, Hogben and Lumb (Reference 6.5)<br />
developed the equations to correlate the significant wave height<br />
(Hs) and the mean zero uncrossing period (Tz) with the observed<br />
data:<br />
Hs = ( 1.23 + 0.44*HOWS)<br />
(meters)<br />
Tz = ( 4.7 +0.32*Tows )<br />
(seconds)<br />
HOws is the wave height and Tows is the period reported by observers<br />
on weather ships.<br />
6-2
Actual recordedwave elevationdata is the most accurate information<br />
available. However, wave records are only available for a few<br />
locations, and typically the time spans of available recorded wave<br />
data are less than 10 years. Even recorded data may not be<br />
complete. The most serious fault in recorded data is that<br />
measurement techniques cannot detect the higher frequency waves.<br />
Wave rider buoys measure wave slope and wave heights are derived<br />
from the slope records. The resolution of these slope measurements<br />
are limited by the dimensions and motion properties of the buoy.<br />
The recorded data does not readily allow detection of the very long<br />
period waves and subsequent data analyses “filter” out the long<br />
period information. Filteringis used to separate “sea” and “swell”<br />
wave spectra. The sea/swell filtering technique is often a simple<br />
truncation of the measured spectrum above and below a selected<br />
frequency. Thus, the higher frequency “sea” part of the spectrum<br />
loses its longer period wave information.<br />
Wind Data<br />
The sources of wind data are the same as for wave data. Older data<br />
tends to be voluntary observations from ships and newer data comes<br />
from measurements on platforms or from weather buoys. Satellites<br />
may provide informationon high altitudewinds by tracking clouds or<br />
from lower level winds by tracking weather balloons.<br />
The older observations are logged anemometer readings and are<br />
typicallyonly the mean wind speed. The height abovewater at which<br />
the wind speed was measured may be unknown. Various analysts have<br />
devised methods to correlate observed wind data to actual measured<br />
data.<br />
Existingoil platforms allowedgathering of extensivewind records,<br />
including gust readings which can be analyzed to derive wind<br />
spectrum information. The presence of the platform has some effect<br />
upon the measured wind velocity, and the location of the anemometer<br />
is very important to the accuracy of the measurements.<br />
6-3
In many cases wind informationmay be available from transmitting<br />
ships or nearby coastal weather stations for areas where wave data<br />
is either skimpy or questionable. For these cases various equations<br />
have been developed to estimate or verify the wave information.<br />
Example equations to relate wind speed to wave height can be as<br />
simple as the “25% Rule”,<br />
H~ = 0.25 * U<br />
where Hs is the significant wave height in feet and U is the<br />
observed wind speed in feet/see. More involved equations include<br />
the wind “fetch” and the wind duration. The wind fetch is the<br />
distance over water that the wind acts. Appendix B presents the<br />
equations developed by Bretschneider to calculate wave height and<br />
period based on wind speed, duration and fetch.<br />
Hindcast Data<br />
Elaborate computer models have been developed to “hindcast” or<br />
recreate weather (wind and wave) records. The hindcast models may<br />
be for a region (such as the North Sea), or the models may be<br />
oceanic or even global. One important consideration in the<br />
developmentof hindcastmodels is the sensitivityof these modelsto<br />
interaction of various parameters. Using available wind and wave<br />
data to correlate the hindcast results can improve the accuracy of<br />
hindcast models.<br />
The hindcast models derive wind information from pressure and<br />
temperatureinformation. Pressuremeasurementsare fairlyaccurate,<br />
and the techniques of combining the pressure readings from many<br />
measurementstationsto produceisobarplots allowsdeterminationof<br />
the pressures over a large region without making measurements at<br />
each grid point. The temperatures measured at coastal weather<br />
stations surrounding the area of interest along with whatever<br />
temperature measurements available from the area can be used to<br />
identify temperature gradients, fronts, etc.<br />
6-4
Wave informationis calculatedfrom wind, accountingfor direction,<br />
duration and fetch. By integrating the weather conditions over<br />
small time steps, a wind and wave history can be built. The<br />
resulting records can be analyzed in a manner similar to that used<br />
with actual wind and wave records to produce wave scatter diagrams<br />
and wave exceedence curves.<br />
6.1.2 Wave and Wind SDectra<br />
Wave and wind spectra define the energy that is being applied to a<br />
structure or vessel. There are many wave spectra formulations and<br />
some of these are discussed in Appendix A. The most general and<br />
therefore most useful wave spectrum formulation is the General<br />
JONSWAP. The General JONSWAP spectra include the Bretschneider<br />
spectra which in turn include the Pierson-Moskowitz spectra.<br />
Reference 6.6 presents a summary of the various wind spectra. The<br />
spectrum recommended in Reference 6,6 is defined as follows:<br />
JONSWAP Wave SDectrum<br />
The JONSWAP (JointNorth Sea Wave Project)spectrumwas derived from<br />
wave measurements in the southern North Sea and is based on older<br />
spectraformulations,Pierson-Moskowitz/Bretschneider/ISSCModified<br />
P-M. The Mean JONSWAP spectrum has fixed parameters and represents<br />
the waves measured during the project. The General JONSWAP<br />
parameters can be varied so that the spectrum can represent either<br />
fully developed seas or developing seas.<br />
The formula for the JONSWAP spectrum is as follows:<br />
s(f) = a (g2/f5)EXP[-1.25(f/fm)-4]qa<br />
where<br />
a = EXP[-.5 (f-fm)2/(sfm)2]<br />
The Mean JONSWAP is defined with the following parameters.<br />
6-5<br />
l-u /J
q = 3.3<br />
s = 0.07, for f< fm<br />
s = 0.09, for f > fm<br />
The Bretschneider spectrum is a subset of the General JONSWAP;<br />
setting the gamma parameter to 1.0 converts the JONSWAP spectrum<br />
into the Bretschneideror ISSCModified P-M spectrum. Also setting<br />
the alpha parameterto 0.0081 convertsthe JONSWAP spectrum intothe<br />
Pierson-Moskowitzspectrum.<br />
As a guideline, the JONSWAP spectrum with gamma = 2 would be an<br />
applicablespectrum for confined regional areas. The Bretschneider<br />
spectrum (JONSWAPwith gamma = 1) would be applicablefor open ocean<br />
(Pacific or Atlantic) areas.<br />
Ochi-Shin Wind Spectra<br />
Ochi and Shin reviewed six wind spectra formulations currently in<br />
use and have created an average wind spectrum to represent the<br />
variation (gusts) of the wind about the mean value. The wind<br />
spectrum represents the average of measured spectra and was<br />
deliberately devised to accurately represent the low frequency<br />
portion of the wind spectrum. The equation has three forms<br />
depending upon the frequency range.<br />
s(f*) =<br />
{<br />
583 f*<br />
420 f*0”70/(l+f*O-3’) ”-5<br />
838 f*/(l+f*0.3s)ll.s<br />
with<br />
f* = f z/uz,<br />
where<br />
f = frequency in Hz,<br />
z = height above sea level in meters, and<br />
Uz= mean wind speed at height z in meters/see.<br />
6“6
6.1.3 Scatter Diaqram<br />
Wave scatter diagrams show the occurrences of combinations of<br />
significantwave heightand averagezero-uncrossingperiod overmany<br />
years.<br />
$iqnificant Heiqht vs Zero-crossinq Period<br />
Irregular waves do not have any consistent pattern of height or<br />
period, but exhibitcompleterandomness. Irregularwave heightsand<br />
periods are usually defined by the statistical properties of the<br />
wave record or by the properties of the energy spectrum which<br />
represents the random sea. The significantwave height is taken to<br />
be four times the standard deviation of the recorded water surface<br />
elevations,or if the sea is representedby a half-amplitudeenergy<br />
spectrum, the significantwave height is four times the square-root<br />
of the area under the spectrum. The average zero-uncrossingperiod<br />
is the average of the time intervals between negative to positive<br />
sign changes in the recorded water surface elevations, or is the<br />
square-root of the area under the spectrum divided by the squareroot<br />
of second moment of the spectrum (frequency in Hz).<br />
The wave height and period distributionovertime can be obtained by<br />
actual wave measurements. The heights and periods of all waves in<br />
a given direction are observed for short periods of time at regular<br />
intervals. A short time intervalof severalhours maybe considered<br />
constant. For this sea state, defined as “stationary”, the mean<br />
zero- uncrossing period, Tz, and the significant wave height, Hs,<br />
are calculated. The Hs and Tz pairs are ordered, and their<br />
probabilitiesof occurrencewritten in a matrix form, called a wave<br />
scatter diagram. A typical wave scatter diagram, presenting<br />
statistical data on the occurrence of significant wave height and<br />
zero-uncrossingperiod for one direction is shown on Figure 6-1 and<br />
further discussed in Appendix B.<br />
6-7
Seasonal Variation<br />
The annual wave scatter diagram is often separated out into monthly<br />
or seasonal (spring, summer, fall and winter) scatter diagrams.<br />
Because a fatigue environment covers many years, the seasonal or<br />
monthly scatter diagram cell values may be added to produce the<br />
annual diagram.<br />
Directional Variation<br />
Sometimes the wave scatter diagram is separated out by direction.<br />
This may be important for fixed structures, because waves from one<br />
directionmay cause a different stress distributionthan waves from<br />
another direction.<br />
Sea and Swell<br />
Sometimes the wave scatter diagrams are separated into “sea” and<br />
“swell”. The sea scatterdiagram shows the significantwave heights<br />
and zero-uncrossingperiodsdefiningsea spectra. The swell scatter<br />
diagram usually shows the heightsand periodsof long period regular<br />
waves. This separated informationcan be helpful in analyzing the<br />
structure, because the swell may be present a large percentage of<br />
the time, and the swell is likely to be from a different direction<br />
than the higher frequency waves producing unique stress<br />
distributions.<br />
6.1.4 Directionality and Spreadinq<br />
The directions that have been referred to up to now have been the<br />
“central”directionof the sea. Irregularwaves are often idealized<br />
as two-dimensionalwith wave crests parallel in the third dimension<br />
and all waves moving forward. Such an irregularsea is called long<br />
crested. In reality, storms occur over a finite area and the wave<br />
heightsdiminishdue to lateral spreading. If such waves meet other<br />
waves from different directions, a more typical “confused” sea is<br />
observed. A confused sea is referredto as a short-crestedsea. The<br />
6-8
waves in a short crested sea approach from a range of directions<br />
centered about the central direction.<br />
Directionality<br />
For a fixed structure the direction of the sea will affect the<br />
stressdistributionwithin the structure. Most fatigueanalysesare<br />
performed for four or eight wave directions. When directional wave<br />
scatter diagrams are available the sea direction can be matched to<br />
the analysisdirection, and the fatiguedamage accumulated. If the<br />
data availabledo not includewave directionality,directionscan be<br />
estimated on the basis of wind roses or hindcasting.<br />
Spreading<br />
In order to model a short crested sea a “spreadingfunction” is used<br />
to distribute the wave energy about the central direction. In<br />
typical analyses the short crested sea is represented by a set of<br />
long crested spectra coming from directions spread over -90 deg to<br />
+90 deg from the central direction<br />
to the specified short-crestedsea<br />
and having a total energy equal<br />
spectrum.<br />
The directional spreadingfunction as defined by Kinra and Marshall<br />
(Reference6.7) is often used in the following form.<br />
D (8) = Cn COSn (1?)<br />
where n is a positive integer and is measured from the central<br />
direction. The coefficient C. should satisfy the following:<br />
A typical n value for wind-driven seas would be 2, while an<br />
appropriatevalue for a limited fetch (restrictedspreading)may be<br />
6-9
4. ~arpkaya<br />
spreading.<br />
(Reference 6.8) provides further<br />
discussion on<br />
A significant effect of short crested seas is that they can cause<br />
response in a direction orthogonalto the central direction, i.e. a<br />
ship may develop considerableroll motion even though the vessel is<br />
headed into the waves.<br />
In the design and analysis of typical offshore platforms (i.e.,<br />
conventional structures in shallow or moderate waterdepths)<br />
spreading is generally neglected. However, for both typical and<br />
nonconventional structures such as the tripod or an extended base<br />
platform (see Figure 6-2) spreadingmay be significant. A platform<br />
with very differentresponsecharacteristicsin two orthogonalaxes,<br />
such as the extended-base platform, may be susceptible to larger<br />
dynamic response in one axis. Even a typical platform, with a<br />
natural period coinciding with the wave force cancellation<br />
frequency, will be subjected to higher wave loading at the<br />
cancellation frequency and neglecting of spreading may not be<br />
conservative.<br />
6.2<br />
STRESS SPECTRUM<br />
A stress spectrum is the stress energy distribution resulting from<br />
loading the structurewith a particular sea spectrum.<br />
6.2.1 Stress RAOs<br />
In order to derive the stress statisticsa stress response spectrum<br />
is developed. The stress response spectrum is the product of the<br />
wave spectrum ordinates times the stress response amplitude<br />
operators squared. The stress response amplitude operators (RAOS)<br />
are the stresses representing a “unit amplitude” regular wave,<br />
obtained by normalizing the input wave heights.<br />
The stress responsesto a set of regularwaves coveringthe complete<br />
frequency (or period) range and the complete direction range are<br />
6-10<br />
[,>—~q
evaluated as explained in Section 5. For a vessel global effects<br />
of port and starboard quartering seas are identical, allowing<br />
reduction of applied loading cases. Similarly, for a platform with<br />
two planes of symmetry several of the eight loading cases (45<br />
deg. intervals)may be combined.<br />
6.2.2<br />
Res~onse Analysis<br />
The response analysis squares the stress RAOS; multiplies them by<br />
the spectrum ordinate; multiplies that product by the spreading<br />
function;andsums/integratesoverdirections the resultsto produce<br />
the stress spectrum.<br />
The stress range spectra is integrated to allow determination of<br />
various statistical parameters, including the zero-uncrossing<br />
frequency, the mean squared value, etc., from which the short-term<br />
probabilitystatistics preconstructed. The ’’Rayleigh’’ distribution<br />
can be used to idealize the stress range associated with a<br />
particular cell (Hs and T) in the scatter diagram. Then, the<br />
fatigue damage associated with each block can be computed, the<br />
cumulative damage thus incorporating the weighting effect of the<br />
joint probability of wave scatter diagram. Since the damage for<br />
each cell is computed numerically, this approach is generally<br />
defined as the “short-termnumerical method.”<br />
The typical loading response exhibits smaller stress cycles<br />
interspersed among larger stress cycles, making it difficult to<br />
identify the number of cycles contributing to fatigue damage.<br />
Rainflow counting is the name of a large class of stress cycle<br />
counting methods often applied to upgrade the short-term<br />
statistics. The rainflow parameter, introduced by Wirsching<br />
(Reference 6.9) is frequently used in upgrading stress spectra<br />
statistics.<br />
The stress range associated with a particular block of the wave<br />
scatter diagram is random in nature and governed by a probability<br />
density function. Such a density function, covering the fatigue<br />
6-11
~. \“<br />
life of a structure,cannot be defined bya closed-formmathematical<br />
function. Most often a numerical long-termdensity function of the<br />
stress range is used to determine the fatigue damage and the method<br />
is identifiedas the “long-termnumerical method”. If the long-term<br />
stress range density function is idealized, an approximate density<br />
function can be used. “Weibull” distribution is one commonly<br />
accepted shape parameter used to describe the long-term stress<br />
density function. The fatigue damage computed is closed<br />
form. Incorporating the Weibull shape parameter is generally<br />
referred to as the “Long-TermClosed-FormMethod”.<br />
6.2.3 Uncertainties and Gaps in Stress SDectrum Develo~ment<br />
There are several important variables contribut<br />
uncertainties in the development of the spectrum.<br />
ng to the<br />
Analysis assumptions substantially influence the calculated<br />
results. The most important of these is the selection of scatter<br />
diagram blocks. While atypical scatterdiagram has40t060 blocks<br />
(each representingthe joint probabilityof Hs and T), these blocks<br />
are often arbitrarily grouped into 10 to 15 super blocks to<br />
facilitate analyses. In addition to the uncertainties introduced<br />
dueto lumpingof these blocks,validityof Rayleighdistributionis<br />
also jeopardizeddue to limitednumber of blocksdefining the entire<br />
environment.<br />
Other analyses uncertainties result from the use or omission of<br />
various parameters (rainflow counting, Weibull d stribution) and<br />
their validity for the problem at hand.<br />
Work carried out by various investigatorshave he” ped enhance the<br />
reliability of spectral fatigue analysis. Chen and Maurakis<br />
(Reference6.10) offer a close form spectralfatigue analysesmethod<br />
that eliminates some of the uncertainties due to analyses<br />
assumptions and computational procedures. The computer program<br />
developed, incorporating the self-contained algorithm, appears to<br />
minimize the uncertainties due to analytical assumptions (i.e.,<br />
6-12
judgement errors) and facilities carrying out of a cost-effective<br />
spectral fatigue analysis.<br />
Some studies show that full-scale service stress data match the<br />
predicted design stresses reasonablywell. However, it should also<br />
be noted that full-scale service stress data may substantially<br />
differ from those predicted during design. This may be especially<br />
true for ships and both the short-term and the long-term service<br />
stress data require a careful scrutiny. Evaluation of full-scale<br />
service stress data on three different ship types (a high-speed<br />
containership, a bulk carrier and a VLCC) by Dalzell et al<br />
(Reference6.11) shows that short-termwave-induced bending moment<br />
do not reasonably fit the Rayleigh distribution. The combined<br />
dynamic stress distributionsfor two of the three ship types did not<br />
fit the Rayleigh or the exponential distributions. Dalzell et al<br />
recommend that additionalresponse calculationsare carried out for<br />
different ship types utilizing Rayleigh and broad-band<br />
distributions. Comparison of response calculations with<br />
experimentaland/orfull-scaleresultsshould indicatethe magnitude<br />
of error and advisabilityof corrective measures.<br />
6.2.4 Decompose into Stress Record<br />
To obtain a stress histogram from the response statistics, the<br />
stress response spectrum for each wave spectrum in the scatter<br />
diagram can be decomposed into a finite Fourierseries. In orderto<br />
produce a realistic stress record, the number of frequencies<br />
required will be on the order of 100. Each component will have an<br />
amplitude defined by the differential stress energy in the<br />
neighborhood of the frequency. Each component will be given a<br />
random phase. By summingthe componentsateach time step, a stress<br />
value is obtained. The stress value is then accumulated into the<br />
stress histogram, accordingto the probabilityof occurrenceof the<br />
particularwave spectrum. The stress histogram can then be used to<br />
evaluate the fatigue life at the hot spot.<br />
6-13
6.3 TIME-DOMAIN ANALYSES<br />
Nonlinear effects, such as submersion/ immersion, velocity squared<br />
drag, mean drift offset, etc., may have a noticeable influence upon<br />
the stresses of a structure. When the nonlinear effects are<br />
substantial, the stresses may be directly calculated from a timedomain<br />
analysis. For a time-domain analysis a discrete set of<br />
regular waves are selected to represent the typical sea spectrum.<br />
The structure response and the stress responses are evaluated by<br />
stepping the waves past the structure in small time increments. At<br />
each time step the Newtonian laws are satisfied.<br />
The regular waves may be selected at equal frequency increments.<br />
Each wave will be the same frequency difference away from its<br />
neighbors, but each wave will have a different height corresponding<br />
to the energy within its frequency increment. Typically, wave<br />
period incrementsshould not be greater than 2 seconds to correctly<br />
define the effects of wave period variability. Wave heights in3 ft<br />
(lm) increments are considered acceptable.<br />
Alternatively, the regular waves may be selected so that they each<br />
have the same energy (height). The area under the sea spectrum is<br />
decided into bands of equal area. Either the centroid frequency<br />
(first frequency moment divialedby area) or the zero-uncrossing<br />
frequency (square-root of the second frequency moment divided by<br />
area) of the frequency band is used as the regular wave frequency.<br />
Regardlessof the selectiontechnique,each regularwave is assigned<br />
a phase using some randomizing method. A number of waves, on the<br />
order of 100, should be selected to insure that the random wave<br />
record does not repeat itself during the “sampling”time.<br />
Since any “bin” in the scatter diagram is characterized by a<br />
characteristic wave height and a characteristic period, another<br />
alternativetechniquemay be used to facilitatethe work. “Bins”of<br />
unequal period (frequency) may also be used to help prevent<br />
repetition of the random wave record.<br />
6-14
6.3.1 Stress Statistics<br />
The resulting stress records are then processed to find the stress<br />
statistics. The significantstress can redetermined as four times<br />
the standarddeviation of the stress values. Stress histogramscan<br />
also be derived from the records.<br />
6.3.2 70 Percentile Spectra<br />
Time-domain analyses tend to be computation intensive, and they<br />
often require costly computer runs. Therefore, the number and<br />
extent of time-domain analyses must be kept within reason by<br />
selecting one or a few representative sea spectra for evaluation.<br />
Selecting the representativesea spectrum and the regular waves to<br />
model it will have an effect upon the resulting fatigue life.<br />
Because the fatigue damage is an accumulation over many years of<br />
exposure to mostly mundane sea conditions, the selected<br />
(representative)sea state must be an average or mean condition,<br />
with a slight hedge toward conservatism. A recommendedselectionis<br />
a spectrumalong 70 percentilewave height line, i.e. from a cell in<br />
the scatter diagram below which lie 70% of the scatter diagram<br />
probabilities. Thezero-uncrossing period would be near the median<br />
on the 70 percentile line with a slight offset to the side that is<br />
expected to produce the greater stresses.<br />
6.4 OVERVIEW AND RECOMMENDATIONS<br />
The long term wave environment, as defined by a wave scatter<br />
diagram, is usually based on measurements and hindcasting.<br />
Measurements should be reviewed as to the extent of area covered,<br />
the time lengthof coverage,and the measurementsystem. Typically,<br />
measurements are made for limited time spans. Accelerometers of a<br />
measurement system may have limitations, preventing accurate<br />
description of wave energy content in all frequency ranges and in<br />
all directions.<br />
6-15
The wave environmentdefinitionsbased on hindcast models are quite<br />
reliable. However,modelingparametersshouldbe carefullyreviewed<br />
to ensure accuracy of the data. The environment is defined by<br />
multiple “bins” in the scatter diagram, each “bin” representing a<br />
significantwave height and a zero uncrossingperiod. Each “bin” is<br />
used to generate a specific wave spectra, defining that seastate.<br />
Since wind fetch and geographic parameters differ from one area to<br />
another,mathematical formulationsdeveloped to define wave spectra<br />
in one area may not be applicable to another area. Thus, as<br />
discussed in Section 6.1 and in Appendix B, P-M, Bretschneider,<br />
ISSC, JONSWAP, etc. wave spectra should be carefully reviewed as to<br />
their applicabilityto a given geographic area.<br />
6-16
Ss@e WaveScatt@rDiagram<br />
s<br />
i<br />
9<br />
n<br />
:<br />
i<br />
c<br />
a<br />
n<br />
t<br />
u<br />
a<br />
v<br />
e<br />
H<br />
e<br />
i<br />
:<br />
t<br />
(m)<br />
12 . . . . . . . . . . . . . ...+-+-.. . .+ . . . . ...++ . . . ...++.+- . . .+-- ..-..+. . . . ...+-... . . .+ . . . . ...+- -. ..-.+<br />
I I I<br />
1 I 1 I I I<br />
I I 1 1 I 1 I ! i<br />
I<br />
11 :. . . . ...+... . . .-+..-... .+- . . . ...+- -. ..-.+-... -..+...-.-.+. . . . ...+.--- --.+-----..+. . ...--+<br />
I I<br />
I I I<br />
I I I I 0.s : 1.0 ~<br />
:<br />
I<br />
i<br />
I<br />
10 ; . . . . ...+....- . .+ . . . . ...+.-+ . . . .+. -.....+. . . . - . .+. ------+. . . . ---+------ .+-- .--..+... . ...+<br />
I<br />
I<br />
I<br />
I<br />
i<br />
i I I<br />
I<br />
I<br />
I<br />
I 1.0 ~ 2.0 ~ 1.5 :<br />
I I<br />
9 . . . . . . . . . . . . . ...+.......+-. . . . . .+--- . ...+--- . - . .+ . . ...--+. . . . ...+-..++. . . . . . . . . . . . . . . ..-+<br />
I 1 I I I I 0.5~ 1.5 ~ 2.5 ~ 3.0 ~ 0.5 [ i I<br />
8 . . . . . . . . . . . . . -.-+..... . .+.-.....+-- . -- ..+----- . .+--- . ...+. - . .-..+-... . . -+---- ...+. -. . ...+<br />
I 1.0 \<br />
I I I<br />
1 I<br />
5.0 ~ 5.5 ~ 2.5 ~ 0.5 ~ 1 I I<br />
7 . . . . . . . .. . . . . . . . . . . . . . . .. . - . . . -.+------ +.- ...-.+... . - . .+ . ...-..+. . . . . ..*..-... .. . . . . . . ..<br />
I<br />
I I I I 1 5.0 ~ 13.0 ~ 11.0~ 2.0 f<br />
I 1 I I<br />
6 . . . . . . . . . . . . . .-.+.-... . .+ . . . . ...+- - . --.-+---- . . .+ . . ...-.+. . . . ...+.-.. . . -+. -.....+- -. ...-+<br />
I I I 0.5 ~ 6.0 ~ 18.0 : 23.0 ~ 8.5 : 1.0 {<br />
1 1 1<br />
I<br />
I<br />
I<br />
5 +. . . . ...+-. . . +-.+..... . .+ . . . . ...+- - . ...++---- . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ...+<br />
1<br />
I I I 4.0 ~ 26.5 ~ 4&5 ~ 26.5 ~ 7.0 ; 2.5 ~ 0.5 ~ 0.5 ~ I 4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . - . . ...+.-.. . . .+. -.....+. . . . ...+.... . . .++- .-...+. -. . ...+<br />
I I 1.5 ~ 39.5 : 79.5 : 63.5 ~ 20.0 ~ 6.o ~ 3.0 ~ 1.5 ~ 0.5 ~ 0.5 :<br />
3; . . . . ...+..++. . .+++-....++. . . . . .+- . . . ...+. . . . . . . . . . . . . . . . . . . . . ..+.-..-. .+-= -s---+--- . ..-+<br />
I 0.5 : 50.0 : 105.0 ~ 95.5 ~ 35.0 [ 11.5 { 5.5 ~ 2.0 ~ 1.5 ~ I I<br />
I 1<br />
2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . - . . ...++... . . . . . . . . . . . . . . . . ...+-... . . .+- . . . ...+. . . ...-+<br />
I 1-5 ~ 59-5 \ 89*O ~ 34.5 : 12-O { 7.0 ~ 4.0 ~ 1.5 ~ 0.5 ~ I<br />
I<br />
1 . . . . . . . . . . . . . ...+..... . . . . . . . . . . . . . . . ...+..... . . . . . . . . . . . . . . . . ...++... . . . . . . . . . . . . . . . . ...+<br />
I<br />
2.5 ~ l&O ~ 8.0 ~ 2.5 ~ 2.5 ~ 1.5 ~ 0.5 : I I I I<br />
04 . . . . ...+... . . ..+...... . . . . . . . . . . . . . . ...+.... . . .+. -.....+. . . . ...+.... . . . . . . . . . . . . . . . . ...+<br />
2 3 4 5 6 7 8 9 10 11 12 13<br />
Zero Up-crossing Perid, Tz (SSC)<br />
W of Occurances*.5<br />
,.-.<br />
Figure 6-1 Typical Wave Scatter Oiagram<br />
?+’ ?4<br />
. ,.,<br />
I<br />
Figure<br />
6-2<br />
Platfo= with Different Dynamic Response<br />
Characteristics in Two orthogonal -is
.,..,,<br />
....- ., ;., ..<br />
....<br />
.-<br />
..<br />
..- ..<br />
(<br />
THIS PAGE INTENTI<br />
ALLY LEFT BLANK<br />
.,.
7.<br />
7.1 BASIC PRINCIPLES OF FATIGUE DAMAGE ASSESSMENT<br />
Fatiguedamage of marine structuresis typicallydetermined usingS-<br />
N curves and the linear cumulative damage rule known as Miner’s<br />
rule. The S-N curves are usually provided in design standards,<br />
where each curve is applicableto specific joint configurations.<br />
The S-N curves applicable to details with complex stress patterns,<br />
such as tubular joint interfaces, require amplification of the<br />
nominal stresses by stress concentration factors (SCFS). The S-N<br />
curves applicable to details with simple stress patterns, such as<br />
hull scantlings, often include geometric effects and therefore can<br />
be used directly with nominal stresses.<br />
Application of Miner’s rule typically implies that the long-term<br />
distribution of stress range is replaced by a stress histogram<br />
consisting of a number of constant amplitude stress range blocks.<br />
Thus, for a stress history covering many stress ranges, each with a<br />
number of cycles (N), damage for each stress block is added to<br />
produce cumulative fatiguedamage. An alternativeto this approach<br />
is based on weighting and summing the probabilitydensity functions<br />
to obtain a long-term probability density function. Total damage<br />
can then be computed based on either numerical integration or the<br />
use of Weibull shape parameter and a closed form solution. Chen<br />
(Reference 4.10) offers a short-term closed form method that<br />
facilitates spectral fatigue analysis. Further discussion on this<br />
subject is presented in Section 6.2.<br />
As discussed in Section 4.1, various recommendations, rules and<br />
standards differ in defining desirable fatigue lives and the<br />
specifics and applications of S-N curves. However, these<br />
recommendations,rules and standards (References 1.5, 1.6, 1.7 and<br />
4.14) generally adhere to the following basic principles of fatigue<br />
damage determination:<br />
7-1
●<br />
Fatiguetest data should be carefullyevaluated andS-N curves<br />
should be generated by statisticalmeans to allow estimation<br />
of failure probabilityand incorporationof conservatism into<br />
the design. Separate S-N curves should be applicable to<br />
different weld details and in some applications to different<br />
profiles.<br />
●<br />
S-N<br />
not<br />
curves include a level of fabricationeffects that should<br />
be exceeded.<br />
●<br />
The cumulative fatigue damage computation should be based on<br />
Miner’s rule, and should consider the damaging effects of all<br />
loadings (both global and local).<br />
Fatigue damage assessment technology has benefitted from the<br />
application of fatigue crack growth data and fracture mechanics<br />
analysis of defects. In addition to predicting fatigue life,<br />
fracture mechanics analysis allows better understanding of various<br />
parameters that affect the behavior of welded joints. In turn,<br />
experimental data and fracture mechanics analysis have allowed<br />
upgrading of recommendedS-N curves (References1.5, 1.7) including<br />
Gurney’s work on the influence of plate thickness (Reference 7.1).<br />
7.2<br />
S-N CURVES<br />
The S-N curves recommended by various rules, recommendations and<br />
codes are based on the application of constant amplitude stress<br />
cycle on various detail/joint geometries in the laboratory until<br />
fatigue failure. Most S-N curves for simple details (stiffener,<br />
cutout, etc.) account for the local notch stress and can be used<br />
with the member nominal stresses. Tubular joints of offshore<br />
structures exhibit a wide variety of joint configurations and<br />
details. Therefore, while the S-N curves account for several<br />
parameters (platethickness,weld profile), theydo not account for<br />
peak stresses, requiring the application of SCF’S on computed<br />
nominal stresses to obtain peak (hot-spot)stresses.<br />
7-2
The S-N curves that can be used directly with the nominal stresses<br />
most often apply to ship structuredetails. Munse’s SSC-318 report<br />
(Reference 1.3) documents the S-N curves for 69 ship structure<br />
details and refers to earlierwork by Jordan and Cochran (Reference<br />
7.2) on in-service performance of ship structure details.<br />
Tubular offshore components have more complex geometries and are<br />
subjected to corrosive ocean environment, requiring careful<br />
assessment of all parameters contributing to fatigue failure and<br />
selection of appropriateS-N curves.<br />
Many design, fabrication and in-service factors affect the fatigue<br />
lives of details/joints. Fatigue cracks in welded joints often<br />
initiate at weld discontinuities introduced during fabrication.<br />
Weld quality problems that contributeto the degradation of fatigue<br />
strength include:<br />
●<br />
●<br />
●<br />
●<br />
●<br />
Planar defects in the body of the weld<br />
Incompletepenetration<br />
Imperfectweld root quality<br />
Imperfectweld toe profile<br />
Development of an embrittled heat affected zone (HAZ)<br />
Fatigue assessment requires definition of the number of applied<br />
stress cycles (N). Welded details/joints subjected to repeated<br />
cyclic stresseswill go through several stages of crack growth. For<br />
each hot-spot stress range (s), failure is assumed to go through<br />
three stages:<br />
●<br />
●<br />
●<br />
First discernible surface cracking (Nl)<br />
First through-wall cracking (N2)<br />
Extensive cracking and end of testing (N3)<br />
Ideally, cracks should be large enough to detect, yet not large<br />
enough to cause failure and alterationof load path. To ensure that<br />
cracks are repairable, the number of cycles to failure in fatigue<br />
7-3
assessmentis typically identifiedas the number requiredto produce<br />
through-wallcracking (N2),which can often be visually detected in<br />
a laboratory environment. To ensure accuracy of results tubular<br />
joints being tested in a laboratory are sometimes pressurized and<br />
the<br />
number of cycles to N2 is tied to the first drop in pressure.<br />
Tests are carried out for numerous stress range blocks to determine<br />
the number of stress cycles needed to reach failure, allowing<br />
development of an S-N curve. An S-N curve is also based on<br />
idealized laboratory conditions that may not fully represent the<br />
actual fatigue life in a marine environment. As discussed in<br />
Section 4.2.2, the S-N data for offshore components are based on<br />
testing of fillet-weldedplates and small-scaletubularjoints. The<br />
test data on Figure 7-1 indicate substantial scatter and allow<br />
development of S-N curves for a 99% confidence 1evel or a 95%<br />
confidence level (representingthe characteristic strength at two<br />
standard deviations).<br />
The use ofan S-N curve based on strictly small specimen data is not<br />
advisable. Smal1 test specimens usual1y do not depict welded<br />
offshore component details accurately as full-scale component<br />
fabricationresidual stresses are substantiallydifferent from test<br />
specimen residual stresses. Further discussion on size effect of<br />
welded joints is presented by Marshall (Reference 7.3).<br />
It is also necessary to consider definition of hot spot stress<br />
1evels. API reconnnendedX and X’-curves (with and without smooth<br />
transition of weld profile at weld toe) are derived from hot spot<br />
stresses obtained from strain gages placed within 0.25 inch (6 mm)<br />
to O.lRt of the weld toe. The hot spot stresses as obtained are<br />
less severe than the local stress concentrations at the weld toe,<br />
but the S-N curve developed accounts for this difference. DEn<br />
Guidance Notes (Reference1.6) defines the hot spot stress as “that<br />
which is as near the weld as possible without being influenced by<br />
the weld profile”.<br />
7-4
The primary factors that influence the fatigue life assessment are<br />
discussed as follows:<br />
7.2.1 Desicm Parameters<br />
The design is optimized to ensure effective resistance of marine<br />
structures to both extreme and operating fatigue loads. Typically<br />
thestructureandjoint/detail configurationsshould redeveloped to<br />
minimize stress concentrations and stress levels, and arranged to<br />
provide easy access to help maintain welding quality. The material<br />
should be selected to have an acceptable chemical composition to<br />
ensure weldability and satisfactorymechanical properties to ensure<br />
notch toughness.<br />
Fabrication specifications should permit only minimized mismatch<br />
tolerances, thereby reducing SCF’S and residual stresses. They<br />
should also controlthe quantity and quality of repair work, thereby<br />
ensuring allowabledefects in weldments comply with specifications.<br />
These design parametersare discussed in Section3. and described in<br />
more detail below.<br />
Material Strenctth<br />
Fatigue strengths of marine structure components are sometimes<br />
assumed to be affected by material strength. Cast steel node or<br />
forged components of a structure have significant fatigue crack<br />
initiation periods and material strength may have an effect on<br />
fatigue lives. However, material strength does not affect the<br />
fatigue life of welded components of marine structures. As-welded<br />
joints of marine structures contain inherent flaws and Maddox<br />
(Reference 7.4) has shown that the fatigue 1ife of such joints is<br />
largely expended in crack propagation. While increased material<br />
strength retardscrack initiation,the rate of crack growth has been<br />
shown to be insensitive to material strength. Experimental work<br />
carried out by Hartt et al (Reference 7.5) on high strength steel<br />
(HSS) specimens in a corrosive ocean environment indicated fatigue<br />
7-5
damage accumulation similar to that of structural steel. Gurney<br />
(Reference 7.6) indicatesthat increased material tensile strength<br />
does not increase fatigue resistance and implies that a fatigue<br />
design approach incorporatingmaterial tensi1e strength is not valid<br />
for welded marine structures.<br />
The effect of initial flaw size on<br />
affecting crack propagation should<br />
size estimatingprocedureby Grover<br />
in assessing fatigue crack growth.<br />
fatigue life and the parameters<br />
be understood. An initial flow<br />
(Reference7.7) is quite helpful<br />
Plate Thickness<br />
Current S-N curves reconunended by DnV (Reference 1.7), DEn<br />
(Reference 1.6) and AWS (Reference 4.13) incorporate a thickness<br />
correction factor. DnV and DEn recommendations largely reflect<br />
early work by Gurney (Reference 7.1) and many test programs<br />
corroborating plate thickness effect corrections proposed by<br />
Gurney. Class B, C, D, E, F, F2, G and N curves are applicable to<br />
non-tubular (including tube-to-plate) joints based on detail<br />
geometry, stressing pattern and method of fabrication/inspection.<br />
While these eight classes are applicablewithout correctionto plate<br />
thicknessupto7/8inch (22MM), class Tcurve (for tubular joints)<br />
is applicable to 1-1/4 inch (32 mm) plate.<br />
The UK DEn Guidance Notes recoimnendspecific size effect (i.e.,<br />
plate thickness) correction factors in the following form:<br />
S = Sb (32/t)l/4<br />
7“1<br />
where<br />
s =<br />
fatigue strength of a joint under consideration<br />
(N/nun2)<br />
7-6
‘b =<br />
fatigue strength of a joint applicable to T curve<br />
for 32 & wall thickness (N/mm2)<br />
t = wall thickness of a joint under consideration (nnn)<br />
Although the tubular joint test data available may be insufficient<br />
to document the size effect throughout the range of plate<br />
thicknesses in use, the data available has been grouped, analyzed<br />
and relative fatiguestrengthdata documented. Tolloczko and Lalani<br />
(Reference7.8) report that sizeeffect is adequatelyrepresentedin<br />
the Guidance Notes (Reference 1.6) and that none of the more than<br />
300 datapoints fall below the applicableS-N curves.<br />
Test results show that plate thickness or scale increases can<br />
adversely affect fatigue strength, perhaps due to increase in weld<br />
toe stresses with an increase in plate thickness. S-N curves<br />
modified to account for thickness-effect of thick plates often<br />
substantially affect the fatigue lives computed. Some experts<br />
consider the applicable plate thickness correction to be mild for<br />
typical nodes. However, additionalwork by Maddox (Reference 7.9)<br />
indicates that thickness correction may be too severe if only the<br />
primary plate thickness is increased. His work on cruciform-type<br />
joints (Figure7-2) indicatesthat the joint proportions ratio (L/B)<br />
has greater effect on fatigue strength than does the primary plate<br />
thickness.<br />
.-<br />
While Maddox’s encouraging results are applicable to joints<br />
subjected to axial tension, increased primary plate thickness<br />
subjected to bending stresses still adversely affects the fatigue<br />
life. A typical joint in most marine structures is likely to be<br />
subjected to substantial bending stresses. Thus, before any<br />
relaxation of plate thickness effect on the S-N curves is attempted<br />
further data are necessary for a range of geometries and combined<br />
loading conditions.<br />
7-7
“FabricationRestrictions<br />
Fabrication specifications and drawings often attempt to minimize<br />
the conditions that may adversely affect fatigue strength of a<br />
detail/joint. Fatigue tests performed on various types of joints,<br />
and fracture mechanics analysis carried out by Maddox (Reference<br />
7.10), indicate that the fatigue life of a joint does not change<br />
appreciablydue to attachmentof a backing bar on a plate. Fatigue<br />
strength also”hasbeen shown to be unaffectedby poor fit-up between<br />
the backing bar andthe plateor by the configurationof the backing<br />
bar. However, it should be emphasized that fatigue strength not<br />
changing appreciably due to attachment of a backing bar or a poor<br />
fit-up may have more to do with the root condition without backing<br />
bar.<br />
7.2.2<br />
Fabrication and Post-FabricationParameters<br />
Fabricationparameterscoverall of the fabrication activities that<br />
affect the quality of welded details/joints. These parameters,<br />
ranging from welder qualificationto heat input and cooling rates,<br />
were identifiedon Figure 3-3 and discussed in Section 3.1.2.<br />
Misalignments<br />
Misalignments adversely affect the fatigue strength of a<br />
detail/joint. When a misalignment between two elements is large,<br />
both elementsmay haveto be improperlydeformedto align them prior<br />
to welding. Such joints incorporatesubstantialresidual stresses.<br />
If the misalignment between two elements is small, they may be<br />
welded as-is, but the misalignmentcauses a stress concentrationdue<br />
to the resulting secondary bending.<br />
Because misalignment increasesthe stress at the weld toe of joints<br />
loaded axially, the stress magnification factor (Kc) can be<br />
correlated to fatigue damage. Fatigue test results for different<br />
levels of misalignment in plate joints and tubulars carried out by
Maddox (Reference 7.11) provide the basis for assessment of<br />
misalignments.<br />
Weld Quality<br />
A significant scatter of fatigue life test data is expected and<br />
appropriatelyaccountedfor. Acharacteristic strengthrepresenting<br />
a 95% confidence level in test data may be used to assess data<br />
points falling substantiallybelow the S-N curve. Such data points<br />
are likely to be due to a problem with the welding procedure or the<br />
welder qualification. Weld quality degradation (and therefore<br />
fatigue life degradation) due to incomplete penetration and poor<br />
weld root quality can be minimized by developing a welding<br />
specification applicable to the specific configuration and closely<br />
adhering to it during fabrication. Weld quality degradation due to<br />
undercut at the weld toe can be similarly minimized.<br />
Weld Toe Profile<br />
The significance of weld profiles on joints subjected to fatigue<br />
loading is controversial. Substantial time and expenditures are<br />
necessary to prepare a favorable weld profile, and weld profiling<br />
may increasewelding costs by as much as 20%. Thus, weld profi1ing<br />
is limited to specific tubular joints of discrete marine systems.<br />
While API RP2A does “not recognize and quantify plate thickness<br />
effects, theAPI S-N curves recognize and quantifyweld profile. As<br />
illustrated on Figure 4-3 in Section 4.1.2, API (Reference 1.5)<br />
recommends the use of an X-curve for welds with a favorable profile<br />
while the X’-curve is reconnnendedfor welds without such a profile.<br />
As il1ustrated on Figure 7-3, substantial preparation, weld bead<br />
shape, applicationof extra weld beads and grinding may be necessary<br />
to allow the use of an X-curve.<br />
7-9<br />
! “-7 /<br />
[/
Fatigue strength of a tubular joint is shown to improvedue to weld<br />
profiling (References 7.12 and 7.13). Weld profiling (including<br />
grinding of weld toe) has two primary benefits:<br />
● It can minimize the potential for crack propagation by<br />
removing inherent crack-like flaws.<br />
● It can reduce stress concentrations by improving local weld<br />
profile.<br />
However, grindingto remove flaws and to provide a smooth transition<br />
between the weld and parent material is not universally accepted as<br />
quantifiable benefit unless the weld toe undercut is sufficient.<br />
Both AWSand API do not require a correctivemeasure if the undercut<br />
of weld toe is less than 0.01 in. (See Figure C 10.7.5, Reference<br />
4.14). DnV (Section3.3.1 , Reference 1.7) states, “the effect of<br />
weld profiling giving the weld a smooth concave profile compared<br />
with the typical triangular or convex shape~ improve the fatigue<br />
properties.”Although DnV accepts the use of an X-curve (in lieu of<br />
a T-curve) provided weld profiling is carried out, it also<br />
stipulates that the effect of profiling on the S-N curves will be<br />
considered for each case separately.<br />
The weld profiles applicabletoAPI XandX’ S-N curves are shown on<br />
Figure 7-3. However, to ensure that the flaws at weld toe are<br />
removed, grinding or AWJ process should result in sufficient<br />
undercut at the weld toe. The minimum undercut recommended by the<br />
DEn Guidance Notes (Reference 1.6) is shown on Figure 7-4.<br />
Further discussion and an excellentoverview of the effects of weld<br />
improvementtechniques is provided by Bignonnet (Reference7.14).<br />
7.2.3<br />
Environmental Parameters<br />
The environment in which fatigue cracks initiate and propagate<br />
substantiallyaffectsfatiguelife. The amplitude,distributionand<br />
7-1o
frequency of loading identify severity of the fatigue environment.<br />
Although a structure’sconfigurationcan be optimizedto reduce the<br />
stress range, the site-specificenvironmental loading controls the<br />
choice of fatigue design and analyses method.<br />
An environmental parameter that affects fatigue is either air or<br />
seawater. Because of the adverse effects of seawater corrosion on<br />
fatigue strength, adesign factor is often applied for fatigue life<br />
in a seawater environment. However, an effective cathodic<br />
protection systemwill reduce or prevent seawater corrosion, and if<br />
such a system is used, the design factor may be deemed unnecessary.<br />
This approach (and its inclusion in various rules, recommendations<br />
and standards) is based on corrosion fatigue test data on welded<br />
plate specimenswith and without cathodic protection.<br />
Environmentaleffects on welded flat plates have been assumed to be<br />
the same as those on tubular joints. However, Wylde et al<br />
(Reference 7.15) have indicated that the corrosive effect of<br />
seawater on tubular joints may be greater than the effect on flat<br />
plate specimens. Althoughdifficultto document, tubular jointsmay<br />
be more susceptibleto environmentaleffects than small welded flat<br />
plates due to scale effects, including initial flaws. Flat plates<br />
may have longer fatigue lives as substantial time will be expended<br />
in initiation of flaws.<br />
7.3<br />
FATIGUE DAMAGE COMPUTATION<br />
State-of-the-art methodology for determining fatigue lives and<br />
designing structures with fatigue lives in excess of the design<br />
lives is primarily based on S-N curves and the cumulative damage<br />
rule. The cumulative damage rule is an approach used to obtain<br />
fatigue damage by dividing the stress range distribution into<br />
constant amplitude stress blocks, assuming that the damage per load<br />
cycle is the same at a given stress range.<br />
7“11
Current recommendations,rules and standardsuniformlyallowthe use<br />
of Miner’s rule to compute the cumulative damage. Applicable<br />
cumulative damage rules are discussed in this section, followed in<br />
Section 7.4 by a discussion of stress spectrum in the context of<br />
fatigue damage computation.<br />
7.3.1<br />
Miner’s Rule<br />
The damage for each constant stress block is defined as a ratio of<br />
the number of cycles of the stress block required to reach failure.<br />
Thus, the Palmgren-Minerlinear damage rule defines the cumulative<br />
damage (D) for multiple stress blocks as equal to:<br />
‘i<br />
D=:T < 1.0<br />
i=l i<br />
As briefly discussed in Section 3.2.5, Miner’s rule can either<br />
overpredict or underpredict the cumulative damage.<br />
One source of inaccuracy regarding cumulative damage is the<br />
applicationof constantamplitudestressblocks; itmay be important<br />
to be able to predict the fatigue damage due to variable amplitude<br />
loading. Another source of inaccuracy is the sequence of loading;<br />
while Miner’s rule cannot account for the loading sequence,<br />
occurrence of large amplitude loads early in fatigue life can<br />
accelerate the rate of crack growth. Another source of inaccuracy<br />
for wide band processes is the choice of cycle counting method,<br />
which is further discussed in Section 7.4.<br />
Despite these sources of potential inaccuracy,Miner’s rule is used<br />
to compute fatigue damage because of its simplicity as well as its<br />
ability to predict fatigue damage conservativelymost of the time.<br />
Other uncertainties in determining wave environment, wave loading<br />
and hot-spot stresses contribute far more to the inaccuracy of<br />
7-12<br />
[7 ‘cd<br />
1/
fatigue damage predictions. Fatigue analysis assumptions also<br />
contribute to the inaccuracyof fatigue damage predictions. As an<br />
example, 10 to 15 stress blocks, each representing a significant<br />
wave height and a zero-crossing point, may be used in the fatigue<br />
analyses. Theuseof40to 50 stress blocks is desirable, but often<br />
considered impracticalfor most analyses.<br />
7.3.2<br />
Alternative Rules<br />
The ability to use servohydraulic testing machines and to apply<br />
computer-controlled loads has allowed testing of a substantial<br />
number of specimens subjected to variable amplitude loading<br />
(References 7.16, 7.17, 7.18 and 7.19). Gerald et al (Reference<br />
7.20) provide an excellent overview on variable amplitude loading.<br />
Some analytical work carried out and many of the test results show<br />
that Miner’s rule is realistic and conservative. However, some of<br />
the test results also show that Miner’s rule may lead to<br />
underpredictionof fatigue damage.<br />
One source of discrepancymay be crack growth fluctuations. Stress<br />
block procedures used in tests result in the application of high<br />
tensile stresses, which can retard crack growth. Test specimens<br />
subjected to random loadings are less likely to have similar high<br />
tensile stresses. Another source of discrepancy is the counting of<br />
stress cycles. Gurney (Reference 7.17) and Trufiakov (Reference<br />
7.21) conclude that small fluctuations superimposed on each stress<br />
cycle add substantiallyto fatigue damage.<br />
Miner’s rule is the acceptedmethod for fatigue damage computation.<br />
However, since alternativesto Miner’s rule have been proposed it is<br />
beneficial to review one such rule.<br />
Gurney proposes a damage rule by expressing the applied stress<br />
spectrum in terms of the maximum stress range (Smax),the number of<br />
cycles (ni) applied at proportions (pi) of smax, and its length<br />
(1 ni)defined as the block 1ength. Gurney’s rule states:<br />
7-13
n<br />
NB=T<br />
1<br />
[p=+pi<br />
Ei<br />
.<br />
‘c<br />
where:<br />
NB. = predicted life in blocks<br />
NC = constant amplitude life at SmaX<br />
NEI =<br />
i =<br />
number of cycles per block 2 ‘i ‘max<br />
lton<br />
This product rule can be compared to Miner’s<br />
NB=~<br />
‘c<br />
x Pi ni<br />
1<br />
where m is the slope of the S-N curve expressed as SmN - constant K<br />
It should be noted that Gurney’s rule may also result in<br />
underprediction of fatigue damage. Study of spectrum shape and<br />
block length (Reference7.22) indicatesthat for long block lengths<br />
Gurney’s rule may be unsafe.<br />
7.4 STRESS HISTORY AND UPGRADED MINER’S RULE<br />
7.4.1 Background<br />
Miner’s linear cumulativedamage rule can be used safely, provided<br />
some of the wave environment uncertainties (including counting of<br />
cycles and evaluatingthe stress ranges compatible with cycles) are<br />
properly accounted for.<br />
7“14
Typically, the sea state represented by joint probabilities of<br />
significant wave heights and characteristic periods (scatter<br />
diagram) is applied to the transfer function to produce the stress<br />
range spectrum. Integration of the spectra provides a number of<br />
statistical parameters, such as the bandwidth, the zero-uncrossing<br />
frequency, etc., allowing development of short-term probability<br />
density functions.<br />
The short-term probabilitydensity function of the stress range for<br />
each significant wave height and its characteristic period is<br />
generally defined by using a Rayleigh distribution. For this<br />
assumption to be valid, (1) a large number of sea states must be<br />
used, and (2) the stress cycles can be considered narrow-banded.<br />
Individualstress cycles are considerednarrow-bandedwhen they are<br />
readily identifiable and there is no ambiguity in counting the<br />
stress cycles. The wide-banded loadings exhibit smaller stress<br />
cycles interspersed among larger stress cycles. Because it is<br />
difficult to define the stress cycles, different cycle counting<br />
methods result in different fatigue damage predictions.<br />
Rainflow counting is the name of a large class of stress cycle<br />
counting methods, including the original rainflow method, Hayes<br />
method, range-pair counting, range-pair-range counting, ordered<br />
overall range counting, racetrack counting and hysteresis loop<br />
counting.<br />
Rainflow counting and other alternatives are briefly discussed in<br />
Sections 7.4.2 and 7.4.3, respectively, to illustrate the options<br />
availableto upgradeMiner’s rule. However, it should be noted that<br />
two very important variables affecting fatigue life computation<br />
should be addressed in any attempt to upgrade Miner’s rule:<br />
(1) S-N curves are based on constant amplitude stress blocks and<br />
should be compared against variable amplitude results.<br />
7-15
(2) Damage computation does not account for stress sequence and<br />
may overpredict fatigue lives ofjoints/details subjected to<br />
large stress amplitude ranges early on, accelerating crack<br />
propagation.<br />
7.4.2 Miner’s Rule IncorDoratinqRainflow Correction<br />
The rainflowcountingprocedure ismore accuratethan other counting<br />
methods because the rainflow procedure is based on counting the<br />
reversals in accordancewith the material stress-strain response.<br />
Modified Miner’s rule uses the rainflowcycle counting procedure but<br />
does not require the stress process to be simulated.<br />
D =#E (Sm)<br />
where:<br />
n =<br />
K=<br />
E(Sm) =<br />
s =<br />
total number of cycles<br />
constant, equal to SmN<br />
the mean value of S<br />
a random variable denoting fatigue stress cycles<br />
If the process<br />
D can be shown<br />
is stationary, Gaussian and narrow band, the damage<br />
that:<br />
where:<br />
u =<br />
ro =<br />
RMS of the process<br />
gamma function<br />
7-16
When the structure response yields narrow-banded stress cycles, the<br />
choice of counting method is imaterial. Even for moderately wide<br />
band stress cycle histories, the various cycle counting methods<br />
produce similar fatigue damage predictions. The choice of counting<br />
method becomes significant only for wide band stress histories with<br />
an irregularityfactor equal to or less than 0.5. The irregularity<br />
factor is a measure of the band width, defined as the ratio of mean<br />
crossings with positive slopes to the number of peaks or valleys in<br />
the stress history.<br />
7.4.3<br />
Other Alternatives<br />
An alternativeapproachto predicting fatigue damage under wide-band<br />
stresses is to use the narrow-band stress approach and apply an<br />
adjustment factor. Assuming a narrow band fatigue stress with the<br />
same RMS, and the same expected rate of zero crossings, fo, as the<br />
wide band stress, a damage estimate can readily be carried out.<br />
Given the spectral density of the stress w(f), the kth moment of of<br />
spectral density function mK is equal to:<br />
f= fK w(f) df , while the<br />
‘K= 0<br />
RMS (Std dev.) = ~= i%,<br />
and the expected rate of zero crossings<br />
with slope<br />
f. = d~/mo<br />
With this equivalent narrow band process, the fatigue damage can be<br />
predicted by the following closed form solution:<br />
7-17
DNB = (f. T/K) (2/2 U)m r (f+ 1)<br />
where<br />
n = f. T<br />
T= design life<br />
Wirsching (Reference 7.23) proposes that the fatigue damage be<br />
expressed as:<br />
o = a ‘NB<br />
where A is the adjustmentfactorto fatiguedamage predicted based on<br />
a narrow-band stress. Thus, the rainflow counting effect to fatigue<br />
damage can be incorporated directly if J is known. An empirical<br />
formula proposed by Wirsching is as follows:<br />
A (E, m) = a(m) + [1-a(m)] (LE)b(m)<br />
where<br />
a(m) = 0.926 - 0.033 m<br />
b(m) = 1.587m - 2.323<br />
Thus the fatigue damage obtained by incorporating the narrow-band<br />
adjustment factor, A provides a closed-form formulation. The<br />
empirical formula allows fatigue damage predictions quite close to<br />
those obtained by incorporatingthe direct rainflow method.<br />
The A parameter introduced by Wirsching is an equivalent rainflow<br />
adjustment factor intendedto correct the slight conservatism of the<br />
Rayleigh distribution. Whether a closed-form or a numerical<br />
integrationis carriedout, short-termstatisticsand the probability<br />
density function allow obtaining of partial damage, weighting and<br />
summing of all damages.<br />
7-18
Following the weighting of the short-term density functions, the<br />
long-term density functions for the structure’s design life are<br />
obtained. While the cumulative damage may be computed through<br />
numerical integration, an approximation is introduced to allow<br />
application of a closed-form solution. Typically, a Weibull shape<br />
parameter (Weibull distribution) is used in predicting cumulative<br />
fatigue damage based on the 1ong-term, closed-form method. This<br />
subject is discussed further in Section 6 and in a comprehensive<br />
paper by Chen and Mavrakis (Reference7.24).<br />
7.5<br />
OVERVIEW AND RECOMMENDATIONS<br />
7.5.1<br />
Application of S-N Curves<br />
The S-N curves used indeterminingfatiguedamage computations should<br />
be compatible with structural details investigated. The S-N curve<br />
including the effect of peak stresses should be used together with<br />
nominal stresses at the detail, while the S-N curve uninfluenced by<br />
the weld profile should be used with nominal stresses increased“by<br />
appropriate SCFS.<br />
Scatter in fatigue test data should also be appropriately accounted<br />
for. One primary parameter affecting scatter of S-N data may be<br />
plate thickness. As plate thickness increases higher localized<br />
stresses will occur near plate surface, accelerating propagation of<br />
fatigue cracks. Consideringthat small specimen S-Ndata needto be<br />
adjusted for scale effects and a reasonable confidence level should<br />
be achieved, S-N curves may be obtained assuming 95% to 97.5%<br />
confidence level and a log normal distribution.<br />
There are other parameters that are difficult to assess yet they<br />
affect the crack growth and fatigue failure, causing substantial<br />
scatter of S-N data points. One importantconsideration is the size<br />
of initialflaw (crack)and another is the number of flaws. Although<br />
further work is necessary, Morgan’s (Reference 7.25) findings on<br />
7-19
interactionof multiple fatigue cracks provide valuable insight into<br />
scatter of S-N data points.<br />
Additional parameterscontributingto the fatigue life uncertainties<br />
are the effects of corrosive sea water environment and the<br />
implications of long-life regime. Although catholically protected<br />
offshore structure components in sea water are assumed to have the<br />
same fatigue resistance as those components in air, the basis for<br />
this assumption is the test data for simple plate specimens. Some<br />
large scale tubular joint tests indicate (Reference 7.15) that the<br />
corrosive effects of seawater on tubular joints may be greater than<br />
the effect on small flat specimens. More test data is necessary to<br />
quantify corrosive effects.<br />
There are limited number of test data in long-life regime. As a<br />
result, some codes do not provide endurance limit, some have a<br />
changing slope and some have a definite plateau at different number<br />
of cycles. These and other uncertainties require further research<br />
work to upgrade current S-N curves. Current research efforts on<br />
fatigue resistance are summarized in Section 9.<br />
The S-N curves recommended byAPI, DEn and DnV (References 1.5, 1.6<br />
and 1.7) may be used in the computation of fatigue damage. While<br />
most early S-N curves were based on AWS data, current DEn curves are<br />
largely based on work at the Welding Institute (primarilyGurney and<br />
Maddox). DEn Guidance Notes also provide tables, allowing the<br />
selection of S-N curves for specific details. For ship structure<br />
details, appropriate DEn S-N curves can be selected based on<br />
judgement in assessing the details and tables. Earlier works by<br />
Munse (Reference 1.3) and Jordan and Cochran (Reference 4.4) can be<br />
used directly or in comparison of component test data for ship<br />
structure details.<br />
The S-N curves given in DEn Guidance Notes are applicable to a base<br />
case plate thickness of 7/8 inch (22 mm), requiring an adjustment of<br />
the S-N curves for thicker plates. Consideringfurther validationof<br />
7-20
thickness effect is necessary and the ship structure plate<br />
thicknesses are not excessive, the correction factor may be<br />
neglected.<br />
The S-N curves reconanendedbyAPI for offshore platforms may be used<br />
in the computation of tubular component fatigue damage. The API X-<br />
curve and the DEn T-curve (identicalto DnV T-curve up to 10 million<br />
cycles for catholically protected areas - see Section 4.2.2)<br />
intersect at about 500,000 cycles and would yield similar lives for<br />
a plate thicknessof 1-1/4 inch (32nmI). Most tubular chord and stub<br />
thicknesses are likely to be greater than 1-1/4 inches and the<br />
applicationof correctedDEn or DnV T-curves to compute fatigue lives<br />
will result in shorter lives and considered to be appropriate.<br />
Consideringthe effectsof plate thickness,weld profile and undercut<br />
on fatigue strength and the S-N curves it may be prudent to reassess<br />
the hot spot stress range concept. Tolloczko et al (Reference 7.8)<br />
recommend modifying the definition of hot spot stress range to<br />
reflect weld toe defects. Then, the S-N curves will reflect only the<br />
size effects.<br />
7.5.2 Fatiaue DamacteComputation<br />
Fatigue lives determined based on S-N curves and Miner’s cumulative<br />
damage rule are uniformlyacceptableto certifyingand classification<br />
agencies. The national and internationalstandards allow the use of<br />
simple cumulative damage rule for the computation of damage. Large<br />
number of test results as well as the in-serviceperformance records<br />
of marine structures indicate adequacy of this approach.<br />
Alternative rules to compute fatigue damage and methods to upgrade<br />
Miner’s rule have been proposed. Although necessary to evaluate<br />
possible benefitsof such alternatives,additionalcomplexity and the<br />
cost should also be considered. Since the S-N curves are developed<br />
based on constant amplitude stress ranges, the effect of variable<br />
7-21
amplitude loading and loading sequence on fatigue life is a valid<br />
concern.<br />
The results obtainedfrom a substantialnumber of specimenssubjected<br />
to variable amplitude loading show that Miner’s rule is appropriate<br />
and generally conservative. Dobson et al (Reference 7.26) studied<br />
loading histories of containershipsbased on recorded service data.<br />
When the stress intensity ranges were expressed as the root-meansquare,<br />
the crack growth of laboratory specimens subjected to<br />
constant-amplitude loading history compared quite well with those<br />
specimens subjected to constant amplitude loading.<br />
Fatigue damage computation is based on stress ranges and number of<br />
cycles and does not account for stress sequence. Since welded<br />
structure fatigue lives are largely expended in crack propagation,<br />
applicationof sufficientnumber of large stress amplitudes early in<br />
fatigue life is likely to accelerate crack propagation and<br />
overpredictingof fatigue life. The uncertainty of stress sequence,<br />
aside, the use of rainflowcounting procedure, based on counting the<br />
reversals in accordancewith the material stress-strainresponse,may<br />
enhance accuracy of damage computation. However, improvement in<br />
accuracy is significantonly for wide band stress histories with an<br />
irregularity factor equal to or less than 0.5. When the structure<br />
response yields narrow-banded stress cycles, the choice of counting<br />
method is innnaterial. Even for moderately wide band stress cycle<br />
histories,the variouscycle countingmethods produce similar fatigue<br />
damage predictions. Although further research is necessary,<br />
especially on the effect of stress sequence, the use of S-N curves<br />
and Miner’s cumulative fatigue damage rule is appropriate.<br />
7-22
:Ilo<br />
I Cn<br />
80<br />
to<br />
40<br />
.<br />
20<br />
T-<br />
m 10<br />
m -k<br />
8<br />
u<br />
A.<br />
c<br />
*<br />
.-i-<br />
4,<br />
1“<br />
,.<br />
z ȯnaxxs<br />
HLL SICKI S<br />
“ $lReSs nrwct<br />
I<br />
oa—<br />
x +<br />
Figure 7-I S-N Curve for a Transverse Butt Weld and Test Data<br />
#’+’7,<br />
—<br />
I<br />
13 20 30 Lo 50<br />
ThiCkn@SS. s mm<br />
Figure 7-2 Theoretical Thickness Effect for a Cruciform Joint<br />
(From Reference 7.9)<br />
.,
n“<br />
CAP PASSES<br />
ROOT = *<br />
{<br />
A A 4 I<br />
u<br />
A) WITH PROFILE CONTROL<br />
1<br />
J 4. AI<br />
B) WITHOUT pROFILE CONTROL<br />
1<br />
Figure 7-3 Weld Profiles for API X and X’ S-N Curves<br />
(From Reference 1.5)<br />
Bmce —<br />
Defect -<br />
Dcpthof grinding<br />
should be 0.5mm<br />
below bottom of<br />
ony visible<br />
und~rcut<br />
B..-<br />
m<br />
Depth<br />
.+<br />
of grinding<br />
should be 0-5 mm<br />
below bottom of<br />
any vlslbie<br />
undercut<br />
Chord<br />
Chord<br />
Defect<br />
in Brace<br />
Defect<br />
in Chord<br />
Gridne mMd lac mngenudly m the plsu mrha 8s m A. WIII ~<br />
Iiuk mpmwm!m m strm@. Grindingmm atcd kku k ptuc<br />
3urhcGssmB.inoIdcrlOrcInOvclm*-<br />
Figure 7-4 DEn Guidence Notes Recommended Weld Profiling and Undercut<br />
(From Reference 1.6)
8.<br />
FATIGUE DUETO VORTEX SHEDDING<br />
This section specificallyaddresses fatigue due to vortex shedding.<br />
Fatigue due to vortex-induced vibrations is different from other<br />
forms of fatigue discussed in previous sections only in its loading<br />
characteristics. Generally, relatively small number of slender<br />
members are susceptible to vortex-induced fatigue. However,<br />
response to vortex shedding cannot be predicted using conventional<br />
dynamic analyses techniques because the problem is non-linear. In<br />
compliancewith project objectives, a brief discussion is presented<br />
on vortex sheddingphenomena,analysisand design, damage assessment<br />
and avoidance. A comprehensive discussion, including example<br />
problems, is presented in Appendix D.<br />
VORTEX SHEDDING PHENOMENON<br />
8.1.1<br />
Background<br />
Amember exposedto fluid flow may be subjectedto unsteadydrag and<br />
lift forces caused by sheddingof vortices. While the vortices shed<br />
are most often due to steadywind or current flow, the phenomena can<br />
occur due to combined wave and current action. Depending on the<br />
member’s natural frequency and the velocity of fluid flow, the<br />
member may experience sustained vibrations.<br />
Many structure members may be susceptible to vortex induced<br />
vibrations (VIV). Relatively large diameter cylindrical brace<br />
members of a fixed offshore platform can be designed to avoid VIV.<br />
Component members of a cargo boom on a ship or the flare structure<br />
on production units (FPSO, platform, etc.) are relatively slender<br />
and can not be readily designed to avoid VIV. Then, they need to be<br />
either designed to have adequate fatigue strength to resist the VIV<br />
over the design life of the structure or provided with devices or<br />
spoilers to modify the vortex shedding and/or member natural<br />
frequencies.<br />
8-1
It should be pointed out that the effect of wind-induced vibration<br />
is often not adequatelyaddressedduring design. The basis for the<br />
issuing of an offshore Safety Notice 7/87 by the U.K. DEn to all<br />
North Sea Operators for reassessment of platform flare boom<br />
structuraladequacywas the discoveryof fatigue cracks in the flare<br />
boom struts. Bell and Morgan (Reference 8.1) report that the<br />
original design documents revealed relatively low fatigue stresses<br />
and high fatigue lives. Reanalyses of the flare boom joints<br />
indicated that the extensive cracking observed may be due to the<br />
combined effect of poor weld quality in the joints and the largerthan-expected<br />
stress cycles due to vortex-induced vibrations.<br />
8.1.2<br />
Vortex Induced Vibration (VIV)<br />
At low fluid velocities (expressed as Reynold’s numbers) the flow<br />
acrossthe cylindricalmember remains stable. As the fluid velocity<br />
increases (i.e.,higherReynold’snumbers)the innermostpart of the<br />
shear layer adjacent to the cylinder moves more slowly than the<br />
outer part of the layer. As a result, the shear layers “roll-up”<br />
into discrete swirling vortices. These vortices are shed<br />
periodically,either in pairs (in-lineflow) or sequentially (crossflow)<br />
from two sides of the cylinder, generating unsteady and very<br />
complex pressure distribution. As illustrated on Figure 8-1 (from<br />
Reference 8.2), the laminar boundary layer goes through several<br />
stages of vortex turbulence with increasing Reynold’s numbers. A<br />
detailed discussion on vortices and pressure distribution is<br />
presented by Marris (Reference8.3).<br />
If the cylindrical member natural frequency (fn) is close to the<br />
vortex sheddingfrequency,vibrationsof the cylinder may affect the<br />
vortices shed. The vortex sheddingfrequency (fv)will no longerbe<br />
dependent onthe Strouhalnumber (St),and is likely to become equal<br />
to the natural frequency of vibration. If this “lock-in” effect<br />
materializes, further increases in the vibration amplitudes will be<br />
observed. To prevent the occurrence of critical velocity (fc),<br />
where the member natural frequency is equal to the vortex shedding<br />
8-2
frequency (i.e. fc = fn = fv), member stiffness and mass may be<br />
modified. The maximum amplitude of oscil1ation for the critical<br />
velocity is an important variable, directly affecting the stress<br />
amplitudes. The maximum amplitude of oscillation of a member<br />
depends on member support conditions and the Ks value, reaching a<br />
value approximately equal to member diameter for simply supported<br />
boundary conditiens. To prevent the lock-in effect, it is desirable<br />
to keep the member natural frequenciesto less than 70%or more than<br />
130% of the vortex shedding frequency, whenever practical.<br />
8.2<br />
ANALYSES AND DESIGN FOR VORTEX SHEDDING<br />
The interactivenature of the vortices shed and the vibration of the<br />
cylinder makes analytical prediction of response to vortex induced<br />
vibration (VIV) extremely difficult. Empirical formulations<br />
(References 8.5, 8.6 and 8.7) have been developed to reflect the<br />
state-of-the-artwith respect to VIV technology. These empirical<br />
approaches incorporate various parameters and are based on the<br />
comparison of specificparametricvalues with experimental results.<br />
Empirical formulations can be effective y used to avoid VIV, but<br />
they are less reliable at predicting the occurrence of VIV and<br />
determining the response amplitudes.<br />
8.2.1<br />
Suscelltibilityto Vortex Sheddinq<br />
Cylindrical members may experience either in-line or cross flow<br />
oscillations for a range of flow velocity and member response<br />
characteristicratios. To define susceptibilityof a member to VIV,<br />
a reduced velocity (Vr) term is introduced:<br />
v<br />
vr=—<br />
fnd<br />
where:
v= flow velocity normal to the cylinder axis<br />
fn = fundamental frequency of the member (H)<br />
d= diameter of the member<br />
Susceptibilityof a member to VIV in air is different than in water<br />
due to the density of air flowing around the member being different<br />
than the density of water. Susceptibility of a member is defined<br />
for in-line and cross-flow oscillations in both environments.<br />
In-line VIVmay occur when:<br />
l*2svr
8.2.2<br />
VIV Response and Stresses<br />
A strategy based on avoidance of VIV is quite feasible for most<br />
marine structures. Primarystructuralmembers are usually designed<br />
to be sturdy enough that they are not susceptible toVIV. However,<br />
some secondary or non-structuralmembers may be susceptible to VIV<br />
in water and in air. An empirical approach proposed by DnV<br />
(Reference 8.7) does not account for the nonlinear relationship<br />
betweenresponseanddamping,therebyyielding conservativeresponse<br />
amplitudes and stresses. To predict response amplitudes more<br />
reliably an approach based on Hallam et at (Reference 8.9) is<br />
recommended.<br />
Cross-flow oscillations due to wind action may not always be<br />
preventable, requiring the members to have sufficient resistance.<br />
An empirical formulation based on a procedure by Engineering<br />
Sciences Data’ Unit ESDU (Reference 8.6) that accounts for<br />
interaction between vortices shed and forces induced is<br />
recommended. This procedure and the basis for estimating maximum<br />
bending stresses for different boundaryconditions are discussed in<br />
Sections D.4 andD.5 of Appendix D.<br />
8.3<br />
FATIGUE DAMAGE ASSESSMENT<br />
All members susceptible to VIV should be assessed for fatigue<br />
damage. First, the fatigue damage due to VIV is calculated. Then<br />
a global fatigue analyses is performed and fatigue determined for<br />
all critical members. The total fatigue damage is equal to the sum<br />
of local (VIV) and global fatigue damage on each member.<br />
Step-by-step determination of both local and global fatigue damage<br />
is discussed further in Section D.6 of Appendix D. Application of<br />
the procedurecould indicatethat the fatigue life is expended after<br />
relatively small number of oscillations, requiring corrective<br />
measures to be taken either in the design process or during<br />
fabrication (devices,spoilers, etc.).<br />
8-5
8.4 METHODS OF MINIMIZING VORTEX SHEDDING OSCILLATIONS<br />
Because the environmental factors that cause vortex-induced<br />
oscillations (wave, current and wind) cannot be controlled,<br />
minimizing the oscillations depends primarily on the physical<br />
characteristicsof the structure.<br />
There are several ways to solve the problem of vortex-induced<br />
oscillations:<br />
●<br />
Controlof structuraldesign (length,diameter, end fixity)to<br />
obtainmember naturalperiodsto avoid the critical velocity.<br />
●<br />
Control of structuraldesign to have sufficiently high values<br />
of effective mass and inherent damping to avoid the critical<br />
velocity.<br />
●<br />
Altering the pattern of the approachingflowto modify vortex<br />
shedding frequency.<br />
Further discussion on this subject is presented in Section D.8 of<br />
Appendix D.<br />
8.5<br />
RECOMMENDATIONS<br />
Fatigue damage due to vortex shedding is best prevented during the<br />
design of the structure by sizing the members (length-to-length<br />
ratio, rigidity, damping, etc.) to ensure that critical velocity<br />
values are avoided. If geometric, design schedule or-economic<br />
constraints preclude resizing of members susceptible to VIV, the<br />
total fatigue damage due to local (VIV) and global response should<br />
be computed and the integrity of those members verified. If a<br />
limited number of members are found to be susceptible to fatigue<br />
failure, the flow around such members may be modified through the<br />
use of devices and spoilers.<br />
8-6
Verification of a member’s structural integrity due to VIV fatigue<br />
is difficult due to the-interactivenature of the vortices shed and<br />
the vibration of the member. State-of-the-artprocedures developed<br />
to determine the responseamplitudesof amember incorporateseveral<br />
approximations. It is reconunendedthat some of the more important<br />
of these approximationsare carefully considered before starting a<br />
VIV analyses:<br />
●<br />
Experimental data used to correlate parameters in the<br />
development of empirical procedures are limited. Published<br />
data is not available for in-line VIV in uniform oscillatory<br />
flOw.<br />
●<br />
Accuratedeterminationof structuraldamping ratios in air and<br />
inwater isdifficult. Thedamping ratios directly affect the<br />
stability parameter and may contribute to either<br />
underestimationor overestimationof the vibration amplitudes<br />
and stresses.<br />
●<br />
Tubulars extendingover multiple supportsneed to reevaluated<br />
by considering support sleeve tolerances and spanwise<br />
correlation of varying lengths and fixity prior to the<br />
determinationof natural frequencies.<br />
8-7<br />
/ y_.
Ra < S REGIME OF UNSE?ARATE~ FLOW<br />
5 TO 15 c Ro < U A FIXED ?AlfI OF F-<br />
VOfiTICES IN WAKE<br />
4G
9.<br />
FATIGUE AVOIDANCE STRATEGY<br />
Most marine structures are designed and analyzed to resist extreme<br />
loadings. Some structures,includingoffshore structures and ships<br />
with special features,are also checked for fatigue. This approach<br />
may be valid for structures in environments not susceptible to<br />
fatigue loadings. A good overall design of marine structures<br />
susceptible to fatigue loading (largeships and tankers, stationary<br />
fixed and floatingstructures,etc.) can be achievedwhen fatigue is<br />
given an equal emphasis to stability, strength and other<br />
considerations during design, long before steel is ordered.<br />
Fatigue design should be both an integral part of an overall design<br />
effort and a part of a strategy covering the entire design life of<br />
the structure. Thus the design, fabrication, inspection and<br />
operationalmaintenanceshould be treated as interactiveparameters<br />
that affect fatigue avoidance strategy.<br />
While most offshore structuressusceptibleto fatigue were properly<br />
analyzed and designed to prevent fatigue failures, ship-shaped<br />
vessels were seldom analyzed and designed for fatigue. The use of<br />
high strength steel in recently constructed vessels proved that an<br />
indirect fatigue design (i.e. member sizing, detailing) is not<br />
sufficient to prevent fatigue failures. As a result, large number<br />
of vessels constructed by reputable firms now incorporate detailed<br />
finite element analysis and design to prevent fatigue failures.<br />
9*1<br />
REVIEW OF FACTORS CONTRIBUTINGTO FAILURE<br />
Mobile vessels and stationary structures differ not only in their<br />
general configuration but also in the nature of applied<br />
environmental loading. A stationary structure’s site-specific<br />
environment usually determines the stress ranges and the number of<br />
stress cycles, and is a major variable affecting fatigue life. The<br />
next most importantvariables are the parameters affecting design<br />
and fabrication quality. While maintenance may not be important<br />
early in design life, it assumes a major role as the structure<br />
9-1<br />
j:) >~
ages. The designer has no control over the environment, but other<br />
factors can be addressed to enhance fatigue quality.<br />
The factors that affect fatigue quality can be reviewed in four<br />
groups. It appears reasonable to assume that each of these four<br />
groups contributes equally to fatigue failure:<br />
●<br />
●<br />
●<br />
●<br />
Design<br />
Fabrication<br />
Maintenance<br />
Operational Loads<br />
The fatigue life of a vessel is similarly affected by the activities<br />
undertakenduringdesign,fabrication,maintenancework and severity<br />
of operational loads. Skaar (Reference9.1) reports that a survey<br />
to assess the approximate importance of design, fabrication,<br />
maintenance and operations indicated that each contributes about<br />
equally to overall quality.<br />
9.2<br />
BASIC FATIGUE AVOIDANCE STRATEGIES<br />
9.2.1 Basic Premises<br />
Review of fatigue failures shows that while relatively few failures<br />
threaten structural integrity, repairs are costly and the cost of<br />
continuous inspection and maintenance is appreciable. A survey of<br />
design configurations and structural details shows that designers<br />
who have access to operational feedback on inspection, repair and<br />
maintenance, generally develop more reliable designs. To ensure a<br />
functional,high-qualitystructure(i.e.,with structural integrity)<br />
that is cost-effective, both capital expenditures (CAPEX) and<br />
operating expenditures (OPEX) should be addressed simultaneously.<br />
The review of marine structures indicate several design<br />
philosophies:<br />
9-2
●<br />
An indirect fatigue design where the design for extreme<br />
Ioadingandexperience-based detailingare intendedto provide<br />
ample fatigue resistance. This approach may be valid for<br />
structures subjected to negligible cyclic loadings.<br />
● ✍<br />
Simplified allowable stress methods based on in-servicedata<br />
and valid theoreticaldevelopments. This approach isvalid as<br />
a design tool to size structure components.<br />
●<br />
Comprehensive fatigue analyses and design methods with<br />
appropriatefatigue strength and stress history models. This<br />
approach, including finite element analyses to accurately<br />
determine the stress distributions, should be used in the<br />
design of all structures susceptible to fatigue failure.<br />
●<br />
Comprehensivefatigue analyses and design methods, taking the<br />
lifetime inspection and maintenance strategies into account.<br />
This is the valid approach to implement a cost-effective<br />
fatigue avoidance strategy.<br />
Design, inspection and maintenance are thus logically treated as<br />
interdependent parts of an overall process contributing to the<br />
quality of a structure.<br />
The other basic premises affecting fatigue avoidance strategies can<br />
be summarized as follows:<br />
●<br />
The fatigue life is usually taken as twice the design life.<br />
The target fatigue lives can be chosen tobe about fiveto ten<br />
times the design life with very little increase in steel.<br />
The additional expenditures caused by the slight increase in<br />
steel cost can be offset many times over by savings in<br />
operating expendituresassociatedwith inspection,repair and<br />
maintenance.<br />
●<br />
Service experience is of utmost importance in the design of<br />
marine structures. The designer should have an access to<br />
9-3
failure data on various structures, including continuous<br />
system stiffening details (i.e., orthotropically stiffened<br />
hul1plate).<br />
●<br />
Typically, stiffening detail failures cause serviceability<br />
problems, affecting the extent of a structure’s repair work<br />
and cost. Unrepaired, they may cause buckling, flooding and<br />
progressive collapse, thereby, resulting in the pollution of<br />
the environment and the loss of structural integrity.<br />
●<br />
Typical tubular interface failures of stationary structures<br />
can cause substantial degradation in structural integrity.<br />
Repairs on location, especially underwater, are extremely<br />
costly and are not always entirely successful.<br />
9.2.2 FatictueAvoidance Strategies<br />
Fatigue avoidancestrategiesfor ships and tankers are both similar<br />
and dissimilar to those for fixed and floating stationary<br />
structures. The primary components of continuous systems (ship<br />
longitudinal girder, semisubmersiblecolumn, etc.) are designed to<br />
provide ample strength, and the redundant load paths provided by<br />
multiple stiffenersmake fatigue more a serviceability problem. A<br />
discrete system such as a fixed platform may have redundancy to<br />
prevent major degradation of the structure, yet redistribution of<br />
load paths will accelerate crack growth in adjacent areas and can<br />
cause failures in these areas. To prevent additional failures,<br />
repair work should not be postponed beyond a reasonable period.<br />
The basic fatigue avoidance strategies are best addressed as the<br />
factors that affect design and maintenance:<br />
9-4
“!M91!<br />
●<br />
Global Configurations<br />
A design strategy that provides a global configuration with<br />
redundancy and minimizes both the applied loads and the<br />
response will enhance structure fatigue life and reduce<br />
maintenance costs.<br />
Both continuous system and discrete system global<br />
configurationscan be optimizedto various degrees to minimize<br />
the effect of applied loads and the response of the structure<br />
to these appliedloads. The dynamic response of the structure<br />
can contributeto substantialcyclic stress (i.e. both global<br />
and local dynamics, including vortex induced vibrations) and<br />
should be minimized.<br />
•~<br />
Joint/Weld Details<br />
The structuraljoint/welddetails should be developed basedon<br />
operatingexperience,analyticalstudiesand assessmentof the<br />
impact of actual fabricationyard work to minimize the stress<br />
concentrations,adversefabricationeffects and stresslevels.<br />
The joint/weld details should be designed to prevent large<br />
stress concentrations. Review of typical joint/detail<br />
failures and analytical parametric studies should be used to<br />
identify both “desirable” and “undesirable”details. Review<br />
of some of the published data on structural detail’failures<br />
(References9.2, 9.3, 4.2 and 4.3) also illustrate that such<br />
fatigue failures can be significantly decreased by avoiding<br />
magnification of stress patterns on a structure detail.<br />
Jordan and Cochran (Reference9.2) surveyed 3,307 failures in<br />
over 50 ships and presented their findings by grouping the<br />
structural details into 12 families The review of details<br />
within each family (twelvefamilies: beam brackets, tripping<br />
brackets, non-tight collars, t. ght collars, gunwale<br />
9-5
connections, knife edge crossings, miscellaneous cutouts,<br />
clearance cuts, deck cutouts, stanchionends, stiffenerends,<br />
and panel stiffener ends) should provide an invaluable<br />
operationalfeedbackto the designer in understandingrelative<br />
susceptibilityof different details to fatigue failure.<br />
●<br />
Material and Fabrication<br />
The material selected, procedures specified and fabrication<br />
specificationsissuedshouldbe compatiblewith each other and<br />
meet the requirements of the intended function of the<br />
structure.<br />
The design effort should ensure selection of material with<br />
chemical composition and material properties applicable for<br />
the structure’s intended function. Welding material and<br />
procedures should be compatible with the structural material<br />
selected. Overall fabrication specifications, covering<br />
fabrication tolerances, repair procedures, etc., should be<br />
developed to meet the target objectives. Specifications<br />
should reflect a balance between cost and fit-for-purpose<br />
approach to quality.<br />
Maintenance<br />
Stationary structures may require a higher degree of design<br />
conservatism than mobile structures to minimize the cost of<br />
maintenance, inspection and repair. Maintenance and inspection<br />
programs should be developed during design to reflect both design<br />
conservatism and functionalityof the structure and its components.<br />
Maintenance,<br />
parameters.<br />
differs from<br />
inspection and repair are interactive in-service<br />
The maintenance and inspection of continuous systems<br />
discrete systems largely in degree of accessibility.<br />
Most continuous systems (such as interiors of hulls, columns and<br />
pontoons)can be routinely inspectedandmaintained. Such units can<br />
be brought to shipyards for scheduled or unscheduled repairs.<br />
9-6
Fatigue avoidance strategy for mobile vessels should consider both<br />
the consequence of limited degradation due to fatigue failure and<br />
the relative ease of routine maintenance and scheduled repairs.<br />
Most discrete systems, such as offshore platforms, are stationary<br />
and their components are generally not accessible for internal<br />
inspection. Thus, inspectionis carried out externally, both above<br />
and below water. Any repair work undertaken is costly and may be<br />
only partially successful. Were regulations impose comprehensive<br />
inspection and maintenance programs, such as in the North Sea, a<br />
fatigue design philosophyaddressingthe inspection and maintenance<br />
issues also facilitates certification of design. Typically,<br />
redundancy and consequence of failure dictate the inspection<br />
intervals. Those areas known to be susceptible to fatigue failure<br />
will require more frequent inspection intervals. Similarly,<br />
inspectionresults should be the basis for altering the recommended<br />
inspection schedule as necessary.<br />
Analysis<br />
Analytical assumptions and the methodology implemented for fatigue<br />
life computations have dramatic effects. The choice of fatigue<br />
analyses appropriate for a specific project depends on the<br />
information available, research gaps, and sensitivity of structure<br />
to fatigue failure. Because fatigue analysis approach is not truly<br />
an avoidance strategy, it is discussed separately in Section 9.4.<br />
9.3<br />
FATIGUE STRENGTH IMPROVEMENTSTRATEGIES<br />
Fatigue strength improvement and fatigue avoidance strategies<br />
benefit from application of an appropriate design philosophy that<br />
allows development of structure and component integrity, and that<br />
facilitatesqualityof construction. The specificmethods discussed<br />
in the section are remedial measures for fatigue strength<br />
improvement.<br />
9-7
9.3.1 Fabrication Effects<br />
The fatigue strength of welded joints/details is lower than the<br />
parent material due a wide range of fabrication effects. Some of<br />
the primary causes for the degradation of fatigue strength are due<br />
to:<br />
●<br />
Increase in peak stresses due to geometrical effects and<br />
discontinuities(stressamplification)andmismatchtolerances<br />
(bending stress) introduced.<br />
●<br />
Residual stresses introduced<br />
excessive heat input, etc.<br />
due to welding, forced fit,<br />
●<br />
Defects introduced in the weld<br />
edge of welds.<br />
material, and undercut at the<br />
Adverse fabrication effects are minimized by addressing<br />
during design and specification writing.<br />
Both<br />
the issues<br />
experience<br />
(operationaland design)and parametricstudiesal1ow developmentof<br />
“desirable” details to minimize the local increase of stresses.<br />
Fabrication specifications are prepared to optimize fabrication<br />
quality without excessive expenditures.<br />
9.3.2 Post-FabricationStrenqth Improvement<br />
Numerouspost-fabricationprocessescan partiallyor totally counter<br />
the fabrication effects that contribute to degradation of fatigue<br />
strength. However, post-fabrication processes may be costly and<br />
should not be incorporated in the design process routinely.<br />
The developmentof fatigue cracks depends largelyon the geometryof<br />
the joint detail and often developat the weld toe. Any mismatchof<br />
parent plates will facilitate propagation of the crack through the<br />
weld until a failure across the throat is observed. Deposition of<br />
extra weld metal in the throat area to decrease the shear stress can<br />
9-8
improve the fatigue strength. The methods available to improve<br />
fatigue strength can be-grouped into two:<br />
●<br />
●<br />
Modification of weld toe profile<br />
Modification of residual stress distribution<br />
Some of the methods in each category are identified on Figure 9-1<br />
and discussed in this section.<br />
Modification of Weld Profile<br />
Both contour grinding of the weld profile and the local grinding of<br />
the weld toe area are recommendedto improve fatigue strength. The<br />
two key objectives in the modification of weld toe profile are:<br />
●<br />
●<br />
Remove defects at the weld toe.<br />
Develop a smooth transition between weld material and parent<br />
plate.<br />
By applying either local grinding or remelting techniques to remove<br />
defects and discontinuities, the fatigue life is increased as a<br />
function of time required for crack initiation. Some applicable<br />
methods are as follows:<br />
● ✍<br />
Grindinq<br />
Full-profile burr grinding, toe burr grinding or localized<br />
disc grinding can be carried out. Considering the time<br />
required for grinding, local-weldtoe grinding has become one<br />
of the most frequently used grinding methods. Careful and<br />
controlledlocal grindingof the weld toe improvesthe fatigue<br />
strength of a specimen in air by at least 30%, equivalent to<br />
an increase in fatigue life by a factor greater than 2.<br />
However, to obtain such a benefit the grinding must extend<br />
about 0.04 inch (1 nun)beneath the plate surface. Typical<br />
defects and correctivemeasures are shown on Figure 9-2.<br />
9-9
●<br />
Controlled Erosion<br />
An alternate weld toe modification technique uses a highcontaining<br />
grit. Under carefully<br />
pressure water jet<br />
controlled conditions the weld toe area can be eroded as<br />
though itwere beingground. Work carried outon filletwelds<br />
with abrasive water jetting (AWJ) by Maddox and Padilla<br />
(Reference9.4) andKing (Reference9.5) indicatethat fatigue<br />
life improvement due to AWJ erosion and toe grinding are<br />
comparable. The S-N curve improvements obtained due to weld<br />
toe abrasivewater jet erosion are illustratedon Figure 9-3.<br />
This approach does not require heat input and can be carried<br />
out quickly, offering an advantage over alternative methods.<br />
●<br />
Remeltinq Techniques<br />
Remeltingweld material to a shallow depth along the weld toe<br />
results in removal of inclusions and helps achieve a smooth<br />
transitionbetweentheweld and the platematerial. Tungsteninert-gas<br />
(TIG) and plasma welding are not practical<br />
techniquesfor routine use, but TIGandplasma dressing can be<br />
used to improve the fatigue strength of selective areas.<br />
TIG welding is based on astringer bead process<br />
is performed on welds made by other processes<br />
region is melted to a shallow depth without<br />
TIG dressing<br />
where the toe<br />
the use of a<br />
filler material. Slag particles in the remelted zone are<br />
brought to the surface, 1caving the weld toe area practical1y<br />
defect free. Ahigh heat inputshould be maintainedto obtain<br />
a good profile and a low hardness. A low hardness in the<br />
heat-affectedzone (HAZ)may be also achieved by a second TIG<br />
pass.<br />
Plasma dressing requires remelting the weld toe using the<br />
plasma arc welding technique. It is very similar to TIG<br />
dressing, but plasma dressing uses a wider weld pool and<br />
higher heat input. This technique is relatively insensitive<br />
9-1o
to the electrode position, so the strength improvements are<br />
better than the improvementsobtained from TIG dressing.<br />
Although overall weld profiling is considered desirable for<br />
tubular intersections, rules and recommendations other than<br />
API do not allow improvement in fatigue strength of a joint<br />
unless weld profiling is accompanied by weld toe grinding.<br />
Tliefatigue strength increaseof welded joints dueto weld toe<br />
grinding in air is considered equally applicable to<br />
catholicallyprotectedwelded joints in seawater. However, in<br />
the absence of cathodic protection, a corrosive environment<br />
helps to initiate fatigue cracks. Thus, without cathodic<br />
protection, fatigue strength improvement due to weld toe<br />
grinding cannot be justified.<br />
The fatigue strength increaseinwelded joints dueto weld toe<br />
grinding is basedon simple plate specimenstested in air and<br />
in seawater (with and without cathodicprotection). However,<br />
extensionof welded plate specimentest data to tubular joints<br />
may not be correct. Hork carried out by Wylde et al<br />
(Reference 9.6) indicates that additional research is<br />
necessary because:<br />
1) The corrosive effect of seawater appears to be greater<br />
on tubular joints than on flat plates.<br />
2) Cathodic protection appears to be less effective on<br />
tubular joints than on flat plates.<br />
Modification of Residual Stress Distribution<br />
A wide range of residual stress techniques are available to<br />
redistribute the fabrication stresses at a welded joint. If large<br />
residual tensile stresses are present at a welded joint, the applied<br />
stress cycle near the weld toe can remain wholly tensile. Thus,<br />
9-11
after<br />
a given number of stress cycles, the stress range to cause<br />
failure is practicallyconstant for a wide range of mean stresses.<br />
The undesirable tensile residual stresses at the weld can be<br />
modified by the following methods to set up desirable compressive<br />
stresses at the weld toe:<br />
●<br />
Stress Relief<br />
Various fatigue tests on simpleplate specimens indicate that<br />
an improvedfatigue strength can be obtained by stress relief<br />
due to post-weld heat treatment (PWHT). However, PIate and<br />
stiffening elements of continuous systems rarely require<br />
stressrelief. Thick tubularjointswith residual stressesas<br />
a result of fabrication work can often benefit from stress<br />
relief. Yet, it is not clear that a complex joint with builtin<br />
constraints can be effectively stress relieved. It is<br />
1ikely that substantial residual strains and stresses wil1<br />
remain at a joint assembly after PWHT.<br />
Localized stress relief may be very beneficial in an<br />
embrittled heat-affected zone (HAZ). Typically, high<br />
localized heat input in a HAZ alters the material properties<br />
and causes reduced fatigue life due to unstable fracture. A<br />
PWHT carriedout to improvetoughnessof the HAZmay partially<br />
restorethe fatiguestrengthof welded joints, as the residual<br />
stresses have an influence in the development of fatigue<br />
cracks. Previous investigations on this subject (Reference<br />
1.8) document influence of PWHT on fatigue.<br />
●<br />
Compressive Overstressinq<br />
Compressiveoverstressingis a technique in which compressive<br />
residual stresses are introduced at the weld toe.<br />
Experimental results and analytical work demonstrate<br />
effectivenessof prior overstressing,but the procedure to be<br />
implementeddoes not appear to be practical for most marine<br />
9-12
structures. A comprehensive discussion of strength<br />
improvementtechniquesby Booth (Reference9.7) is recotmnended<br />
for further review of compressive overstressing.<br />
Peening is a cold working process intended to produce surface<br />
deformations to develop residual compressive stresses. When<br />
impact loading on the material surface would otherwise cause<br />
the surface layer to expand laterally, the layer underneath<br />
prevents such surface layer expansion, creating the<br />
compressiveresidualstressesat the surface. Typical peening<br />
methods are hamer peening, shot peening and needle peening.<br />
Further discussion on peening techniques and their relative<br />
benefits is provided by Maddox (Reference 9.8).<br />
9.3.3<br />
Comt)arisonof StrenctthImrirovementStratecties<br />
Strength improvement techniques are time consuming and costly and<br />
they should be applied selectively. Comparison of different<br />
techniques allows assessment of their effectiveness and cost. The<br />
recommended strength improvement strategy depends on the<br />
characteristics of the structure (global and local) and the<br />
preference for one technique over others based on effectiveness,<br />
cost and fabricationyard characteristics.<br />
Some of the more important comparisons of various approaches<br />
available to improve fatigue strength of weld details subjected to<br />
a wide range of stresses are as follows:<br />
●<br />
Full profile burr grinding is preferable to toe burr grinding<br />
only, or disc-grinding only, because it results in higher<br />
fatigue strength even at a substantial cost penalty.<br />
Disc grinding requires the least time and cost. However, it<br />
produces score marks perpendicular to the principal stress<br />
direction, making this technique less effective than others.<br />
9-13
A second pass with polishingdisc is considered advisable. A<br />
complete chapter on weld toe grinding by Woodley (Reference<br />
9.9) provides a detailed discussion on grinding techniques.<br />
●<br />
Using a high-pressure abrasive water jet (AWJ) process for<br />
controllederosion of the weld toe area can be as effectiveas<br />
grinding. Its simplicity, speed and non-utilization of heat<br />
make controlled erosion very promising. Work carried out by<br />
King (Reference9.5) indicatethat AWJ process is suitable for<br />
a range of material removal applications, includingweld toe<br />
dressing, gouging and weld edge preparation.<br />
●<br />
A wider weld pool makes plasma dressing less sensitive to the<br />
positioning of the electrode relative to the weld toe,<br />
compared with TIG dressing. Therefore, the fatigue strength<br />
improvementobtained from plasma dressing is generally better<br />
than that obtained from TIG dressing.<br />
Both methods are suitable for automation and cost-effective<br />
application.<br />
●<br />
Review of grinding, remeltingand peening techniques indicate<br />
substantial scatter of fatigue strength improvements.<br />
Typically the best fatigue strength improvementsare achieved<br />
byTIG dressing and hannnerpeening. Toe disc grinding is the<br />
least effective technique. Figure 9-4, obtained from<br />
Reference 9.7, provides a good comparison of various fatigue<br />
strength improvementtechniques.<br />
9.4 FATIGUE ANALYSIS STRATEGIES<br />
9.4.1 Review of Uncertainties.Gaps and Research Needs<br />
There are many uncertainties in a fatigue analysis, carried out to<br />
determine the fatigue lives of marine structure components. To<br />
ensure validity of analysis the first objective is to accurately<br />
predict the stress-historyfor the lifetime of the structure. The<br />
9-14
second objective is to accurately evaluate the fatigue strength of<br />
the structure components and to calculate the cumulative fatigue<br />
damage basedon stress-historyand fatigue strength. While some of<br />
the uncertaintiesoccur in nature,others are caused by shortcomings<br />
in simulating the actual behavior.<br />
Uncertainties in Predicting Stress Histor.v<br />
It is necessaryto model the actual structure as closely as possible<br />
to determine the applied loads and the response of the structure to<br />
these applied loads. Since marine structures are typically<br />
indeterminate structures, stresses are strongly dependent on the<br />
structuralconfiguration,necessitatingcareful simulationof actual<br />
member and joint behavior.<br />
a) Hydrodynamic Loads Model<br />
The ship structureloads model allows the use of strip methods<br />
or 3-D flw solutions to determine the wave loads. The<br />
accuracyof the wave loaddeterminationdepends on the ability<br />
to accurately define the wave force coefficients, marine<br />
growth, wave steepness, hydrostatic effects and hydrodynamic<br />
effects.<br />
The loads on a stationary semisubmersible or fixed platform<br />
are typically determined from Morison’s equation. Fixed<br />
platform loads are largely affected by the accuracy of wave<br />
inertia and drag force coefficients, wave steepness, marine<br />
growth and the shieldingeffectof componentmembers.’ The use<br />
of a stick model is valid for a fixed platform, the use of a<br />
stick model for a structure made up of large members will<br />
result in inaccurate loads.<br />
Becauselargememberswill disturb the flow, leading to highly<br />
frequency dependent diffraction, a three-dimensional<br />
diffraction theory is often used to determine the wave force<br />
componentsto directly accountfor the effect of one member on<br />
9-15
others. Extensive analytical and experimental work provides<br />
validation of techniques used to generate the loads.<br />
Fcrstandard vesselswith aforward speed, stripmethods often<br />
provide the desirable accuracy. Although the diffraction<br />
methods are still considered largely a research tool by many,<br />
they are now used as an analyses and design tool by others.<br />
Limited amount of available data on wave-induced and dynamic<br />
impact (i.e. slawaning)loading on vessels and the vessel<br />
responsedo not facilitatecalibrationof analysesmodels. It<br />
is necessary to obtain sufficient data for various vessel<br />
types for an extended period. Boylston and Stambaugh<br />
(Reference 9.10) recommended program to obtain loading<br />
computer records, based on vessel strains for at least three<br />
vessel types over a five-year period, should provide<br />
sufficient data on probabilistic loadings and the vessel<br />
response.<br />
b)<br />
Mass. Motions and Stiffness Models<br />
There are few uncertainties in developing an accurate mass<br />
model. The motionsmodel, however, is largely affected by the<br />
assumptions made to define the motions and stiffness models<br />
and the analyses techniques chosen. The uncertainties built<br />
into these models that allow the definition of nominal<br />
stresses are:<br />
linearizationof drag term<br />
definitionof joint releases,complexityof joint, joint<br />
flexibility etc.<br />
definition of strongbacks and global versus local<br />
distribution of loads<br />
added mass<br />
appurtenancesmodelling<br />
structuraldamping (for bottom-supported structures)<br />
foundationmatrix (for bottom-supportedstructures)<br />
9-16
elative slippage--betweenjacket legs and piles.<br />
Additionaluncertaintiesintroduceddue to assumptionsmadeon<br />
analyses techniques, are:<br />
applicationof regular or random waves<br />
applicationoftime-domainor frequencydomain solutions<br />
use of deterministic versus spectral analyses<br />
While some of the uncertainties relate to analytical<br />
simulation of actual conditions, others reflect the<br />
uncertainties in both the nature and in simulation. Most<br />
analysis and modeling uncertaintiescan be minimized, and the<br />
current state-of-knowledge and tools available facilitate<br />
obtaining accurate nominal stress distributions.<br />
Since the structuredynamic responses (bothglobal and local,<br />
including vortex induced vibrations) contribute substantial<br />
cyclic stresses, it is extremely important to minimize the<br />
uncertainties in simulating structure responses.<br />
c)<br />
Hot Snot Stresses<br />
Peak stresses can be reasonably well defined by the use of<br />
physical models and finite element analyses. However, for<br />
most analysis and design work the time and cost constraints<br />
necessitate the use of empirical formulations to obtain the<br />
SCFS and define the hot-spot stresses.<br />
Al1 empirical formulations have application 1imits and the<br />
accuracy of the SCFS computed depend on several variables.<br />
More finiteelementwork is requiredto define the interaction<br />
of parameters for a wide range of joint geometries to upgrade<br />
existing empirical formulations.<br />
9-17
d) Stress SDectrum<br />
Hot-spot stresses combined with the long-term effects of the<br />
environment allow development of the stress spectrum.<br />
Randomnessof ocean environmentmakes both the short and longterm<br />
prediction of sea states quite difficult. The<br />
uncertainties of nature that influence the life-time stress<br />
history of a stationary structure are:<br />
- Use of full scatter diagram ofHs andT<br />
- Variations ofT<br />
- Percentage of occurrence estimates<br />
- Wave directionality<br />
- Interactionof wave and current<br />
For some site-specific stationary structures, a good existing<br />
databasemay allow comprehensivehindcastingstudies to predict both<br />
short- and long-term environment with a reasonable certainty. A<br />
reliability-based full probabilistic fatigue analysis al1ows<br />
selection of the degree of reliability that affects the fatigue<br />
life, such as the environmental loading, size and distribution of<br />
defects, fatigue strength, etc. However, even commonly used<br />
spectralfatigueanalyses,which isdeterministic, (i.e. application<br />
of only probabilisticenvironmentalconditions),the desirablelevel<br />
of uncertainty for the environment can be chosen to be compatible<br />
with the other factors that affect the computed fatigue life.<br />
For oceangoing ships which move through various site-specific<br />
environmentsin a singleroute, the stress history is very difficult<br />
to define. A full probabilisticreliabilityanalysis, orthe useof<br />
conservativeupper bound conditions,is necessary to account for the<br />
many different routes over the the uncertainties regarding the use<br />
of very different routes over the life of the vessel as well as<br />
route changes due to extreme environmental conditions.<br />
9-18
h!!!u<br />
Fatiguestrength is not analyzedbut determinedfrom laboratorytest<br />
specimens. The experimental work that allows the definition of<br />
fatigue strength and the S-N curves require substantial further<br />
work. Some of the basic variables contributing to the uncertainty<br />
fatigue strength include the effects of:<br />
Geometry (weld profile, toe discontinuity, etc.)<br />
Defect type, size and location<br />
Definition of fatigue failure (Nl, N2) in S-N data<br />
Size on S-N data<br />
Assumption of a linear model and log normal distribution for<br />
N<br />
Environment (corrosion,cathodic protection, etc.)<br />
Load amplitude and sequence<br />
Fabrication residual stresses<br />
Post-fabricationprocedures to increase fatigue strength<br />
-.<br />
Due to large uncertainties in each of the items listed, the fatigue<br />
strength data show a very large scatter, requiring the use of<br />
somewhat conservative S-N curves. The available test data on high<br />
stress range-low cycle fatigue failure is limited. Thus, the S-N<br />
curves for the 1000 to 10,000 cycle range are less reliable than the<br />
high cycle ranges.<br />
While additional work is necessary to better define geometric<br />
variations, the recent research has shown that there are also some<br />
uncertainties regarding the:<br />
●<br />
●<br />
●<br />
Beneficialeffect of weld profilewithout weld toe grindingor<br />
remelting<br />
Assumption of catholically protected joints in sea water<br />
having the same fatigue strength in air<br />
Classification of joints based on geometry rather than load<br />
pattern<br />
9“19
Cumulativefatiguedamagecomputationshave been and still are based<br />
on Miner’s linear cumulativedamage rule. Alternative stress cycle<br />
(rainflow)countingmethods have allowed reduction of uncertainties<br />
for wide-band loading. Gurney’s rule provides an alternative to<br />
Miner’s rule. However, the most important research gap in the<br />
computation of fatigue damage is the sequence of loading. The wave<br />
1oading, which is of stochastic nature, have been simulated by<br />
Markow matrix (Reference9.11) to carry out fatigue test of plates<br />
under stochastic and constant amplitude loading (Reference 9.12).<br />
These initial tests indicate fatigue strength properties for<br />
constant amplitude and spectrum loading may be different. Unti1<br />
more research is carried out on loading sequence it should be<br />
presumed that a certain number of large amplitude stress cycles<br />
during the beginning of a structure’s life would be likely to<br />
accelerate the fatigue crack growth of most defects. A series of<br />
tests being carried out at Technical University of Denmark<br />
(Reference 9.13) should provide more definitive conclusions on<br />
fatigue life of welded joints subjected to spectrum loading under<br />
various corrosive conditions.<br />
9.4.2<br />
Recent Research Activities<br />
Extensive fatigue research activities were carried out in the<br />
1980s. A large percentage of these activitieswere carried out in<br />
Europe, addressing the parameters affecting fatigue life of<br />
joints/detailsin the extremeNorth Sea environment. Other research<br />
activities carried out in the United States and elsewhere indicate<br />
that the research activities are often complementary and generally<br />
avoid duplication of effort.<br />
The fatigue research activities are generally carried out in two or<br />
three phases over multiple years. While some research activities<br />
were completed, others will continue into early 1990s. These<br />
research activities may be grouped into following areas and the<br />
relevant activities are sunanarizedon Figure 9-5.<br />
9-20
●<br />
Stress concentrationfactors; includingcollating of existing<br />
data, calibration of SCF equations and development of<br />
parametric equations.<br />
●<br />
Fatigue analysis and design methods; includingfinite element<br />
analysis procedures and application of fatigue design rules.<br />
●<br />
Fatigue resistance; including simple plate S-N curves and<br />
complex details, S-N curves for stiffened joints and S-N<br />
curves for different materials.<br />
●<br />
Effect of various parameters on fatigue life; including the<br />
effect of cathodicprotectionin seawater,plate thicknessand<br />
weld profile effects.<br />
●<br />
Fatigue<br />
life improvementtechniques.<br />
●- Fatigue<br />
damage,<br />
life determination; including review of cumulative<br />
assessment of random loading and low cycle fatigue.<br />
9.4.3 Cost-EffectiveAnalysis Strategies<br />
Acost-effective analysesstrategy is relativelyeasy to develop for<br />
any marine structure. First, the structure configuration and the<br />
likely marine environment should be assessed to determine<br />
susceptibility of the structure to fatigue. Second, structure<br />
configuration and operational response characteristics should be<br />
assessed to determinethe desirable analyses techniques to generate<br />
the loads and to determine the response of the structure.”<br />
Although computer cost is an important variable in developing an<br />
analysis strategy, computer cost should be assessed in conjunction<br />
with engineering time and effort as well as the time available to<br />
completethe fatigueanalysisand design. Most important,design is<br />
an iterative process and structural changes will invariably occur<br />
during fatigue analysis. Thus, fatigue analysis should be treated<br />
9-21
as a parametric study intended to identify the fatigue-susceptible<br />
areas for improvement.<br />
Considering that small increases in steel used can appreciably<br />
increase fatigue lives, it is recommended that the target fatigue<br />
lives (at least for a screening effort) be taken as five to ten<br />
times the design life while most rules and reconunendationsspecify<br />
a factor of two between fatigue and design life. Then, changes<br />
introduced during design that has an impact on applied loads and<br />
stress distributions can be readily accommodated.<br />
9.5<br />
RECOMMENDATIONS<br />
Fatigue avoidance strategiesadopted and the design tools used have<br />
served as well. However, further efforts are necessary in carrying<br />
out more research, in developing further improvements in analyses<br />
and design, and in upgrading the rules and regulations to<br />
incorporate the research results.<br />
Recommendationspresented in Section 5 through8 provided the basis<br />
for further in-depth discussions in Section 9. Applicable<br />
references in each section are listed in Section 10. Some of the<br />
primary recommendations are listed as follows:<br />
●<br />
Although “allowable stress” methods may be used as a<br />
“screening process,” a detailed fatigue analysis is often<br />
necessary.<br />
●<br />
Assessment of various empirical equations indicate that the<br />
UEGequations yield conservativepredictionofSCFs for awide<br />
range of geometry. However, empirical equations provided by<br />
UEG, Efthymiou,Kuang and others should be reviewed for joint<br />
geometry and loading condition to allow selection of most<br />
appropriate equation.<br />
●<br />
The long-term wave<br />
models are quite<br />
environment definitions based on hindcast<br />
reliable. However, modeling parameters<br />
9-22
should be carefully reviewed and the model calibrated to<br />
ensure the reliability of data.<br />
● ✍<br />
The S-N curvesused indetennining fatiguedamage computations<br />
should be compatiblewith structural details investigated.<br />
●<br />
Considering the effect of size, weld profile and undercut on<br />
fatigue strengthand S-N curves, it may be prudent to reassess<br />
the hot spot stress range concept. The definition of hot spot<br />
stressrange can be modified to reflect the weld toe defects.<br />
●<br />
The use of Miner’s cumulativefatiguedamage rule with the S-N<br />
curves is appropriate. Further research, especial1y on the<br />
effects of stress sequence and counting of stress reversals,<br />
is considered necessary.<br />
9.5.1 Research Priorities<br />
Whether designinga supertankeror an offshoreplatform, significant<br />
fai1ure modes can be identified, environmental 1oads generated,<br />
structure response characteristics determined, and stress<br />
superpositionscompatiblewith the environmentand the failuremodes<br />
computed. Although strength statistics for these structures can be<br />
expressed in terms of means and variance, lack of sufficient<br />
statistical data on loads, stresses and strength prevent full<br />
probabilistic fatigue analyses. A development of a semiprobabilisticanalysisapproachapplicableto<br />
various structuresand<br />
that does not require a distribution shape is desirable.<br />
While a typical fatigue damage assessment is based on fatigue<br />
strength data yielding S-N curves, such an assessment can also be<br />
made based on fracture mechanics and crack growth laws. While the<br />
damage assessment is based on propagationof individual crack, work<br />
carried out by Morgan (Reference 9.14) has indicated possible<br />
interactionof multiple cracks. Thus, further work is necessary to<br />
obtain data on interaction of cracks as well as interaction of<br />
parameters affecting development of S-N curves.<br />
9-23<br />
..2I‘7
Additional areas requiring further research are sununarizedas<br />
follows:<br />
Parallel study of weld profile and weld toe defects.<br />
Analytical study of existing data for weld toe defect stress<br />
levels and through-thicknessstress levels.<br />
Identification of the type and magnitude of the errors<br />
introduced in laboratory work and development of appropriate<br />
means to normalize test data.<br />
Further assessment of empirical equations. Available test<br />
data should be further evaluated, incorporating necessary<br />
correction of data, and reliability and limitation of<br />
equations revised, as necessary.<br />
Carrying out of additional tests in both air and in ocean<br />
environment to fill the gaps in existing research.<br />
Development of NDE methods to quantify residual stresses<br />
introducedduring fabrication.<br />
Further study of long-term wave environment.<br />
Further assessment of stress sequence on fatigue life.<br />
9.5.2<br />
Rules and Regulations<br />
Existing rules, regulations and codes are adequate and generally<br />
conservative. However, differences exist between various rules,<br />
regulations and codes, including omissions and inconsistencies.<br />
Research data obtained in the 1980s was the basis for revisions<br />
introduced into the 4th Edition of Guidance Notes (1990). Similar<br />
effort has been initiated to revise API RP 2A. Some of the recent<br />
studies published (References 7.8 and 5.20) follow a deliberate<br />
format to facilitate extraction of data to upgrade existing rules<br />
9-24<br />
-m’<br />
..7 ,
and regulations. These and other study results should prove<br />
valuable in revision and upgrading of rules and regulations.<br />
9-25
GROUP<br />
Modification of<br />
Weld Profile<br />
Modification of Residual<br />
Stress Distribution<br />
METHOD<br />
a. Local and Contour<br />
Grinding<br />
a. Stress Relief<br />
b. Controlled Erosion<br />
c. Remelting Techniques<br />
- TIG Dressing<br />
- Plasma Dressing<br />
b. Compressive<br />
Overstressing<br />
- Local Compression<br />
- Spot Heating<br />
c. Peening<br />
- Shot Peening<br />
- Hammer Peening<br />
- Needle I%ening<br />
Figure 9-1 Typical Methods to Improve Fatigue Strength
UNDERCUT<br />
CWCK-LIKE<br />
DEFECT & .’i ~HyDROGEN<br />
,:- ,.<br />
-. ,?,<br />
i-,. .<br />
HAz<br />
clUCK<br />
~SION<br />
LINE<br />
~LL<br />
PROFIU<br />
TOE GRINDING<br />
IACK<br />
OF _ETm,,ON FUSION<br />
i<br />
!<br />
-t-<br />
O. S-l. Omm<br />
Figure g-2 Typical Weld Toe Defects and CorrectiveMeasures<br />
\<br />
● m weld toes erad<br />
by abrasve.wter<br />
jet<br />
x’<br />
●<br />
A$-welded<br />
“+%<br />
/ 8’-<br />
Figure 9-3 Fatigue Life ImprovementDue to Weld Toe Abrasive<br />
Water Jet Erosion<br />
(FroITIReference 9.4)
400 ‘,, . .<br />
350 - I I<br />
300r<br />
[ ~<br />
,05- I<br />
104 105<br />
\<br />
\\<br />
106<br />
Endurance, cycie$<br />
\ <<br />
\<br />
d 445<br />
“ - Siaot heated<br />
FuIIv bttfr ground<br />
S~Ot Deened<br />
OWflOaded at<br />
232N/mm2<br />
-“ i<br />
=A$+elded<br />
1Q7<br />
I<br />
...<br />
400<br />
I<br />
300 – TIG dressed (high tensile steel)<br />
m<br />
: 200 .<br />
2<br />
g 150–<br />
5<br />
L TOOdiw ground :<br />
! 100–<br />
Overlaadcdat 232N/mm2<br />
50<br />
5<br />
350-<br />
wEl-<br />
Id 1135<br />
2 345<br />
\<br />
\ \ ASw,~~<br />
I<br />
I<br />
234j<br />
106. 107<br />
Endurance,<br />
cycies<br />
Figure g-q Comparison of Fatigue<br />
(From<br />
Ref erer-tce<br />
Strength Improvement Techniques<br />
9.7)
SOMEOF THERELEVANT FATIGUERESEARCHPROJECTS<br />
TOPIC<br />
SUBJECT<br />
INVESTIGATOR<br />
COMPLETION &<br />
COMMENTS<br />
I<br />
STRESS CONCENTRATION<br />
Calibration of various SCF equations<br />
with Sillpletubularjoint test data<br />
Lloyd’s Registry<br />
1988<br />
STRESS CONCENTRATION<br />
Development of SCF equations for<br />
Internally stiffened complex<br />
tubular joints<br />
Lloyd’s Registry<br />
1990<br />
STRESS CONCENTRATION<br />
Development of SCF equations for<br />
multiplanar joint CHS and RHS”sections<br />
Univ. of Delft<br />
1992<br />
FATIGUE STRENGTH<br />
Assessment of S-N curve for lnternallystlffened<br />
tubular joints<br />
TMI/tiEL<br />
& Lloyd’s<br />
1990<br />
FATIGUE STRENGTH<br />
Development of S-N curve for slngleslded<br />
closure welds on tuhulars<br />
TW1/Glasgow<br />
I.lrtiversity<br />
1989<br />
FATIGUE STRENGTH<br />
Extend the range of hot spot stress<br />
approach to low cycle fatigue<br />
Unlverlty<br />
of Delft<br />
1989<br />
PARAMETERS AFFECTING<br />
FATIGUE LIFE<br />
Study of the chord plate thickness<br />
and other geometric parameters<br />
TWI<br />
1992<br />
PARAMETERS AFFECTING<br />
FATIGUE LIFE<br />
Study of the plate thickness and weld<br />
profile effects on fatigue llfe<br />
Florida Atlantlc<br />
University<br />
1990<br />
PARAMETERS AFFECTING<br />
FATIGUE LIFE<br />
Study of the weld profile effects and<br />
assessment of AtiS/APIrequirements<br />
TWI/EWI<br />
1989<br />
PARAMETERS AFFECTING<br />
FATIGUE LIFE<br />
Study of cathodic protection effect<br />
on tubular joint fatigue<br />
TWI/SSNTEF<br />
1991<br />
Figure 9-5 Summary of Relevant<br />
Page 1 of 2<br />
Research Activities<br />
(.jd
TOPIC<br />
SUBJECT<br />
INVESTIGATOR<br />
COMPLETION &<br />
COMMENTS<br />
PARAMETERSAFFECTING<br />
FATIGUE LIFE<br />
Study of the effect of seawater on<br />
thick welded long-llfe joints<br />
TWI<br />
1992<br />
PARAMETERS AFFECTING<br />
FATIGUE LIFE<br />
Application of fatigue life improvement<br />
1.<br />
techniques<br />
TWI<br />
FATIGLJEANALYSIS<br />
Correlation of simple p]ate S-y curves for<br />
complex details and F.E. analyses procedure!<br />
EW1/TWI<br />
Non-tubular details<br />
“only;through 1992<br />
FATIGUE ANALYSIS<br />
Development of SCF, S-N curves and<br />
materials data for cast steel pocles<br />
Wimpey<br />
1988<br />
FATIGUE ANALYSIS<br />
Reviewof cumulative damage of welded joints<br />
TWI<br />
?<br />
FATIGUE ANALYSIS<br />
Fa~igue of welded joints under random<br />
loading and Imp?lcatlonson the accuracy<br />
of Miner’s Rule<br />
TMI<br />
1990<br />
FATIGUE ANALYSIS<br />
Effect of loading sequence on<br />
fatigue life<br />
Technical Univ.<br />
of Denmark<br />
1991 ‘<br />
Figure 9-5 Summary of Relevant Research Activities<br />
Page’2of 2<br />
---c-
10. REFERENCES<br />
1.1 Fatigue Handbook - Offshore Steel <strong>Structure</strong>s, Edited by<br />
A. Almar-Naess, Tabin Publishers, Trondheim, Norway, 1985.<br />
1.2 Fatigue Analyses of Tankers, Technical Report No. RD-89020F,<br />
Research and Development Department, American Bureau of<br />
<strong>Ship</strong>ping, December 1989.<br />
1.3 Munse, W.H., Wilbur, T.W., Tellaliar, M.L., Nicoll, K., and<br />
Wilson, K., Fatigue Characterization of Fabricated <strong>Ship</strong><br />
Details for Design, <strong>Ship</strong> <strong>Structure</strong> ConnnitteeReport SSC-318,<br />
1983.<br />
1.4 Wirsching, P.H., “Probability Based Fatigue Design Criteria<br />
for Offshore<strong>Structure</strong>s,”Final ProjectReport onAPI PRAC81-<br />
15, University of Arizona, Tucsonj 1983.<br />
1.5 RecommendedPracticefor Planning,Designing and Constructing<br />
Fixed Offshore P1atform,API ReconmnendedPractice 2A (RP 2A),<br />
EighteenthEdition,American Petroleum Institute,Sept. 1989.<br />
1.6 Offshore Installations: Guidance on Design, Constructionand<br />
Certification, U.K. Department of Energy (UK DEn) Guidance<br />
Notes, Fourth Edition, June 1990.<br />
1.7 Fatigue Strength Analysis for Mobile Offshore Units,<br />
Classification Notes, Note No. 30.2, Det Norske Veritas,<br />
August 1984.<br />
1.8 Design of Tubular Joints for Offshore <strong>Structure</strong>s, Underwater<br />
EngineeringGroup, UEG Publication UR33, 1985.<br />
1o-1
3.1 Soyak, J. F., Caldwell, J. W., and Shoemaker,A.K., “Fatigueand<br />
FractureToughness Characterizationof SAW and SMAA537 Class<br />
I <strong>Ship</strong> Steel Weldments,”<strong>Ship</strong> <strong>Structure</strong> ConunitteeReportSSC-<br />
303, 1981.<br />
3.2 Pense, A.W., “Evaluationof FractureCriteria for <strong>Ship</strong> Steels<br />
and Weldments,”<strong>Ship</strong><strong>Structure</strong><strong>Committee</strong>ReportSSC-307, 1981.<br />
3.3 Williams, A.K., and Rinne, J.E., “FatigueAnalysis procedure<br />
of Steel Offshore <strong>Structure</strong>s”, Proceedings of Institute of<br />
Civil Engineers, Part 1, November 1976.<br />
3.4 Kuang, J.G., Potvin, A.B., Leick, R.D., and Kahrlich, J.L,<br />
“Stress Concentration in Tubular Joints,” Journal of Society<br />
of Petroleum Engineers,August 1977.<br />
3.5 Smedley, G.P., and Wordsworth, A.C., ‘Stress Concentration<br />
.-<br />
Factors of Unstiffened Tubular Joints,” European Offshore<br />
Steels Research Seminar, The Welding Institute, Cambridge,<br />
England, 1978.<br />
3.6 Worsworth, A.C., “Stress Concentration Factors at K and KT<br />
Tubular Joints,” Fatigue in Offshore Structural Steel, ICE,<br />
London, 1981.<br />
4.1 Rules for Buildingand ClassingSteel Vessels,American Bureau<br />
of <strong>Ship</strong>ping, 1988.<br />
4.2 Thayamballi,A.K., “FatigueScreening for Tankers,” Report RD<br />
90005, American Bureau of <strong>Ship</strong>ping, Research and Development<br />
Division, May 1990.<br />
10-2
4*3<br />
Munse, W.H., Wilbur, T.U., Tellalian, M.L., Nicoll, K., and<br />
Wilson, K., Fatigue Characterization of Fabricated <strong>Ship</strong><br />
Details for Design, <strong>Ship</strong> <strong>Structure</strong> ConanitteeReport SSC-318,<br />
1984.<br />
4.4<br />
Jordan, C.R., and Cochran, C.S., ‘In-Service Performance of<br />
StructuralDetails,”<strong>Ship</strong> <strong>Structure</strong>Conunittee Report SSC-272,<br />
1978.<br />
4.5<br />
Jordan,C.R., andCochran,C.S., “FurtherSurvey of In-Service<br />
Performance ofStructural Details,” <strong>Ship</strong> <strong>Structure</strong> <strong>Committee</strong><br />
Report SSC-294, 1980.<br />
4.6<br />
Chen, Y-K., Chiou, J-W, and Thayamballi, A.K, “Validation of<br />
Fatigue Life Prediction Using Containership Hatch-Corner<br />
Strain Measurements,” SNAME Transactions, Volume 94, 1986,<br />
pp. 255-282.<br />
4.7<br />
Luyties,W.H., and Geyer, J.F., “TheDevelopment of Allowable<br />
Fatigue Stresses in API RP 2A,” Nineteenth Annual Offshore<br />
TechnologyConference,OTC Paper No. 5555, Houston, TX, 1987.<br />
4.8<br />
Rules for Building and Classing Mobile Offshore Drilling<br />
Units, American Bureau of <strong>Ship</strong>ping, 1980.<br />
4*9<br />
Wirsching, P.H., “Digital Simulation of Fatigue Damage in<br />
Offshore <strong>Structure</strong>s,” Computational Method for Offshore<br />
<strong>Structure</strong>s, Editors: H. Armen and S.G. Stiansen, ASME, 1980.<br />
4.10<br />
Chen, Y.N., and Mavrakis, S.A.,“Closed-FormSpectral Fatigue<br />
Analysis for Compliant Offshore <strong>Structure</strong>s,” Journal of <strong>Ship</strong><br />
Research.<br />
4.11<br />
Wirsching, P.H., “Probability Based Fatigue Design Criteria<br />
15, University of Arizona, Tuscon, 1983.<br />
10-3<br />
for Offshore<strong>Structure</strong>s,nFinal ProjectReport onAPI PRAC81-<br />
-+?-
4.12 Daidola, J.C., and Basar, N.S., ‘probabilistic Structural<br />
Analysis of <strong>Ship</strong> Hull Longitudinal Stresses,” <strong>Ship</strong> <strong>Structure</strong><br />
ConunitteeReport SSC-301, 1981.<br />
4.13 Wells, A.A., “The Waning of Fitness-for-Purpose and the<br />
Concept of Defect Tolerance,” International Conference on<br />
Fitness for Purpose Validation of bleldedConstructions, The<br />
Welding Institute,Volume 1, Paper No. 33, London, 1981.<br />
4.14 Structural Welding Code, American Welding Society, AWS D1.1,<br />
1988.<br />
4.15 Specifications for the Design, Fabrication and Erection of<br />
Structural Steel for Buildings, American Institute of Steel<br />
Construction, 1989.<br />
4.16 Rules for Steel <strong>Ship</strong>s, Oil Productionand Storage Vessels,Det<br />
norske Veritas, Part 5, Chapter 9, 1986.<br />
4.17 Rules for Classificationof Fixed Offshore Installations,Det<br />
norske Veritas, January 1989.<br />
4.18 Rules and Regulationsfor the Classificationof Fixed Ofshore<br />
Installations, Lloyd’s Register, Part 4, Steel <strong>Structure</strong>s,<br />
July 1988.<br />
4.19 Gurney, T.R., “The Influence of Thickness of the Fatigue<br />
Strength of Welded Joints,” BOSS Conference, London, August<br />
1979, Paper No. 41.<br />
4.20 Lotsberg, I., and Anderssson, H., “Fatigue in Building Codes<br />
Background and Applications,H Chapter 11, Fatigue Handbook -<br />
Offshore Steel <strong>Structure</strong>s, Edited by A. Almar-Naess, Tapir,<br />
1985.<br />
10-4
4.21<br />
Weidler, J.B., and Karsan, D.I., ‘Design, Inspection and<br />
Redundancy Invest~nt Versus Risk for Pile-Founded Offshore<br />
<strong>Structure</strong>s,”ProceedingsOf The InternationalSymposiumOn The<br />
Role Of Design, Inspectionand RedundancyinMarine Structural<br />
Reliabi” ity, Williamsburg,VA, November 1983.<br />
4.22<br />
Capanog” u, C. “Design of Floating Offshore Platforms,”<br />
Proceed ngs Of The International Symposium On The Role Of<br />
Design, Inspection and Redundancy in Marine Structural<br />
Reliability, Paper No. 19, Williamsburg, VA, November 1983.<br />
4.23<br />
ISSC Design Philosophy<strong>Committee</strong> (1983),Design Philosophy of<br />
Marine<strong>Structure</strong>sReport,International<strong>Ship</strong>buildingProgress,<br />
Volume 30, No. 346, June 1984.<br />
4.24<br />
4.25<br />
SEALOAD,A Programfor Wave, Wind and Current Load Generation,<br />
Earl and Wright Developed SEADYN System Component, 1990.<br />
SHIPMOTION, A Program for <strong>Ship</strong> Motion Analysis, An American<br />
Bureauof<strong>Ship</strong>pingDevelopedABS/DAISY System Component, 1989.<br />
—<br />
4.26<br />
Mansour,A.E. andThayamballi,A., “Computer-AidedPreliminary<br />
<strong>Ship</strong> StructuralDesign,” <strong>Ship</strong> <strong>Structure</strong><strong>Committee</strong> ReportSSC-<br />
302, 1981.<br />
5.1<br />
Fukusawa, T., Fujino, M., Koyanagi, M. and Kawamura, T.,<br />
“Effectsof Axial Forceson Deck Stress in Case of Slammingof<br />
Large Bulk Carrier,” JSNAJ, Vol. 155, 1984.<br />
5.2<br />
Liapis, S. and Beck, R.F., “Seakeeping Computations Using<br />
Time-Domain Analysis,” 4th International Conference on<br />
Numerical <strong>Ship</strong> Hydrodynamics,Washington, 1985.<br />
5*3<br />
Papanikolaou, A., Zaraphonitis, G. and Perras, P., “On<br />
Computer-AidedSimulationsof Large Amplitude Roll Motions of<br />
<strong>Ship</strong>s in Waves and of Dynamic Stability,” IMAEM ’87, Varna,<br />
1987.<br />
10-5<br />
.-,—, :,<br />
/’ .,< 7’<br />
>—
5.4 Papanikolaou,A. and Zaraphonitis, G., “On An Improved Near<br />
Field Method For the Evaluationof Second-Order ForcesActing<br />
on 3D Bodies in Waves,” IMAEM ’87, Varna, 1987.<br />
5.5 Hooft, J.P., “Hydrodynamic Aspects of Semisubmersibles,”<br />
Doctoral Thesis From Delft Technical University, The<br />
Netherlands, 1972.<br />
5.6 Capanoglu,C.C., “Designof a CompliantTension LegPlatform-<br />
Naval Architectural and Structural Design Considerations,”<br />
Marine Technology, Vol. 16N0. 4, October 1979, pp. 343-353.<br />
5.7 Garrison,C.J., ‘Hydrodynamicsof LargeDisplacementFixedand<br />
Floating <strong>Structure</strong>s in Waves,” Report No. 80-102, December<br />
1980, C.J. Garrison Associates.<br />
5.8 Sircar, S., Rager, B.L., Praught, M.W. and Adams, C.J., ‘A<br />
Consistent Method for Motions, Strength and Fatigue Analysis<br />
of TLPs,” Proceedings of Seventh International Offshore<br />
Mechanics and Arctic Engineering Conference, OMAE, Houston,<br />
February 1988.<br />
5.9 Bishop, J.R., “Wave Force Investigations at the Second<br />
Christchurch Bay Tower,” NMIReport R177, 1984.<br />
5.10 Bishop, J.R., “Wave Force Data From the Second Christchurch<br />
BayTower,” Proceedingsof OffshoreTechnologyConference,OTC<br />
Paper No. 4953, Houston, 1985.<br />
5.11 Tickell, R.S. and Bishop, J.R., “Analyses of Wave and Wave<br />
Forces at the Second ChristchurchBay Tower,” Proceedingsof<br />
OffshoreMechanicsandArctic Engineering,OMAE, Dallas, 1985.<br />
5.12 Bea, R.G., Pawsey, S.F. and Litton, R.W., “Measured and<br />
Predicted Wave Forces on Offsfhore Platforms,” Twentieth<br />
Offshore Technology Conference, OTC 5787, Houston, TX, 1988.
5.13 Rodenbush, G., “Random Directional Wave Forces on Template<br />
Offshore Platforms,” Proceedings of the Eighteenth Offshore<br />
Technology Conference, OTC 5098, Houston, TX, 1986.<br />
5.14 Rodenbush, G. and Forristall, G.Z., “An Empirical Model for<br />
Random Directional Wave Kinematics Near the Free Surface,”<br />
Proceedings of the Eighteenth Offshore Technology Conference<br />
OTC 5097, Houston, TX, 1986.<br />
5.15 API PRAC PROJECT 83-22 Report, ‘Implementation of a<br />
Reliability-Based API RP 2A Format,” American Petroleum<br />
InstituteJanuary 1985.<br />
5.16 Kint, T.E., and Morrison, D.G., “Dynamic Design and Analysis<br />
Methodology for Deepwater Bottom-Founded <strong>Structure</strong>,”<br />
Proceedings of the Twenty-Second Offshore Technology<br />
Conference, OTC 6342, Houston, TX, 1990.<br />
5.17 Digre, K.A. Brasted, L.K., and Marshall, P.W., “TheDesignof<br />
the Bullwinkle Platform.” Proceedings of the Twenty-First<br />
Offshore Technology Conference, OTC 6050, Houston, TX, 1989.<br />
5.18 Larrabee, R.D., “Dynamics Analysis,” ASCE Continuing<br />
Education, Structural Reliability of Offshore Platorms,<br />
Houston, TX, 1989.<br />
5.19 Efthymiou, M. et al, “Stress Concentrations in T/Y and<br />
Gap/Overlap K-Joint,” The Forth International Conference on<br />
Behavior of Offshore <strong>Structure</strong>s, Amsterdam, The Netherlands,<br />
1985.<br />
5.20 Ma, S.Y.A. and Tebbet, I.E., “New Data on the Ultimate<br />
Strength of Tubular Welded K-Joints Under Moment Loads,”<br />
TwentiethAnnual OffshoreTechnologyConference,OTC PaperNo.<br />
5831, Houston, TX, May 1988.<br />
10-7
5.21 -”’ ” ”---”---” ‘“--”------”-<br />
tilDsZeln,M.B., “StressConcentrationin TubularK-Jointswith<br />
Diameter Ratio Equal to One, “PaperTS 10, Proceedingsof the<br />
Third International ECSC Offshore Conference on Steel in<br />
Marine <strong>Structure</strong>s (SIMS 87), Delft, the Netherlands, 1987.<br />
5.22<br />
Tolloczko, J.A., and Lalani,M., “The Implicationsof New Data<br />
in the Fatigue Life Assessment of Tubular Joints,” Twentieth<br />
Annual Offshore Technology Conference, OTC Paper No. 5662,<br />
Houston, TX, May 1988.<br />
5.23<br />
Lalani,M. etal, “ImprovedFatigue Life Estimationof Tubular<br />
Joints, “EighteenthAnnualOffshoreTechnologyConference,OTC<br />
Paper No. 5306, Houston, TX, 1986.<br />
5.24<br />
Gibstein, M.B., “Parametric Stress Analysis of T-Joints,”<br />
Paper No. 26, European Offshore Steels Research Seminar,<br />
Cambridge, November 1978.<br />
5.25<br />
Wordsworth,A.C., “Aspectsof Stress Concentration Factors at<br />
Tubular Joints,” Paper TS8, Third InternationalECSC Offshore<br />
Conferenceon Steel in Marine <strong>Structure</strong>s (SIMS87), Delft, The<br />
Netherlands, 1987.<br />
6.1<br />
Hoffman, D., and Walden, D.A., “EnvironmentalWave Data for<br />
Determining Hull Structural Loadings, A Report to <strong>Ship</strong><br />
<strong>Structure</strong>s Cotmnittee,SSC-268, 1977.<br />
6.2<br />
Bian Jiaxi and Yuan Yeli, “Meteo-OceanographicStudy of JHN<br />
Blocks 16/06 in South China Sea”, China Ocean Technology<br />
Company (COTC), Ltd., July 1988.<br />
6.3<br />
Meteo-Oceanographic Study of ACT Blocks 16/04-16/08 and<br />
PhillipsArea 15/11 in South China Sea. A Report Prepared by<br />
Snamprogetti for ACT Operators Group, March 1988.<br />
10-8
6.4 Study of Environmental Design Criteria For JHN’s Offshore<br />
China Lufeng Project, A Report to Earl and Wright, Shezhen<br />
China Ocan Technology Company (COTC), January 1991.<br />
6.5 N. Hogben and F.E. Lumb, “OceanWave Statistics,”Hjnistryof<br />
Technology, National Physical Laboratory, London: Her<br />
Majesty’s Stationary Office, 1967.<br />
6.6 M.K. Ochi and Y.S. Chin, “Wind Turbulent Spectra for Design<br />
Consideration of Offshore <strong>Structure</strong>s,” Offshore Technology<br />
Conference 1988, OTC 5736.<br />
6.7 Kinra, R., and Marshall, P., “Fatigue Analyses of Cognac<br />
Platform,’’Journalof PetroleumEnergy,SPE 8600, March 1980.<br />
6.8 Sarpkaya, T. and Isaacson, M., “Mechanics of Wave Forces on<br />
Offshore <strong>Structure</strong>s,”Van Nostrand Reinhold Co., 1981.<br />
6.9 Wirsching, P.H., “Digital Simulation of Fatigue Damage in<br />
Offshore <strong>Structure</strong>s,” Computational Method for Offshore<br />
<strong>Structure</strong>s,H. Armenand S.G. Stiansen,Editors,ASMB, 1980.<br />
6.10 Chen, Y.N., and Maurakis, S.A., “Close Form Spectral Fatigue<br />
Analysis of Compliant Offshore <strong>Structure</strong>s, “Journal of <strong>Ship</strong><br />
Research.<br />
6.11 Dalzell, J.F., Maniar, N.M., and Hsu, M.W., “Examination of<br />
Service and Stress Data of Three <strong>Ship</strong>s for Developmentof Hull<br />
Girder Load Criteria,” <strong>Ship</strong> <strong>Structure</strong> <strong>Committee</strong> Report SSC-<br />
287, 1979.<br />
7.1 Gurney, T.R., “The Influence of Thickness of the Fatigue<br />
Strength of Welded Joints,” Proceedings of 2nd International<br />
Conference on Behavior of Offshore <strong>Structure</strong>s (BOSS ‘79),<br />
London, 1979.<br />
10-9
“7.2<br />
Jordan, C.R., and Cochran, C.S., “In-Service Performance of<br />
<strong>Ship</strong> <strong>Structure</strong>Details,” <strong>Ship</strong><strong>Structure</strong>ConnnitteeReport,SSC-<br />
272, 1978.<br />
7*3<br />
Marshall, P.W., “Size Effect in Tubular Welded Joints,” ASCE<br />
Annual <strong>Structure</strong>s Congress, Houston, TX, October 17, 1983.<br />
7.4<br />
Maddox, S.J., “Assessingthe Significanceof Welds Subject to<br />
Fatigue,” Welding Journal, 1070, 53 (9) 401.<br />
7.5<br />
Hartt, W.H., Rengan,K. and Sablok,A.K., “FatigueProperties<br />
of Exemplary High-Strength Steels in Seawater,” Twentieth<br />
Annual Offshore Technology Conerence, OTC 5663, Houston, TX,<br />
May 1988.<br />
7.6<br />
Gurney, T.R., “Fatigue of Welded <strong>Structure</strong>s,” Cambridge<br />
University Press, 2nd. Edition, 1978.<br />
7.7<br />
Grover, J.L., “Initial Flaw Size Estimating Procedures for<br />
Fatigue Crack Growth Calculations,” InternationalConference<br />
on Fatigue of Welded Constructions, Paper No. 15, Brighton,<br />
England, April 1987.<br />
7.8<br />
Tolloczko, J.A. and Lalani, M., “The Implication of New Data<br />
on the Fatigue Life Assessment of Tubular Joints,” Twentieth<br />
Annual Offshore TechnologyConference,OTC 5662, Houston, TX,<br />
May 1988.<br />
7.9<br />
Maddox, S.J., “The Effect of Plate Thickness on the Fatigue<br />
Strength of Fillet Welded Joints,” Welding Institute Report<br />
1987, 7814.01.<br />
7.10<br />
Maddox, S.J., “A Study of the Fatigue Behavior of Butt Welds<br />
Made on Backing Bars,” Proceedings5th European Conference on<br />
Fracture Life Assessment of Dynamically Loaded Materials and<br />
<strong>Structure</strong>s,” Lisbon, 1984, 197-215.<br />
1o-1o
7.11<br />
Maddox, S.J., “Fitnessfor PurposeAssessment of Misalignment<br />
in Transverse Butt Welds Subject to Fatigue Loading,”Welding<br />
InstituteMembers Report 279/1985.<br />
7.12<br />
Marshall, P.W., “Stateof the Artinthe U.S.A.,“ Paper PSIof<br />
Proceedings of the Third International ECSC Offshore<br />
Conferenceon Steel in Marine <strong>Structure</strong>s (SIMS87), Delft,The<br />
Netherlands, 1987.<br />
7.13<br />
Dijkstra, O.D. et al, “The Effect of<br />
Weld Profile on the Fatigue Behavior<br />
Joints,” Seventeenth Annual Offshore<br />
OTC 4866, Houston, TX, May 1985.<br />
Grinding and a Special<br />
of Large Scale Tubular<br />
Technology Conference,<br />
7.14<br />
Bignonnet, A., “Improving the Fatigue Strength<br />
<strong>Structure</strong>s,” Paper PS4 of the Proceedings of<br />
International ECSC Offshore Conference on Steel<br />
<strong>Structure</strong>s (SIMS87), Delft, netherlands, 1987.<br />
of Welded<br />
the Third<br />
in Marine<br />
7.15<br />
Wylde, J.G., Booth, G.S., and Iwasaki, T., Fatigue Tests on<br />
Welded TubularJoints in Air and in Sea Water,” Proceedingsof<br />
International Conference on Fatigue and Crack Growth in<br />
Offshore <strong>Structure</strong>s, I Mech E, 1986.<br />
7.16<br />
Gurney, T.R., “Further Fatigue Tests on Fillet Welded Joints<br />
Under Simple VariableAmplitude Loading,”Appendix A, Welding<br />
InstituteMembers Report 182/1982.<br />
7.17<br />
Gurney, T.R., “Fatigue Tests on Fillet Welded Joints Under<br />
VariableAmplitude Loading,”Welding InstituteMembers Report<br />
293/1985.<br />
7.18<br />
Gurney,T.R., “SomeVariableAmplitudeFatigue Testson Fillet<br />
Welded Joints,” Paper No. 65, International Conference on<br />
Fatigueof Welded Constructions,Brighton, England, 7-9 April<br />
1987.<br />
10-11 /<br />
.7,3<br />
~ --’
7.19 Niemi, E.J., “FatigueTests on Butt and Fillet Welded Joints<br />
Under VariableAmplitude Loading,” Paper No. 8, International<br />
Conference on Fatigue of Welded Constructions, Brighton,<br />
England, 7-9 April 1987.<br />
7.20 Gerald, J. et al, “Comparison of European Data on Fatigue<br />
Under Variable Amplitude Loading,” Paper TS48 of the Third<br />
International ECSC Offshore Conference on Steel Marine<br />
<strong>Structure</strong>s (SIMS 87), Delft, Netherlands, 1987.<br />
7.21 Trufiakov, V.I., and Kovalchuk, V.S., “The Estimation of the<br />
Fatigue Crack Propagation Rate Under Bicyclic Loading,U IIW<br />
Document X111-1139-84, 1984.<br />
7.22 Gurney, T.R., “The Influence of Spectrum Shape on Cumulative<br />
Damage of Plateswith FilletWelded EdgeAttachments,‘Welding<br />
InstituteReports7816.03/86/495.2, 1987 and 7920.01/87/555.1<br />
7.23 Wirsching, P.H., “Digital Simulation of Fatigue Damage in<br />
Offshore <strong>Structure</strong>s,” Computational Method for Offshore<br />
<strong>Structure</strong>s, H. Armen and S.G. Stiansen, Editors, ASME, 1980.<br />
7.24 Chen, Y.N. and Mavrakis, S.A., ‘Close Form Spectral Fatigue<br />
Analysis for Compliant Offshore <strong>Structure</strong>s,” Journal of <strong>Ship</strong><br />
Research.<br />
7.25 Morgan, H.G., “Interaction of Multiple Fatigue Cracks”,<br />
InternationalConference on Fatigue of Welded Constructions,<br />
Paper No. 35, Brighton, England, April 1987.<br />
7.26 Dobson, W.G., Broderick, R.F., Wheaton, J.W., Giannotti,<br />
J. and Stambaugh, K.A., “Fatigue Considerations in View of<br />
Measured Spectra,” <strong>Ship</strong> <strong>Structure</strong>s ConnnitteeReport SSC-315,<br />
1983.<br />
10-12
8.1 Bel1, E.R.G. and Morgan, D.G., “Repair and Analysis of<br />
Cracking in the-hiurchisonFlare Boom,” Twentieth Annual<br />
Offshore Technology Conference, OTC 5814, Houston, TX, May<br />
1988.<br />
8.2 Strouhal, V., Uber Eine Besondere Art der Tonerregung,<br />
Ann. Physik, Leipzig, 1878.<br />
8.3 Marris, A.W., “AReview of Vortex Streets, Periodic Waves and<br />
Induced Vibration Phenomena,” Journal of Basic Eng., V. 86,<br />
1964, pp 185-194.<br />
8.4 Sarpkaya, T., and Isaacson, M., Mechanics of Wave Forces on<br />
Offshore <strong>Structure</strong>s,Van Nostrand Reinhold, New York, 1981.<br />
8.5 Hunt, R.J., “Practice For Establishing If A <strong>Structure</strong> Will<br />
Undergo Vortex-InducedVibrations,”CONFIDENTIALSIPM Report,<br />
EPD/112, June 1987.<br />
8.6 Engineering Sciences Data Unit, “Across Flow Response Due to<br />
Vortex Shedding”, Publication No. 78006, London, England,<br />
October, 1978.<br />
8.7 Det Norske Veritas, “Rules for Submarine Pipeline Systems”,<br />
Oslo, Norway, 1981.<br />
8.8 Zedan, M.F., Yeung, J.Y., Ratios, H.J., and Fischer, F.J.,<br />
“Dynamic Response of a Cantilever Pile to Vortex Shedding in<br />
RegularWaves,”Proceedingsof OffshoreTechnology Conference,<br />
OTC Paper No. 3799, Houston, May 1980.<br />
8.9 Hallam, H.G., Heaf, N.J., and Wootton, F.R., “Dynamics of<br />
Marine <strong>Structure</strong>s,”CIRIA Underwater EngineeringGroup Report<br />
UR8 (2nd Edition), 1977.<br />
10-13 -,<br />
“7 ----1,<br />
,, d<br />
,>
9.1 Skaar, K.T., “ContributingFactorsto<strong>Ship</strong>Qual ity”,Reportof<br />
ConnnitteeV.3-Service experience - <strong>Ship</strong>s, Proceedings of the<br />
Tenth International <strong>Ship</strong> and Offshore <strong>Structure</strong>s Congress,<br />
Volume 2, Lyngby, Denmark, August 1988.<br />
9.2 Jordan, G.P. and Cochran, C.S., “In-Service Performance of<br />
StructuralDetails,”<strong>Ship</strong> <strong>Structure</strong>ConnnitteeReport SSC-272,<br />
1978.<br />
9.3 Notes on Structural Failure in <strong>Ship</strong>s, Report NO. 19, Lloyd’s<br />
Register of <strong>Ship</strong>ping, 1962.<br />
9.4 Maddox, S.J., and Padilla, J.A., “Fatigue Life Improvementby<br />
Water Jet Erosion”,Welding InstituteMembers Report 280/1985.<br />
9.5 King, C.G, “Abrasive Water Jetting: A New Aid to Welded<br />
Fabrications,” OTC Paper No. 5817, 20th Annual Offshore<br />
Technology Conference, Houston TX, May 1988.<br />
..<br />
9.6 Wylde, J.G., Booth, G,S., and Iwasaki, T., “Fatigue Tests<br />
Welded Tubular Joints in Air and Sea Water”, Proceedings<br />
International Conference on Fatigue and Crack Growth<br />
Offshore <strong>Structure</strong>s, I Mach E, 1986, pp. 155-170.<br />
on<br />
of<br />
in<br />
9*7 Booth, G.S., “Chapter 2 - A Review of Fatigue Strength<br />
Improvement Techniques”, Improving the Fatigue Strength of<br />
Welded Joints, The Welding Institute, 1983.<br />
9.8 Maddox, S.J., “Improvingthe FatigueStrength of Welded Joints<br />
by Peening”, Metal Construction, 1985, 17(4) pp 220-224.<br />
9.9 Woodley, C.C., “Chapter 4 PracticalApplications of Weld Toe<br />
Grinding”, Improving the Fatigue Strength of Welded Joints,<br />
The Welding Institute, 1983.<br />
9.10 Boylston, J.W., and Stambaugh,K.A., “Developmentofa Plan to<br />
Obtain In-ServiceStill Water Bending Moment Information for<br />
10-14<br />
~ Zy<br />
f-‘<br />
,-+
StatisticalCharacterization,”<strong>Ship</strong><strong>Structure</strong>ConnnitteeReport<br />
SSC-319, 1984. -<br />
9.11<br />
Krenk, S. and Gluver, H., “A Markow Matrix for Fatigue Load<br />
Simulation and Rainflow Range Evaluation,” Symposium on<br />
Stochastic Structural Dynamics, Urbana, Illinois, 1988.<br />
9.12<br />
Krenk, S. and Thorup, E., “Stochasticand Constant Amplitude<br />
Fatigue Test of Plate Specimens with a Central Hole,” Report<br />
No. R 242, Department of Structural Engineering, Technical<br />
University of Denmark, 1989.<br />
9.13<br />
Agerskov, H. and Aarkrog, P., “Fatigue Investigation on<br />
Offshore Steel <strong>Structure</strong>s Under Spectrum Loading,”<br />
InternationalSyposiumon Offshore Brazil ’89,Riode Janeiro,<br />
Bvazil, August 1989.<br />
,.<br />
9.14<br />
Morgan, G.G., “Interaction of Multiple Fatigue Cracks,”<br />
InternationalConference on Fatigue of Welded Constructions,<br />
Brighton, England, 7-9 April 1987.<br />
*IJ.S. G.P.0:1993-343-273:80107 10-15<br />
,.— .-
COMMllTEE<br />
ON MARINE STRUCTURES<br />
Commission on Engineering and Technical Systems<br />
National Academy of Sciences - National Research Council<br />
The COMMITTEE ON MARINE STRUCTURES has technical cognizance over the interagency<br />
<strong>Structure</strong> <strong>Committee</strong>’s research program.<br />
Peter M. Palermo Chairmanj Alexandria, VA<br />
Mark Y. Berman, Amoco Production Company, Tulsa, OK<br />
Subrata K. Chakrabarti, Chicago Bridge and Iron, Plainfield, IL<br />
Rolf D. Glasfeld, General Dynamics Corporation, Groton, CT<br />
William H. Harttl Florida Atlantic University, Boca Raton, FL<br />
Alexander B. Stavovy, National Research Council, Washington, DC<br />
Stephen E. Sharpe, <strong>Ship</strong> <strong>Structure</strong> <strong>Committee</strong>, Washington, DC<br />
LOADS WORK<br />
GROUP<br />
Subrata K. Chakrabarti Chairman, Chicago Bridge and Iron Company, Plainfield, IL<br />
Howard M. Bunch, University of Michigan, Ann Arbor, Ml<br />
Peter A. Gale, John J. McMullen Associates, Arlington, VA<br />
Hsien Yun Jan, Martech Incorporated, Neshanic Station, NJ<br />
Naresh Maniar, M. Rosenblatt & Son, Incorporated, New York, NY<br />
Solomon C. S. Yim, Oregon State University, Corvallis, OR<br />
MATERIALS WORK GROUP<br />
William H. Hartt Chairman, Florida Atlantic University, Boca Raton, FL<br />
Santiago Ibarra, Jr., Amoco Corporation, Naperville, IL<br />
John Landes, University of Tennessee, Knoxville, TN<br />
Barbara A. Shaw, Pennsylvania State University, University Park, PA<br />
James M. Sawhill, Jr., Newport News <strong>Ship</strong>building, Newport News, VA<br />
Bruce R. Somers, Lehigh University, Bethlehem, PA<br />
Jerry G. Williams, Conoco, Inc., Ponca City, OK<br />
c-3 $olo7&
SHIP STRUCTURE COMMITTEE PUBLICATIONS<br />
SSC-351 An Introduction to Structural Reliability Theorv by Alaa E. Mansour<br />
1990<br />
SSC-352 Marine Structural Steel Touahness Data Bank by J. G. Kaufman and<br />
M. Prager 1990<br />
SSC-353 Analvsis of Wave Characteristics inExtreme Seas by WilliamH.Buckley<br />
1989<br />
SSC-354 Structural Redundance forDiscreteand Continuous Systems by P.K.<br />
Das and J.F.Garside 1990<br />
SSC-355 Relation of Inspection Findinqs to Fatique Reliability by M. Shinozuka<br />
1989<br />
SSC-356 Fatique Performance Under Multiaxial Lod by Karl A. Stambaugh,<br />
Paul R. Van Mater, Jr.,and WilliamH.Munse 1990<br />
SSC-357 Carbon Equivalence and Weldability of Microalloyed Steels by C. D.<br />
Lundin, T. P. S. Gill, C. Y. P, Qiao, Y. Wang, and K. K. Kang 1990<br />
SSC-358 Structural Behavior After Fatique by Brian N. Leis 1987<br />
SSC-359 Hydrodynamic Hull Dampinq (Phase 1) by V. Ankudinov 1987<br />
SSC-360 Use of Fiber Reinforced Plastic inMarine <strong>Structure</strong>sby EricGreene<br />
1990<br />
SSC-361 Hull Stra~pinq of <strong>Ship</strong>= by Nedret S. Basar and Roderick B. Hulls 1990<br />
SSC-362 <strong>Ship</strong>board Wave Heiqht Sensor by R. Atwater 1990<br />
SSC-363 Uncertainties in Stress Analysis on Marine <strong>Structure</strong>s by E. Nikolaidis<br />
and P. Kaplan 1991<br />
SSC-364 Inelastic Deformation of Plate Panels by Eric Jennings, Kim Grubbs,<br />
Charles Zanis, and Louis Raymond 1991<br />
SSC-365 Marine Structural Inteqrity Proqrams (MSIP) by Robert G. Bea 1992<br />
SSC-366 Threshold Corrosion Fatique of Welded <strong>Ship</strong>building Steels by G. H.<br />
Reynolds and J. A. Todd 1992<br />
None <strong>Ship</strong> <strong>Structure</strong> <strong>Committee</strong> Publications - A Special Bibliography<br />
c-q<br />
f@lo7-s9
U.S.Department<br />
of Transportation<br />
United States<br />
CoastGuard<br />
/47<br />
●<br />
●<br />
,*c%,,<br />
#<br />
~<br />
@<br />
%,J<br />
SSC-<strong>367</strong><br />
FATIGUE TECHNOLOGY<br />
ASSESSMENT AND STRATEGIES<br />
FOR FATIGUE AVOIDANCE IN<br />
MARINE STRUCTURES<br />
APPENDICES<br />
his dmumcnt has been approvsd<br />
for public rdsaseandsalwits<br />
distribution is unlimited<br />
SHIP STRUCTURE<br />
COMMITTEE<br />
1993 C-,1 %r@llz’””<br />
.,-Q#q<br />
/ ‘$’”
~HIP STRUCTURF CO MMllTEF<br />
The SHIP STRUCTURE COMMllTEE is constituted to prosecute a research program to improve the hull structures of ships and other<br />
marine structures by an extension of knowledge pertaining to design, materials, and methods of construction.<br />
RADM A. E. Henn, USCG (Chairman)<br />
Chief, Office of Marine Safety, Security<br />
and Environmental Protection<br />
U, S, Coast Guard<br />
Mr. Thomas H. Peirce Mr. tit T, Hailer Dr. Donald Mu<br />
Marine Research and Development Associate Administrator for <strong>Ship</strong>- Senior Vrce President<br />
Coordinator building and <strong>Ship</strong> Operations<br />
Ameri=n Bureau of <strong>Ship</strong>ping<br />
Transportation Development Center Maritime Administration<br />
Transport Canada<br />
Mr. Alexander Malakhoff Mr. Thomas W. Allen CDR Stephen E. Sharpe, USCG<br />
Director, Structural Integrity Engineering Officer (N7) Executive Director<br />
Subgroup (SEA 05P) Military Sealift Command<br />
Naval Sea Systems Command :Y1.%%;s?m’”ee<br />
CONTRACTING OFFICER TECHNICAL REPRESENTATIVE<br />
Mr. William J. Siekierka<br />
SEA05P4<br />
Naval Sea Systems Command<br />
SHIP ST<br />
RUCTLJRFSUR~OMMllTFF<br />
The SHIP STRUCTURE SLIBCOMMllTEE acts for the <strong>Ship</strong> <strong>Structure</strong> <strong>Committee</strong> on technical matters by providing technical<br />
coordination for determinating the goals atrd objectives of the program and by evaluating and interpreting the results in terms of<br />
structural design, construction, and operation.<br />
AMERICAN BUREAU OF SHIPPING NAVAL S EA SYSTEMS COM MAND TRANSPORT CANADA<br />
Mr. Stephen G. Arntson (Chairman) Dr. Robert A. Sielski<br />
Mr. John Grinstead<br />
Mr. John F, ConIon<br />
Mr. Charles L Null Mr. Ian Bayly<br />
Dr. John S, Spencer Mr. W. Thomas Packard Mr. David L, Stocks<br />
Mr, Glenn M, Ashe Mr. Allen H. Engle Mr. Peter Timonin<br />
&llLITARY SEAI IFT COM MAND jYIARITIME ADMINI STRA TIO~ U, S, COAST GU ARD<br />
Mr, Robert E. Van Jones Mr. Frederick Seibold CAPT T. E. Thompson<br />
Mr. Rickard A Anderson Mr. Norman 0, Hammer<br />
CAPT W. E, Colburn, Jr.<br />
Mr. Michael W. Touma<br />
Mr. Chao H. Lin<br />
Mr. Rubin Scheinberg<br />
Mr, Jeffrey E. Beach Dr. Walter M. Maclean Mr. H. Paul Cojeen<br />
SHIP STRUCTURE SUBCOMMITTEE LIAISON MEMBERS<br />
~Y<br />
LCDR<br />
Bruce R. Mustain<br />
~s -<br />
~n<br />
Mr. Alexander<br />
B. Stavovy<br />
U. S. M FRCHANT MARINE ACAO EMY<br />
Dr. C. B, Kim<br />
~Y<br />
NATIONAL ACADEMY OF SCIENCES -<br />
COMMllTEE ON MARINE STRUCTURES<br />
Mr. Peter M. Palermo<br />
Dr. Ramswar<br />
Bhattacharyya<br />
~SEARCHCOUNCll<br />
STATE UMNIV~~SITY OF NEW YORK<br />
Dr. Martin<br />
Prager<br />
Dr. W. R. Porter<br />
SOCIETY OF NAVAL ARC HITECTS AND<br />
MARINE ENGINEERS<br />
AMERICAN IRON AND STEEL INSTITUTE<br />
Mr, Alexander D. Wilson<br />
DEPARTMENTOF NATIONAL DEFENCE - CANADA<br />
Dr. William<br />
Sandberg<br />
Dr, Neil G. Pegg<br />
OFFI CE OF NAVAL RESEARCH<br />
Dr.<br />
Yapa D, S. Rajapaske
Member Agencies:<br />
United States Coast Guard<br />
Naval Sea Systems Command<br />
Maritime Administration<br />
American Bureau of Sh@ping<br />
Military Sealifi Command<br />
Transpmt Canada<br />
<strong>Ship</strong><br />
<strong>Structure</strong><br />
<strong>Committee</strong><br />
An Interagency Advisory<strong>Committee</strong><br />
May 17, 1993<br />
Address Correspondence to:<br />
Executive Director<br />
<strong>Ship</strong> <strong>Structure</strong> <strong>Committee</strong><br />
U.S. Coast Guard (G-Ml/R)<br />
2100 Second Street, S.W.<br />
Washington, D.C. 20593-0001<br />
PH: (202) 267-0003<br />
FAX: (202) 267-4677<br />
SSC-<strong>367</strong><br />
SR-1324<br />
FATIGUE TECHNOLOGY ASSESSMENT AND STRATEGIES FOR FATIGUE<br />
AVOIDANCE IN MARINE STRUCTURES<br />
.—..<br />
This report synthesizes the state-of–the-art in fatigue<br />
technology as it relates to the marine field. Over the years<br />
more sophisticated methods have been developed to anticipate the<br />
life cycle loads on structures and more accurately predict the<br />
failure modes. As new design methods have been developed and<br />
more intricate and less robust structures have been built it has<br />
become more critical than ever that the design tools used be the<br />
most effective for the task. This report categorizes fatigue<br />
failure parameters, identifies strengths and weaknesses of the<br />
available design methods, and recommends fatigue avoidance<br />
strategies based upon variables that contribute to the<br />
uncertainties of fatigue life.<br />
This set of Appendices includes more in-depth presentations of<br />
the methods used in modeling the loads from wind and waves,<br />
linear system response to random excitation, stress concentration<br />
factors, vortex shedding and fatigue damage calculation.<br />
G.P.Y&n<br />
..<br />
A. E. HENN<br />
Rear Admiral, U.S. coast Guard<br />
Chairman, <strong>Ship</strong> <strong>Structure</strong> <strong>Committee</strong><br />
.-.
T*chnical<br />
Report Documentation Poge<br />
1. Report No. 2. Governmtmt Acc*ssioti No. 3. Rocip,~ntrs Catalog Ne.<br />
4. Title and subtitle 5. Rcportiko<br />
June1992<br />
FATIGUEDESIGNPROCEDURES 6.Performing Organization Cod=<br />
7. AuthorIs)<br />
8. PorfaminQ 0r9anizatisn R9part No.<br />
CuneytC.Capanoglu<br />
SR-1324<br />
9. Prrfnrmin9 Or~anization Namm rnd AdAmss 10. Work Unit No. (TRAIs)<br />
EARL AND WRIGHT 11. Controctar Gr~nt N*.<br />
180 Howard Street<br />
DTCG23-88-C-20029<br />
San Francisco, CA 94105<br />
13. TYP. of R-part ondPried C-word<br />
~2. 5POm~Or~mQ Aq~n=yNa~~ ~mdAddr*~~<br />
<strong>Ship</strong> <strong>Structure</strong> <strong>Committee</strong><br />
Final Repoti<br />
U.S. Coast Guard (G-M)<br />
2100 Second Street, SW<br />
IS. supplementary Notes<br />
Washington, DC 20593<br />
I<br />
14. Sponsoring Agency Cod.<br />
Sponsoredbythe<strong>Ship</strong><strong>Structure</strong> <strong>Committee</strong>anditsmembersagencies.<br />
G-M<br />
ABSTRACT<br />
This report provides an up-to-date assessment of fatigue technology, directed specifically toward<br />
the marineindustry, A comprehensive overview of fatigue analysis and design, a global review of<br />
fatigue including rules and regulations and current practices, and a fatigue analysis and design<br />
criteria, are provided as a general guideline to fatigue assessment. A detailed discussion of all<br />
fatigue parameters is grouped under three analysis blocks:<br />
● Fatigue stress model, covering environmental forces, structure response and loading, stress<br />
response amplitude operations (RAOS] and hot-spot stresses<br />
● Fatigue stress history model covering long-term distribution of environmental loading<br />
● Fatigue resistance of structures and damage assessment methodologies<br />
The analyses and design parameters that affect fatigue assessment are discussed together with<br />
uncertainties and research gaps,to provide a basis for developing strategies for fatigue avoidance.<br />
Additional in-depth discussions of wave environment, stress concentration factors, etc. are<br />
presented in the appendixes.<br />
17. Key Words<br />
18. Distribution Statwn.nt<br />
Assessment of fatigue technology,<br />
fatigue stress models, fatigue Available from:<br />
stress history models, fatigue National Technical information Serv.<br />
resistance, fatigue parameters U. S, Department of Commerce<br />
and fatigue avoidance strategies Springfield, VA 22151<br />
19. .lecunty Cl=$sif. (~f this r-pert)<br />
Unclassified<br />
I<br />
I<br />
20. s~curity Classif. (of this POW)<br />
Unclassified<br />
Form DOT F 1700.7 (8-72) Reproduction Oi completed page authorized<br />
I<br />
21. No. O{ PU9*S I<br />
22. Pri CC<br />
194 EXCI,<br />
Appendixas
‘ METRIC CONVERSION FACTORS<br />
ApproMimMcCofrvmsiortsto M-ttie<br />
Measures<br />
Approximate Convwsions from Mstric Measures<br />
Svmbol Whoa Yss Know Multiply br 1. Find Srmbd<br />
LEUGTH<br />
To Find<br />
{n<br />
rt<br />
d<br />
mi<br />
LENGTH<br />
iriches “2.5<br />
Ieet 30<br />
V*lds 0.9<br />
mi!0s 1.6<br />
AflEA<br />
cent imtsm<br />
Csnl imtercl<br />
mslers<br />
kilcamsters<br />
cm<br />
cm<br />
m<br />
km<br />
\m m<br />
h<br />
m<br />
cm<br />
mil!inwm.m 0.04<br />
cen!imstem 0.4<br />
MStelm 3.3<br />
mstms 1.1<br />
hilcrrmtsm 0.6<br />
AREA<br />
inchea<br />
inchea<br />
feet<br />
#s<br />
miles<br />
in<br />
in<br />
{1<br />
yd<br />
mi<br />
square inche9 6.5<br />
qu~mfeet 0,03<br />
squareyard- 0.0<br />
wrtmm miles 2.8<br />
WI*S 0.4<br />
MASS<br />
{wci~ht)<br />
square<br />
centimeters<br />
square Iwlem<br />
SqUam ms!em<br />
sqtmm<br />
hectares<br />
kilwnalms<br />
cm2<br />
#<br />
~2<br />
kmz<br />
ha<br />
3quH0centkmtms 0.16<br />
equem rmdem 1.2<br />
!#rIIDmkihmwmem 0.4<br />
tmctmY4 {10,000 m2) 2.6<br />
MASS<br />
[woighl]<br />
squme inches<br />
square yamrs $<br />
squsm miles<br />
mi2<br />
■cms<br />
or<br />
lb<br />
ounces<br />
2B<br />
pumds 0.45<br />
ad Imm 0.9<br />
@UI lb}<br />
grsms<br />
kibgmms<br />
Iwnms<br />
Q<br />
kg<br />
I<br />
o<br />
kg<br />
t<br />
grwm 0.035<br />
kifmm 2.2<br />
tmlnes [1OOOho) 1.1<br />
csmces<br />
Smmds<br />
Oz<br />
14<br />
VOLUME<br />
VOLUME<br />
TrwP<br />
tee9pms 5<br />
tnbte~poons 15<br />
fluid ounces 30<br />
cups 0.24<br />
pints 0.4?<br />
quarts 0.95<br />
galtmil 3,8<br />
cubic !Set 0.03<br />
cubtc yerds 0.76<br />
mill ili!em<br />
mitli litem<br />
mit Iilitem<br />
liters<br />
liters<br />
Iilers<br />
liters<br />
cubic meters<br />
cubic melevs<br />
ml<br />
ml<br />
ml<br />
I<br />
1<br />
I<br />
I<br />
~3<br />
~3<br />
ml<br />
!<br />
1<br />
I<br />
~3<br />
~3<br />
rni m tiwm 0.03<br />
Iitsrs 2.1<br />
Iitem 1.m<br />
Iiims 0.26<br />
cubic mstem 35<br />
cubic memrs 1.3<br />
TEMPERATURE<br />
[S12SCi~<br />
Ituid unCOS 1103<br />
pints<br />
m<br />
qualm<br />
qt<br />
gallons<br />
gsl<br />
chic reet i?<br />
cubic yalds vd3<br />
TEMPERATUFIE<br />
‘F Fahrenheit 5/9 Iaflm<br />
tempemtum subtrmct ing<br />
32}<br />
{exsci)<br />
.1 m = 2.54 P2xacllvl. F#wother enad co.versdws and more Nrtanicd tables, see MM Mqsc. PIIM. 2W<br />
Unts of We, ghW md.Mea5uIe5. P. Ice 3-2.25,SD Cal-log !h. Ct3.102!36.<br />
“c<br />
Celsius<br />
9/5 [then<br />
Fahranhefit ‘F<br />
tsmpemlum mkl32)<br />
Wwerstum<br />
Celsius<br />
“c<br />
temmmaaure ‘F<br />
‘F 32 98.6 212<br />
-40 0 40 00 t20 160 Zcm<br />
1 I I 4 I # 1 I 1 # l,t+; i:l,ll, t~<br />
l’; 1 1 1 I<br />
-$: -20 0 20 40 60 “ 60 IOD<br />
37<br />
Oc
FATIGUE TECHNOLOGY<br />
ASSESSMENT AND STRATEGIES<br />
FOR FATIGUE AVOIDANCE IN<br />
MARINE STRUCTURES<br />
APPENDICES<br />
CONTENTS<br />
APPEN131X A Review oftheOcean Environment<br />
APPENDIX B Review ofLinearSystem Response toRandom Excitation<br />
,—<br />
APPENDIX C StressConcentrationFactors<br />
APPENDIX D VortexShedding Avoidance and FatigueDamage Computation
APPENDIX A<br />
REVIEW OF OCEAN ENVIRONMENT<br />
CONTENTS<br />
A.<br />
REVIEW OF OCEAN ENVIRONMENT<br />
A.1<br />
A.2<br />
IRREGULAR WAVES<br />
PROBABILITY CHARACTERISTICS OF WAVE SPECTRA<br />
A.2.1<br />
A.2.2<br />
Characteristic Frequencies and Periods<br />
Characteristic Wave Heights<br />
A.3<br />
WAVE SPECTRA FORMULAS<br />
A.3.1<br />
A.3.2<br />
A.3.3<br />
A.3.4<br />
Bretschneider and ISSC Spectrum<br />
Pierson-MoskowitzSpectrum<br />
JONSWAP and Related Spectra<br />
Scott and Scott-Wiegel Spectra<br />
A.4<br />
SELECTING A WAVE SPECTRUM<br />
A.4.1<br />
A.4.2<br />
Wave Hindcasting<br />
Direct Wave Measurements<br />
A.5<br />
A.6<br />
A.7<br />
A.8<br />
A.9<br />
WAVE SCAITER DIAGRAM<br />
WAVE EXCEEDANCE CURVE<br />
WAVE HISTOGRAM AND THE RAYLEIGH DISTRIBUTION<br />
EXTREME VALUES AND THE WEIBULL DISTRIBUTION<br />
WIND ENVIRONMENT<br />
A.9.1<br />
A.9.2<br />
A.9.3<br />
Air Turbulence, Surface Roughness and Wind Profile<br />
Applied, Mean and Cyclic Velocities<br />
Gust Spectra<br />
A.1O<br />
REFERENCES
A. REVIEH OF OCEANENVIRONMENT<br />
The ocean environment is characterized by waves, wind and current.<br />
The waves are typically irregular (confused or random seas). Some<br />
waves are generated locally by the wind, and some waves are generated<br />
great distances away. The wind is unsteady, with gusts. The wind<br />
varies with height above water. The current is caused by the wind,<br />
by waves, by the tide, and by global temperature differences. The<br />
current varies with depth. All of these characteristics vary with<br />
time.<br />
A.1 IRREGULARWAVES<br />
Irregular waves (a random sea) can be described as the sum of an<br />
infinite number of individual regular (sinusoidal)waves of different<br />
amplitude, frequency, and phase (Figure A-l). Therefore, the<br />
randomly varying sea surface elevation can be represented by a<br />
Fourier series.<br />
N<br />
n(t) = z<br />
i=l<br />
a*cos(wi*t+Oi) i<br />
—<br />
where n(t) is the water surface elevation measured from<br />
still water level,<br />
ai<br />
is the amplitude of each component regular wave,<br />
w. is the frequency of each component regular wave,<br />
1<br />
$j is the phase ang’ e of each component regular<br />
wave, and<br />
t<br />
is time.<br />
A-1<br />
,/-—,<br />
w!
The most distinctive feature-of a random sea is that it never repeats<br />
its pattern and it is impossible to predict.its shape. -Therefore,<br />
total energy is used to define a particular sea. The energy (E) in<br />
an individual regular wave per unit surface area is,<br />
E = +*p*g*a2<br />
and the total energy of the sea is the sum of the energies of the<br />
constituent regular waves.<br />
N<br />
E = $*P*g* z a:<br />
i=l<br />
The total energy of the sea is distributed according to the<br />
frequencies of the various wave components. The amount of energy per<br />
unit surface area within the small frequency band (tii,wi+dm) is,<br />
E{wi) = +*D*g*ai2*d~<br />
The total energy of the sea is then the sum of the energies within<br />
the individualwave components. If the sea is made up of an infinite<br />
number of waves, the energies of the waves form a smooth curve, and<br />
the above summationmay be replaced by an integral.<br />
The smooth distribution of the wave energy is called the energy<br />
spectrum or wave spectrum of the random sea, and is often designated<br />
as S(W).<br />
A wave spectrum is normally depicted as a curve with an<br />
ordinate of energy and an abscissa of frequency. A typical wave<br />
spectrum has a central peak with a tapered energy distribution either<br />
side of the peak.<br />
A-2
The recommended form of displaying a wave spectrum is with an<br />
ordinate of %*a2 and an abscissa of M, radial frequency. However,<br />
since the engineer will encounter wave spectra equations in a number<br />
of forms, using various bases and units, the applicable conversion<br />
factors are provided in the following sections.<br />
Suectrum Basis<br />
The recommended spectrum basis is half amplitude squared or energy.<br />
Often spectrum equations having a different basis are encountered.<br />
Before any statistical calculations are performed with a spectrum<br />
equation, the equation should be converted to the recommended basis.<br />
For a “half amplitude” or “energy” spectrum, the basis is one-half<br />
times the amplitude squared.<br />
S(M) = **n*<br />
S(m)dm = E / (pg)<br />
where,<br />
s<br />
is the spectral ordinate,<br />
(IJ<br />
is the radial frequency,<br />
is the wave amplitude of the constituent wave of<br />
frequency,<br />
E<br />
is the energy content of the constituent wave of<br />
frequency, M.<br />
For an “amplitude” spectrum, the basis is amplitude squared.<br />
A-3
s(w) = 2*(4*112) -<br />
For a “height” spectrum, the basis is height squared.<br />
S(u) = h2<br />
s(u) = 8*(+*n2)<br />
where,<br />
h<br />
is the height of the constituent wave of<br />
frequency, M.<br />
For a “height double” spectrum, the basis is two times the height<br />
squared.<br />
s(w) = s*h2<br />
,. ,<br />
S(u) = 16*(%*r12)<br />
The basis of the spectrum must be determined before the spectrum<br />
used in an analysis, because the ordinate of one representation<br />
the spectrum may be as much as 16 times as great as the ordinate<br />
another representation.<br />
is<br />
of<br />
of<br />
Units<br />
The spectrum equation may be expressed in terms of radial frequency,<br />
circular frequency, or period. Conversion between circular frequency<br />
and radial frequency is accomplished by multiplying by the<br />
constant, 27T.<br />
u . 2T*f<br />
s(w) = s(f) / (21r)<br />
A-4
where,<br />
f is the circular frequency.<br />
The conversion between period and radial frequency is more<br />
complicated.<br />
111 = 2T/T<br />
S(f) = T2*S(T)<br />
s(w) = T2*S(T) / (2Ir)<br />
where,<br />
T is the period.<br />
When converting between period and frequency, the abscissa axis is<br />
reversed. Zero period becomes infinite frequency, and infinite<br />
period becomes zero frequency.<br />
Wave spectrum equations may be used with any length units by<br />
remembering that the spectrum ordinate is proportional to amplitude<br />
squared or height squared.<br />
—<br />
‘(m)meter = (0.3048)2*S(~)feet<br />
The mathematical formulation for the wave spectrum equation will<br />
often include the significant height squared or the gravitational<br />
constant squared, which when entered in the appropriate units will<br />
convert the equation to the desired length units.<br />
A.2 PROBABILITY CHARACTERISTICSOF WAVE SPECTRA<br />
The characteristics of ocean waves are determined by assuming that<br />
the randomness of the surface of the sea can be described by two<br />
A-5
common probability distributions, the Gaussian (or normal)<br />
distribution and-the Rayleigh distribution. These probability<br />
distributions are used to define the distribution of wave elevations, n,<br />
and of wave heights, H, respectively.<br />
A.2.1<br />
CharacteristicFrequencies and Periods<br />
For design purposes sea spectra equations are selected to represent<br />
middle aged seas that would exist some time after a storm, yet which<br />
are still young enough to have a good dispersion of wave<br />
frequencies. The primary assumption about the design seas is that<br />
the wave elevations follow a Gaussian or normal distribution.<br />
Samples of wave records tend to support this assumption. In<br />
conjunction with the Gaussian distribution assumption, the wave<br />
elevations are assumed to have a zero mean. Digitized wave records<br />
tend to have a slight drift of the mean away from zero, usually<br />
attributed to tide or instrument drift. The Gaussian distribution<br />
assumption is equivalent to assuming that the phase angles of the<br />
constituentwaves within a wave spectrum, are uniformly distributed.<br />
The Gaussian distribution allows one to calculate statistical<br />
parameters which are used to describe the random sea. The mean<br />
elevation of the water surface is the first moment of the Gaussian<br />
probability density function. The mean-square is the second moment<br />
taken about zero, and the root-mean-square is the positive square<br />
root of the mean-square. The variance is the second moment taken<br />
about the mean value. The standard deviation is the positive square<br />
root of the variance. Since the wave elevations are assumed to have<br />
a zero mean value, the variance is equal to the mean-square, and the<br />
standard deviation is equal to the root-mean-square. In present<br />
practice, the area under a random wave energy spectrum is equated to<br />
the variance.<br />
In a similar way, the characteristic frequencies and periods of a<br />
wave spectrum are defined in terms of the shape, the area, and/or the<br />
area moments of the ~*a2 wave spectrum. Depending upon the<br />
A-6<br />
i<br />
ij<br />
-,,
particular wave spectrum formula, these characteristic periods may or<br />
may not-reflect any real period. The area and area moments are<br />
calculated as follows.<br />
Area:<br />
Nth Area Moment:<br />
mn=f~w<br />
‘*S(m)dw<br />
The characteristic frequencies and periods are defined as follows.<br />
mm:<br />
Peak frequency<br />
The peak<br />
5(w)<br />
is<br />
frequency is<br />
a maximum.<br />
the frequency at which the spectral ordinate,<br />
‘P:<br />
Peak period<br />
The peak<br />
period is the<br />
S(M) is a maximum.<br />
period corresponding to the frequency at which<br />
T = 2il/um<br />
P<br />
Tm:<br />
Modal period<br />
The modal period is the period at which S(T) is a maximum. Since the<br />
spectrum equations in terms of frequency and in terms of period<br />
differ by the period squared factor, the modal period is shifted away<br />
from the peak period.<br />
A-7<br />
/‘f
T: v.<br />
Visually Observed Period, or Mean Period, or Apparent<br />
Period<br />
The visually observed period is the centroid of the S(u) spectrum.<br />
The International <strong>Ship</strong> <strong>Structure</strong>s Congress (ISSC) and some<br />
environmental reporting agencies have adopted Tv as the period<br />
visually estimated by observers.<br />
Tv = 2~*(mo/m2) 4<br />
Tz :<br />
Average Zero-uncrossing period or Average Period<br />
The average zero-uncrossing period is the average period between<br />
successive zero up-crossings. The average period may be obtained<br />
from a wave record with reasonable accuracy.<br />
Tz<br />
= 2n*(mo/mz) 4<br />
Tc :<br />
Crest Period<br />
The crest period is the average period between successive crests.<br />
The crest period may be taken from a wave record, but its accuracy is<br />
dependent upon the resolution of the wave measurement and recording<br />
equipment and the sampling rate.<br />
Tc<br />
= 2~*(mz/mq) %<br />
T5:<br />
Significant<br />
Period<br />
The significantperiod is the average period of the highest one-third<br />
of the waves. Some environmental reporting agencies give the sea<br />
characteristicsusing Ts and Hs, the significant wave height. There<br />
are two equations relating Ts to Tp.<br />
Ts = 0.8568*TP, Old<br />
A-8
T~ = 0.9457*T P’<br />
New<br />
The first equation applies to original Bretschneider wave spectrum,<br />
and the second is the result of recent wave studies (See Reference<br />
A.I).<br />
The peak period, Tp, is an unambiguous property of all common wave<br />
spectra, and is therefore the preferred period to use in describing a<br />
random sea.<br />
A.2.2<br />
CharacteristicWave Heiqhts<br />
From the assumption that the wave elevations tend to follow a<br />
Gaussian distribution, it is possible to show that the wave heights<br />
follow a Rayleigh distribution. Since wave heights are measured from<br />
a through to succeeding crest, wave heights are always positive which<br />
agrees with the non-zero property of the Rayleigh probability<br />
density. From the associated property that the wave heights follow a<br />
Rayleigh distribution, the expected wave height, the significantwave<br />
height, and extreme wave heights may be calculated. The equation for<br />
the average height of the one-over-nth of the highest waves is as<br />
follows.<br />
—<br />
where:<br />
H,,n/ (mo) + = 2* [2*ln(n)]%+<br />
n*(2~)%*{l-erf[(ln(n)}%l}<br />
m<br />
o<br />
is the variance or the area under the energy<br />
spectrum,<br />
In<br />
is the natural logrithm,<br />
erf<br />
is the error function, (the error function is<br />
explained and tables of error function values are<br />
available in mathematics table books.)<br />
A-9
The characteristic wave heights of a spectrum are related to the<br />
total energy in the spectrum. The energy is proportional to the area<br />
under the +*a2 spectrum.<br />
Ha:<br />
Average Wave Height<br />
The average or mean height of all of the waves is found by setting<br />
n=l.<br />
Ha= 2.51*(mO)%<br />
H~: Significant Height<br />
The significant height is the average height of the highest one-third<br />
of all the waves, often denoted as H 1/3”<br />
Hs = 4.00*(mO)*<br />
H max:<br />
Maximum Height<br />
The maximum height is the 1-gest wave height expected<br />
number of waves, (n on the order of 1000), or over a<br />
period, (t on the order of hours).<br />
among a large<br />
long sampling<br />
The maximum wave height is often taken to be the average of the<br />
l/1000th highest waves.<br />
H = 7.94*(m )% = 1.985*Hs<br />
1/1000 o<br />
Using the one-over-nth equation and neglecting the second term gives<br />
the following equation.<br />
or<br />
H<br />
1/n<br />
= 2*[ln(n)]+*(mO)%<br />
A-10
H = 2*[2*ln(n) ]+*Hs<br />
1/n<br />
For n = 1000, this gives,<br />
H<br />
1/1000<br />
= 7.43*(mO)+ = 1.86*Hs<br />
For a given observation time, t, in hours, the most probable extreme<br />
wave height is given by the following equation.<br />
H max =<br />
2*[2*mO*ln(3600*t/Tz)]%<br />
The 3600*t/Tz is the average number of zero up-crossings in time, t.<br />
-><br />
A.3<br />
WAVE SPECTRA FORMULAS<br />
The Bretschneider and Pierson-Moskowitz spectra are the best known of<br />
,...<br />
the one-dimensional frequency spectra that have been used to describe<br />
ocean waves. The JONSWAP spectrum is a recent extension of the<br />
Bretschneider spectrum and has an additional term which may be used<br />
to give a spectrum with a sharper peak.<br />
A.3.1<br />
Bretschneider and ISSC Spectrum<br />
The Bretschneider (ReferenceA.2) spectrum and the spectrum proposed<br />
as a modified Pierson-Moskowitz spectrum by the Second<br />
International<br />
<strong>Ship</strong> <strong>Structure</strong>s Congress (Reference A.4) are identical.<br />
The<br />
Bretschneider equation in terms of radial frequency is as follows.<br />
S(U) = (5/16)*(Hs)2*(um”/w5)*exp [-1.25*(U/Mm)-’ ]<br />
where:<br />
A-n<br />
H~ is the significantwave height, and<br />
‘m<br />
is the<br />
frequency of maximum spectral energy.<br />
The Bretschneider equation<br />
may be written in terms of the peak period<br />
instead of the peak frequency, by substituting urn=2iT/T.<br />
P<br />
S(u) = (5/16)*(ti)2* [(21r)4/(u5*(T)4)]*<br />
exp[-l~25*(2~q/(~*T )4] p<br />
A.3.2<br />
Pierson-Moskowitz Spectrum<br />
The Pierson-Moskowitz (Reference A.4) spectrum was created to fit<br />
North Atlantic weather data. The P-M spectrum is the same as the<br />
Bretschneider spectrum, but with the H* and Mm dependence merged<br />
into a single parameter. The frequency used in the exponential has -<br />
also been made a function of reported wind speed. The equation for<br />
the Pierson-Moskowitz spectrum is as follows.<br />
S(u) = a*g2/~S)*exp[-~*(~0/~)4]<br />
where:<br />
a<br />
: 0.0081<br />
B<br />
= 0.74<br />
&l<br />
o<br />
= g/u<br />
and, U<br />
is the wind speed reported by the weather ships.<br />
The Pierson-Moskowitz spectrum equation may be obtained from the<br />
Bretschneider equation by using one of the following relations<br />
between H~ and Wm.<br />
A-12<br />
--;”,<br />
:“,< L.
H~ = 0.1610*g/(um)2<br />
‘m = o.4o125*g/(H#<br />
An interesting point that may be noted is that if B<br />
were set equal<br />
to 0.75 instead of 0.74, the w would be the frequency<br />
o<br />
corresponding to the modal period, Tm.<br />
A.3.3<br />
JONSWAP and Related Spectra<br />
The JONSWAP wave spectrum equation resulted from the Joint North Sea<br />
Wave Project (Reference A.5). The JONSWAP equation is the original<br />
Bretschneider wave spectrum equation with an extra term added. The<br />
extra term may be used to produce a sharply peaked spectrum with more<br />
energy near the peak frequency. The JONSWAP spectrum can be used to<br />
represent the Bretschneider wave spectrum, the original Pierson-<br />
Moskowitz wave spectrum, and the ISSC modified P-M spectrum. The<br />
full JONSWAP equation is as follows.<br />
s(w) = (aj*g2u5 ) *exp[-1.25*W/um-’)*ya<br />
where:<br />
a = exp [-**(M-Wm)2 / (U*WM)2]<br />
‘m<br />
is the frequency of maximum spectral energy.<br />
The Joint<br />
North Sea Wave Project recommended the following mean<br />
values to represent the North Sea wave spectra.<br />
Y = 3.3<br />
0 = 0.07, for W
a<br />
= 0.09, for UJ>wm<br />
The<br />
U<br />
value of Q is found by integrating the spectrum and adjusting<br />
to give the desired area.<br />
The<br />
Bretschneider equation and the ISSC equation can be obtained by<br />
setting the following parameter values.<br />
Y = 1.0<br />
a = (5/16 )*( Hs)2*(#/g2<br />
The Pierson-Moskowitz equation is obtained from the further<br />
restriction that Hs and Urn are related.<br />
H~ = 0.1610*g/(wm)2<br />
or<br />
‘m =<br />
0.140125* (g/H5)+<br />
or<br />
a = 0.0081<br />
When Y is set to one the JONSWAP term is effectively turned off.<br />
Without the JONSklAP term, the wave spectrum equation can be<br />
mathematically integrated to give the following relationships among<br />
the characteristicwave periods.<br />
Tp = 1.1362 *TM<br />
‘P<br />
= 1.2957 *TV<br />
‘P = 1.4077 *TZ A-14
‘P = 1.1671*TS .<br />
For Y = 1, the fourth area moment is infinite. The crest period,<br />
T~, is therefore zero.<br />
For values of y other than one, the JONSWAP equation cannot be<br />
mathematically integrated. The period relationships as a function<br />
of Ycan be calculated by numerical integration of the wave spectrum<br />
equation over the range from three-tenths of the peak frequency to<br />
ten times the peak frequency.<br />
The shape of the JONSWAP spectrum can be further adjusted by changing<br />
the values of e. The e values are sometimes varied when the JONSWAP<br />
spectrum is used to fit measured wave spectra.<br />
.-<br />
A.3.4<br />
Scott and Scott-Wieqel Spectra<br />
The Scott (Reference A.6) spectrum was also formulated to fit North<br />
Atlantic weather data. The Scott spectrum is the Derbyshire<br />
(Reference A.7) spectrum with S1ight modifications to the constants<br />
in the equation. The spectrum equation is as follows.<br />
s(w)<br />
= 0.214*(Hs)2*exp[-(~-wm)/ {0.065*(U-Mm+0.26)}4]<br />
for -0.26 < OJ-Wm < 1.65<br />
= o, elsewhere.<br />
where<br />
A-15<br />
2.3
H~ is the significant height,<br />
‘m = 3.15*T-l+&Wl*T-2,<br />
T<br />
is the characteristic period of the waves.<br />
The timis the frequency of the peak<br />
unfortunately, the period, T, used in the<br />
correspond to any of the mathematical<br />
spectrum. The equation for ~mwas derived<br />
data.<br />
spectral energy, but<br />
equation for ~mdoes not<br />
characteristics of the<br />
as a curve fit to real<br />
The Scott-Wiegel spectrum is a Scott spectrum modification that was<br />
proposed by Wiegel (Reference A.8). The constants are adjusted to<br />
match the equation to a “100-year storm” wave condition. The new<br />
equation is as follows.<br />
S(UJ) = 0.300* (H~)*exp[-(w-wm)4/<br />
{0.0.353* (M-WM+0.26) } ]<br />
The umin<br />
equation.<br />
this equation is 1.125 times that specified for the Scott<br />
A.4 SELECTING A WAVE SPECTRUM<br />
Information about the random sea characteristics in a particular area<br />
is derived by either ‘wave hindcasting’ or by direct wave<br />
measurement. For many areas of the world’s oceans, the only data<br />
available is measured wind speeds and visually estimated wave<br />
heights. Sometimes the estimated wave heights are supplemented by<br />
estimated wave periods. For afew areas of intense oil development,<br />
such as the North Sea, direct wave measurement projects have produced<br />
detailed wave spectra information.<br />
A-16
A.4.1<br />
Wave Hindcastinq<br />
Wave hindcasting is a term used to describe the process of estimating<br />
the random sea characteristics of an area based upon meteorological<br />
or wind data. Various researchers (ReferencesA.2, A.4, A.6, A.7 and<br />
A.8) have attempted to derive a relationship between the wind speed<br />
over a recent period of time and the spectrum of the random sea<br />
generated by the particular wind. The wind speed data is usually<br />
qualified by two additional parameters, the duration that the wind<br />
has been blowing at that speed and the fetch or distance over open<br />
ocean that the wind has been blowing.<br />
A set of equations as derived by Bretschneider (ReferenceA.2), which<br />
relate wind speed, duration and fetch are as follows.<br />
g*H#<br />
= 0.283*tanh[0..125*(g*F/Uz)””42]<br />
g*Ts / (2r*U) = 1.2*tanh[0.077*(gF/U2)””42<br />
...<br />
g*t minl” =<br />
6.5882*exp{[0.161*A2-0.3692*h+2.024]%<br />
+ O.8798*A}<br />
where<br />
u<br />
is the wind speed,<br />
F<br />
is the fetch,<br />
A<br />
= ln[g*F/U2],<br />
t min<br />
is the minimum duration for which the fetch will<br />
determine the significant height and period, and<br />
tanh<br />
is the hyperbolic tangent.<br />
A-17<br />
-) s“
If the wind duration is less than tmin, then the third equation is<br />
used to find the fetch which would correspond to tmin = t.<br />
For a fully arisen sea, the above equations simplify to the<br />
following.<br />
g*Hs/U2 = 0.283<br />
g*Ts/(2T*U) = 1.2<br />
Other relationships have been developed in the references. Often<br />
specialized weather/wave research companies have developed elaborate<br />
wave hindcasting models to derive the wave spectra characteristics<br />
for particular areas. However, the assumptions incorporated into<br />
these models have very profound impact on the outcome.<br />
A.4.2<br />
Direct Wave Measurements<br />
By installing a wave probe or a wave buoy in the ocean area of<br />
interest, wave elevation histories may be directly measured. The<br />
elevation of the sea at a particular point is either recorded by<br />
analog means or is sampled at short time intervals (typically one<br />
second) and recorded digitally. The wave elevations are usually<br />
recorded intermittently, ie. the recorder is turned on for say 30 min<br />
every four hours.<br />
The wave records are then reduced by computer, and the wave<br />
characteristics are summarized in various ways. Two common ways of<br />
summarizing the data are as a wave scatter diagram and/or as a wave<br />
height exceedance diagram.<br />
The wave scatter diagram is a grid with each cell containing the<br />
number or occurances of a particular significant wave height range<br />
and wave period range. The wave period range may be defined in terms<br />
of either peak period or zero-uncrossingperiod.<br />
A-la<br />
w
The wave height exceedance diagram is a curve showing the percentage<br />
of the wave records for which the significant wave height was greater<br />
that the particular height.<br />
M<br />
WAVE SCATTER DIAGRAM<br />
Wave scatter diagrams show the occurances of combinations of<br />
significant wave height and average zero-uncrossing period over an<br />
extended time period such as many years.<br />
Wave height distribution over time can be obtained by actual wave<br />
measurements. The heights and periods of all waves in a given<br />
direction are observed for short periods of time at regular<br />
intervals. A short time interval of several hours may be considered<br />
constant. For this sea state, defined as “stationary”,the mean zero<br />
up-crossing period, Tz, and the significant wave height, Hs, are<br />
calculated. The Hs and Tz pairs are ordered and their probabilities<br />
of occurance written in a matrix form, called a wave scatter diagram.<br />
Sometimes wave scatter diagrams are available for the sea and for the<br />
swell. The sea scatter diagram includes the sea spectra generated<br />
locally. The swell scatter diagram contains the swell spectra (or<br />
regular waves) generated far from the area, days before. Due to<br />
greater energy losses in high frequency waves and the continual phase<br />
shifting caused by viscosity, the energy in irregular seas tends to<br />
shift toward longer periods, and the spectra becomes more peaked as<br />
time passes. The energy in the swell is concentrated about a single<br />
long period/low frequency, and often the swell is treated as a single<br />
regular wave.<br />
A typical wave scatter diagram, presenting statistical data on the<br />
occurance of significant wave height and zero up-crossing period per<br />
wave direction is shown on Figure A-2.<br />
A-19
Sample Wave scatter Diagram<br />
s<br />
12 +..-..+---.-+-----+-w---+-----+-_---+-----+-----+-.---+.----+-.-_-+<br />
i 1111111 11111<br />
9 11 +---..+----.+-----+-..--+----.+-----+----.+-----+-.---+-.---+-----+<br />
n 111111 10.511.01 I I I<br />
i 10 +-----+-----+-..--+----.+.----+-..--+-----+-.---+-----+---..+----.+<br />
f 111[11 11.012.011.51 I I<br />
i 9 +----.+.----+...--+--...+-----+-.-.-+-----+..---+--..-+---.-+---..+<br />
c 1111 10.511.512.513.010.51 I I<br />
a 8 +-----+-----+-----+-----+-----+-----+-----+-----+-----+--.--+---.-+<br />
n 1111 11.015.015.512.510.51 I I<br />
t 7 +-----+-----+-----+-----+-----+-----+-----+-----+-----+---.-+-----+<br />
1111 I 5.0 113.0 111.0I 2.0 I I I I<br />
w 6 +-----+-----+-----+-----+-----+-----+-----+-----+-----+-----+-.---+<br />
a II I 0.5 I 6.0 118.0 123.0 I 8.5 I 1.0 I I I I<br />
v 5 +-----+-----+-----+-----+-----+-----+-----+-----+---.-+-----+-----+<br />
e II I 4.0 126.5 148.5 126.5 I 7.0 I 2.5 I 0.5 I 0.5 I I<br />
4 +-----+-----+-----+-----+-----+-----+-----+-----+-..--+-----+-----+<br />
H I 1 1.5 139.5 179.5 163.5 120.0 I 6.0 I 3.0 I 1.5 I 0.5 I 0.5 I<br />
e 3 +-----+-----+-----+-----+-----+-----+-----+-----+-----+---.-+-----+<br />
i I 0.5 150.0 I105.OI95.5 135.0 111.5 I 5.5 I 2.0 I 1.5 I I I<br />
—<br />
9 ,2+-----+-----+-----+.----+-----+-----+-----+-----+-----+.----+---..+<br />
h I 1.5 159.5 189.0 134.5 112.0 I 7.0 I 4,0 I 1.5 I 0.5 I I I<br />
t 1 +-----+-----+-----+-----+-----+-----+-----+-----+-.----+-----+-----+<br />
12.5118 .018.012.512.511.510.5 I I i I I<br />
(m) O +-----+-----+-----+-----+-----+-----+-----+-----+-----+-----+-----+<br />
2 3 4 5 6 7 8 9 10 11 12 13<br />
Zero Up-crossing Period, Tz (see)<br />
Sum of Occurances 999.5<br />
Figure A-2 A Typical Wave Scatter Diagram for the Central North Sea<br />
A-20
Using the significant wave height and zero up-crossing period from<br />
the wave scatter diagram and selecting a representative sea spectrum<br />
formulation, the energy of each sea state can be reconstructed.<br />
A*6 WAVE EXCEEDANCE CURVE<br />
A wave exceedance curve shows the number (percentage)of waves that<br />
are greater than a given wave height ‘for consistent wave height<br />
intervals. Table A-1 shows the type of data contained on a wave<br />
exceedance curve.<br />
Wave Height (ft)<br />
Cl<br />
5<br />
10<br />
15<br />
20<br />
25<br />
30<br />
35<br />
40<br />
Number of Waves<br />
35,351,396<br />
3,723,300<br />
393,887<br />
41,874<br />
4,471<br />
480<br />
51<br />
5<br />
1<br />
(N)<br />
Table A-1 Wave Exceedance Data for Campos Basin<br />
(Number of Waves from Northeast)<br />
This data can be plotted on semi-log paper and closely approximated<br />
by a straight line plot. Typically, a wave exceedance H-N curve can<br />
be defined with the following equation.<br />
H<br />
= Hm+mz * 109 f!h<br />
where<br />
Hm<br />
is the maximum wave height for the design life,<br />
‘z<br />
is the slope of the H-log N curve, -Hm/log Nh,<br />
MIC:500400CC-A<br />
A-21
Nm is the total number of waves in the design life, and<br />
Nh is the<br />
H.<br />
number of occurances of waves with height exceeding<br />
A.7 WAVE HISTOGRAM AND THE RAYLEIGH DISTRIBUTION<br />
Actual wave height measurements can be plotted to show the number of<br />
waves of a given height at equal wave height intervals. The<br />
histogram obtained can be defined by a simple curve.<br />
A simple curve that fits most wave histograms is the Rayleigh<br />
distribution. Past work have shown that the Rayleigh distribution<br />
often allows accurate description of observed wave height<br />
distributions over a short term.<br />
The Rayleigh distribution is typically given as,<br />
,...,<br />
p(Hi) = 2*Hi *EXP(-H12/~2) * (1/~2)<br />
where<br />
P(Hi)<br />
is the wave height percentage of occurances,<br />
Hi<br />
is the wave heights at constant increments,<br />
~ 2 is the average of all wave heights squared.<br />
A.8 EXTREMEVALUESAND THE WEIBULL DISTRIBUTION<br />
For design purposes an estimate of the maximum wave height (extreme<br />
value) is required. The Rayleigh distribution provides such an<br />
estimate over a short duration. However, in order to estimate the<br />
extreme wave that may occur in say 100 years, the Weibull<br />
distribution is often used.<br />
A-22
The equation for the Weibulldistribution is as follows.<br />
P(H) = 1 - EXP[ -( (H-E)/e )= ]<br />
where<br />
P(H)<br />
is the cumulative probability,<br />
H<br />
is the extreme height,<br />
E<br />
is the location parameter that locates one end of<br />
the density function,<br />
e<br />
is the scale parameter, and<br />
a<br />
is the shape parameter.<br />
By plotting the wave exceedance data on Weibull graph paper, the<br />
distribution can be fit with a straight line and the extreme value<br />
for any cumulative probability can be found by extrapolation.<br />
A.9<br />
WIND ENVIRONMENT<br />
The wind environment, source of most ocean waves, is random in<br />
nature. The wind speed, its profile and its directionality are<br />
therefore best described by probabilistic methods.<br />
A.9.1<br />
Air Turbulence, Surface Roughness and Wind Profile<br />
Air turbulence and wind speed characteristics are primarily<br />
influenced by the stability of the air layer and terrain. For<br />
extreme wind gusts the influence of stability is small, making<br />
A-23
turbulence largely a function of terrain roughness. In an ocean<br />
environment, the wave profi1e makes prediction of wind<br />
characteristics more difficult. As the wind speed increases, the<br />
wave height also increases, thereby increasing the surface<br />
roughness. A surface roughness parameter is used as a measure of the<br />
retarding effect of water surface on the wind speed.<br />
A simple relationship developed by Charnock (ReferenceA.9) is often<br />
used to define the surface roughness parameter and the frictional<br />
velocity in terms of mean wind speed. Further discussion on surface<br />
roughness parameter and drag factor is presented in an ESDU document<br />
(Reference A.10).<br />
Full scale experiments carried out by Bell and Shears (Reference<br />
All) may indicate that although turbulence will decay with the<br />
distance above sea surface, it may be reasonably constant to heights<br />
that are applicable for offshore structures.<br />
Considering that wind flow characteristics are primarily influenced<br />
by energy loss due to surface friction, the mean wind profile for an<br />
ocean environment may be assumed to be similar to that on land and to<br />
follow this power law:<br />
v mz<br />
= Vmzl (z/zL) u<br />
where:<br />
v mz<br />
= mean wind velocity at height z above LAT<br />
v mzl<br />
= mean wind velocity of reference height above<br />
LAT<br />
= height at point under consideration above<br />
IAT<br />
A-24
Z1 = reference height, 30 ft (10 M), above LAT<br />
(typical)<br />
u<br />
=<br />
height exponent, typically 0.13 to 0.15.<br />
A.9.2<br />
Applied, Mean and Cyclic Velocities<br />
The random wind velocity at height z can be thought of as a<br />
combination of time-averaged mean velocity, Vmz, and a time varying<br />
cyclic component, vz (t).<br />
Vz (t) = Vmz + v~ (t)<br />
A range of mean and associated cyclic wind speeds can be extracted<br />
from an anemogram and divided into one- to four-hour groups over<br />
which the cyclic component of the wind speed is approximately<br />
equal. By describing cyclic wind speeds associated with an average<br />
value of the mean component of the wind speed over a particular<br />
period of time, a number of pairs of mean and associated cyclic<br />
speeds can be obtained. In addition to the applied, mean and cyclic<br />
wind speeds shown on Table A-2, their probability of occurrence is<br />
necessary to generate a scatter diagram. If sufficient data are not<br />
available, the number of occurrences can be extrapolated based on<br />
similar data. Table A-2 is given only to illustrate the wind make-up<br />
and the uncertainties associated with wind data.<br />
A-25
Applied Wind<br />
Speed Vz(t)<br />
ft/s (m/s)<br />
Mean Wind<br />
Speed Vmz<br />
ft/s (m/s)<br />
.---------_<br />
Cyclic Wind Probability<br />
Speed vz(t) of<br />
ft/s (m/s) Occurrence %<br />
4.26 (13)<br />
78.7 (24)<br />
101.7 (31)<br />
134.5 (41)<br />
154.0 (50)<br />
180.4 (55)<br />
29.5 (9)<br />
62.3 (19)<br />
78.7 (24)<br />
101.7 (31)<br />
124.6 (38)<br />
131.2 (40)<br />
13.1 (4) 16.7<br />
16.4 (5) 45.8<br />
23.0 (7) 12.5<br />
32.8 (10) 16.7<br />
39.4 (12) 4.2<br />
49.2 (15) 4.2<br />
Table A-2 Applied, Mean<br />
Extreme Gust Environment<br />
and Cyclic Wind Speed Distribution for an<br />
A.9.3<br />
Gust Spectra<br />
The power spectral density function provides information on the<br />
energy content of fluctuating wind flow at each frequency<br />
component. A study of 90 strong winds over terrains of different<br />
roughness in the United States, Canada, Great Britain, and Australia<br />
at heights ranging from 25 feet (8m) to 500 feet (150m) allowed<br />
Davenport (Reference A.12) to propose a power density spectrum of<br />
along-wind gust (the longitudinal component of gust velocity).<br />
A modified version of the Davenport spectrum, due to Harris<br />
(Reference ??), is given by:<br />
m=qf<br />
kV;<br />
(2 + f’)$i’<br />
where:<br />
A-26
n<br />
= fluctuating frequency 2<br />
S(n)<br />
= power density [(m/see )/Hz]<br />
k<br />
= surface roughness drag factor corresponding<br />
to the mean velocity at 30 ft (lOm) (i.e.<br />
0.0015)<br />
v<br />
=<br />
mean hourly wind speed at 30 ft (lOm) m<br />
f<br />
=<br />
non-dimensionalfrequency (nL/V ) m<br />
L<br />
=<br />
length scale of turbulence ( 1200 to 1800m,<br />
typical)<br />
-,<br />
The Harris spectra may be used to develop the wind spectra for each<br />
one of the mean wind speeds associated with the scatter diagram.<br />
...<br />
A-27
A.1O<br />
REFERENCES<br />
A*I<br />
Mechanics of Wave Forces on Offshore <strong>Structure</strong>s, Sarpkaya,<br />
T. and Isaacson,M., Van Nostrand Reinhold Company, 1981.<br />
A.2<br />
Bretschneider, C.L., “Wave Variability and Wave Spectra for<br />
Wind-Generated Gravity Waves,” Beach Erosion Board, Corps<br />
of Engineers, Technical Memo No. 118, 1959, pp. 146- 180.<br />
A.3<br />
Report of <strong>Committee</strong> 1, “Environmental Conditions,”<br />
Proceedings Second International <strong>Ship</strong> <strong>Structure</strong>s Congress,<br />
Delft, Netherlands, July 1964, Vol. 1, pp. 1:5-1:13.<br />
A.4<br />
Pierson, W.J. and Moskowitz, L., “A Proposed Spectral Form<br />
for Fully Developed Wind Seas Based on the Similarity<br />
Theory of S.A. Kitaigordsky,” Journal of Geophysical<br />
Research, Vol. 69, No. 24, Dec 1964.<br />
A.5<br />
Rye, H., Byrd, R., and Torum, A., “Sharply Peaked Wave<br />
Energy Spectra in the North Sea,” Offshore Technology<br />
Conference, 1974, No. OTC 2107, pp. 739-744.<br />
A*6<br />
Scott, J.R., “A Sea Spectrum for Model Tests and Long-Term<br />
<strong>Ship</strong> Prediction,” Journal of <strong>Ship</strong> Research, Vol. 9, No. 3,<br />
Dec. 1965, pp. 145-152.<br />
A.7<br />
Derbyshire, J., “The One-Dimensional Wave Spectrum in the<br />
Atlantic Ocean and in Coastal Waters,” Proceedings of<br />
Conference on Ocean Wave Spectra, 1961, pp. 27-39.<br />
A.8<br />
Wiegel, R.L., “Design of Offshore <strong>Structure</strong>s Using Wave<br />
Spectra,” Proceedings Oceanology International, Brighton,<br />
England, 1975, pp. 233-243.<br />
A-28
‘GiiQH1<br />
REGULAR<br />
WAVES<br />
TI T2 T3<br />
I<br />
T4<br />
T5<br />
..-.<br />
SWL<br />
HI<br />
H3<br />
n /<br />
t<br />
IRREGULAR<br />
WAVES<br />
Figure A-I Regular and Irregular Waves
APPENDIX B<br />
REVIEW OF LINEAR SYSTEM RESPONSE<br />
TO RANDOM EXCITATION<br />
CONTENTS<br />
B. REVIEW OF LINEAR SYSTEM RESPONSE TO RANDOM EXCITATION<br />
B.1 GENERAL<br />
B.1.l<br />
B.1.2<br />
B.1.3<br />
Introduction<br />
Abstract<br />
Purpose<br />
B.2 RESPONSE TO RANDOM WAVES<br />
B.2.1 Spectrum Analysis Procedure<br />
8.2.2 Transfer Function<br />
6.2.2.1 Equations of Motions<br />
6.2.2.2 Response Amplitude Operator<br />
6.2.3 Wave Spectra<br />
6.2.3.1 Wave Slope Spectra<br />
6.2.3.2 Wave Spectra for Moving Vessels<br />
B.2.3.3 Short-Crested Seas<br />
B.2.4 Force Spectrum<br />
B.2.5 White Noise Spectrum<br />
B.3 EXTREME RESPONSE<br />
B.3.1<br />
B.3.2<br />
Maximum Wave Height Method<br />
Wave Spectrum Method<br />
B.4 OPERATIONAL RESPONSE<br />
B.4.1<br />
B.4.2<br />
Special Family Method<br />
Wave Spectrum Method
B. REVIEW OF LINEAR SYSTEM RESPONSE TO RANDOM EXCITATION<br />
B.1 GENERAL<br />
B.1.l<br />
Introduction<br />
Spectral analysis is used to determine the response of linear systems<br />
to random excitation. In the case of offshore structures, the random<br />
excitation comes from either irregular waves or winds. Typical<br />
offshore systems subjected to spectral analysis include ships,<br />
semisubmersibles, jack-ups, tension-leg platforms and bottomsupported<br />
fixed platforms. Responses of interest include motions,<br />
accelerations, and member internal forces, moments, and stresses.<br />
Floating units are evaluated by spectral analysis for motions in<br />
random seas. The strength and structural fatigue integrity are often<br />
assessed with spectral analysis.<br />
B.1.2<br />
Abstract<br />
A spectral analysis combines a set of regular wave response amplitude<br />
operators, RAOS, with a sea spectrum to produce a response<br />
spectrum. Characteristics of the response may be calculated from the<br />
response spectrum, and a random sea transfer function can be derived.<br />
For certain spectral analyses, the sea spectrum-must be modified to<br />
produce a wave slope spectrum or to adjust the sea spectrum for<br />
vessel speed. A spreading function can be applied to the sea<br />
spectrum to model a short-crested random sea.<br />
A wave force spectrum can be created directly from force RAOS and the<br />
sea spectrum.<br />
A regular wave transfer function is found as the solution to the<br />
equations of motion. The regular wave transfer function can be.<br />
expressed in terms of RAO and a phase angle.<br />
B-1
Awhite noise function-may be used to represent a very broad banded<br />
input spectrum, if the response spectrum is narrow banded.<br />
The extreme response can be calculated from a given extreme wave, or<br />
the extreme response may be statistically derived from a set of<br />
spectral analyses.<br />
The sea spectra used in the computation of random sea response can be<br />
reduced in number by selecting a smaller family of representative<br />
spectra, or by creating a set of mean spectra.<br />
0.1.3 m<br />
The purpose of this appendix is to provide a background of the<br />
spectral analysis method and to clarify the concept of a response<br />
spectrum and how its properties are derived.<br />
P,<br />
B.2 RESPONSE TO RANDOM WAVES<br />
The spectral analysis method is a means of taking the known response<br />
of an offshore structure to regular waves and determining the<br />
structure’s response to a random sea. The input to the spectral<br />
analysis method is the response amplitude per unit wave amplitude (or<br />
equally, the response double amplitude per unit wave height) for a<br />
range of periods or frequencies of regular waves. These ratios of<br />
response amplitude to wave amplitude are known as “Response Amplitude<br />
Operators” or just “RAOS.” The response of the offshore structure is<br />
first obtained for a set of unit amplitude, regular, sinusiodal<br />
waves. The regular wave response may be obtained either from model<br />
tests or from empirical or theoretical analyses.<br />
A wave energy spectrum is selected to represent the random sea. Wave<br />
spectra are described in Appendix A. The wave spectrum represents<br />
the distribution of the random sea’s energy among an infinite set of<br />
regular waves that when added together create the random character of<br />
B-2 /l-l-
the sea. By assuming thatthe response is linear, the response of<br />
the offshore structure to a regular wave is equal to the RAO times<br />
the regular wave amplitude. By assuming that the response to one<br />
wave does not affect the response to another wave, the response of<br />
the offshore structure to a random sea is the sum of its responses to<br />
each of the constituent regular waves in the random sea. The<br />
response is therefore a collection of responses each with a different<br />
amplitude, frequency, and phase.<br />
The energy of each constituent wave is proportional to the wave<br />
amplitude squared. The energy of the response to a constituentwave<br />
of the random sea is proportional to the response squared, or is<br />
proportional to the RAO squared times the wave amplitude squared.<br />
The response energy may also be represented by a spectrum from which<br />
characteristics of the response may be derived. From the response<br />
spectrum characteristics and the wave spectrum characteristics, a<br />
“transfer function” can be obtained which relates the response and<br />
wave characteristics.<br />
B.2.1<br />
Spectrum Analysis Procedure<br />
The spectral analysis procedure involves four steps: 1) obtaining<br />
the response amplitude operators, 2) multiplying the wave spectrum<br />
ordinates by the RAOS squared to get the response spectrum, 3)<br />
calculating the response spectrum characteristics, and 4) using the<br />
response spectrum characteristics to compute the random sea response<br />
transfer function.<br />
The RAOS are usually calculated for a discrete set of wave<br />
frequencies, and the discrete RAOS are then fit with a curve to<br />
produce a continuous function. The singular term “RAt)” is used both<br />
to signify a single response amplitude to wave amplitude ratio and to<br />
signify the continuous function through all of the RAOS. Any<br />
response that is linearly related (proportional) to wave amplitude<br />
may be reduced to an RAO function. Typical responses are motions,<br />
accelerations, bending moments, shears, stresses, etc.<br />
B-3
Multiplication of the wave spectrum ordinates by the RAO squared is<br />
simple. The two underlying assumptions are that the response varies<br />
linearly with wave amplitude and the assumption that the response to<br />
a wave of one frequency is independent of the response to waves of<br />
other frequencies.<br />
Response spectrum characteristics are taken from the shape of the<br />
spectrum or are calculated from the area under the response spectrum<br />
and the area moments of the response spectrum. Typical<br />
characteristics are significant response amplitude, maximum response<br />
amplitude, mean period of the response, and peak period of the<br />
response spectrum.<br />
The random sea transfer function is the ratio of a response spectrum<br />
characteristic to a wave spectrum characteristic. A random sea<br />
transfer function is usually presented as a function of the random<br />
sea characteristic period. A typical transfer function might be the<br />
ratio of maximum bending moment amplitude per unit significant wave<br />
height. The transfer function is useful for estimating the response<br />
to another wave spectrum with similar form but different amplitude.<br />
.<br />
B.2.2<br />
Transfer Function<br />
A transfer function converts input to output for linear systems. A<br />
transfer function is graphically represented in Figure B-1. A<br />
transfer function can relate motion response to the height of<br />
incident waves directly, or a transfer function can relate motion<br />
response to wave force, or a transfer function can relate member<br />
stresses to wave or wind force.<br />
For typical applications to the design of offshore structures, the<br />
input energy forms are waves, current and wind. The desired output<br />
forms are static displacements, dynamic displacements, and member<br />
stresses.<br />
B-4
B.2.2.1 Equation of Motions<br />
By assuming that the motions are small enough that the inertial,<br />
damping and spring forces can be summed linearly, the equation of<br />
motion can be formulated.<br />
M*X + D*X + K*X = F(x,t)<br />
where M is the mass matrix which includes the structure mass<br />
properties plus the hydrodynamic added mass effects,<br />
D<br />
is the linearized damping matrix which includes the<br />
viscous damping, the wave damping, and the structural<br />
damping effects,<br />
-.><br />
K<br />
is the stiffness matrix which includes the waterplane<br />
spring properties,<br />
moorings or tendons,<br />
the structure and any<br />
the restoring properties of<br />
and the stiffness properties of<br />
foundation,<br />
x<br />
is the system displacement vector,<br />
i<br />
is the system velocity vector = (dx/dt),<br />
—<br />
x<br />
is the system acceleration vector, = (d2x/dt2), and<br />
F<br />
is the force vector which may be calculated from<br />
empirical methods such as Morrison’s equation or from<br />
diffraction theory methods.<br />
The equations of motion can be solved with frequency domain or time<br />
domain techniques. The frequency domain solution involves the<br />
methods of harmonic analysis or the methods of Laplace and Fourier<br />
transforms. The time domain solution involves the numerical solution<br />
by a time step simulation of the motion.<br />
B-5
B.2.2.2 Response Amplitude Operator<br />
The solution of the equations of motion result in a transfer<br />
function. The motion transfer function has an in-phase component and<br />
an out-of-phase component. The transfer function is usually<br />
represented in complex form,<br />
x(u)<br />
= A*[XI(UJ) + i*XO(U)]<br />
or in angular form,<br />
x(u)= A*[XI*cos(Ut) +XO*sin(mt)]<br />
where<br />
x<br />
is the total response,<br />
A<br />
is the wave height,<br />
XI<br />
is the in-phase component of the response for unit<br />
wave height, and<br />
Xo<br />
is the out-of-phase component of the response for unit<br />
wave height.<br />
From this equation, the response amplitude operator (amplitude per<br />
unit wave height), is found to be,<br />
RAO = SQRT (X12 +X02),<br />
and the phase of the harmonic response relative to the wave is,<br />
o = ATAN (XO/XI).<br />
The response can be written in terms of the RAO and phase as,<br />
B-6<br />
,. I
X(w) = A*RAO(w)*cos(wt+ O(W)).<br />
When a spectral analysis is applied to the transfer function the<br />
wave amplitudes, A, become a function of wave frequency, u, and<br />
the X(U) is replaced by the differential slice of the response<br />
power density spectrum.<br />
SR(m)*dm = [A(W)*RAO(U)]2<br />
or<br />
SR(w)*dw = A2(u)*RA02(w)<br />
or<br />
SR(~)*dw = S(w)*du*RA02(w)<br />
Thus, Sf(w) = S(U)*RA02(W)<br />
The<br />
the<br />
response spectrum S(W)<br />
RAO squared.<br />
is therefore just the sea spectrum times<br />
For multiple-degree-of-freedom systems, there is coupling between<br />
some of the motions, such as pitch and heave. For example, to obtain<br />
the motion or motion RAO for heave of a point distant from the center<br />
of pitch rotation, the pitch times rotation arm must be added to the<br />
structure heave. This addition must be added with proper<br />
consideration of the relative phase angles of the pitch and heave<br />
motions, and therefore, such addition must be performed at the<br />
regular wave analysis stage. The combined heave (w/pitch) RAO can<br />
then be used in a spectral analysis to obtain the heave spectrum and<br />
heave response characteristicsat the point.<br />
6-7
0.2.3 Wave Spectra<br />
The wave spectrum used in the spectral analysis may be an idealized<br />
mathematical spectrum or a set of data points derived from the<br />
measurement of real waves. When a set of data points are used, a<br />
linear or higher order curve fit is employed to create a continuous<br />
function. Custom wave spectra for specific regions are often<br />
provided as one of the conventional idealized spectra with parameter<br />
values selected to match a set of measured wave data. For areas<br />
where there is little wave data, wave height characteristics are<br />
estimated from wind speed records from the general area.<br />
B.2.3.1 Wave Slope Spectra<br />
For certain responses, particularly the angular motions of pitch and<br />
roll, the RAO is often presented as response angle per unit wave<br />
slope angle. For these cases the wave spectrum in amplitude squared<br />
must be converted to a wave slope spectrum. The maximum slope of any<br />
constituent wave of the spectrum is assumed to be small enough that<br />
the wave slope angle in radians is approximately equal to the tangent<br />
of the wave slope. The water depth is assumed to be deep enough (at<br />
least one-half the longest wave length) that the wave length is<br />
approximately equal to:<br />
(g/2~)*T2 or 2~g/~2.<br />
By using the Fourier series representation of the wave spectrum,<br />
selecting one constituent wave, and expressing the wave equation in<br />
spatial terms instead of temporal terms, the wave slope is derived as<br />
follows.<br />
rI=a*cos(2rx/L) = a*cos(x~2/g)<br />
dq/dx = -(a~2/g)*sin(x~2/g)<br />
B-8<br />
LJt Y
[dn/dx]max= aW2/g .<br />
Squaring the equation to<br />
get the slope squared,<br />
[dn/dx]2 = a2*(~4/g2)<br />
Therefore, the wave spectrum equation must be multiplied by (~4/g2)<br />
to obtain the slope spectrum. The wave slope angle spectrum is the<br />
wave slope spectrum converted to degrees squared, i.e., multiplied by<br />
(180/r)2.<br />
B.2.3.2 Wave Spectra for Moving Vessels<br />
For self-propelled vessels or structures under tow, the forward speed<br />
of the vessel or structure will have an effect upon the apparent<br />
frequency of the waves. The apparent frequency of the waves is<br />
usually referred to as the encounter frequency. For a vessel heading<br />
into the waves the encounter frequency is higher than the wave<br />
frequency seen by a stationary structure. For a vessel moving in the<br />
same direction as the waves, the encounter frequency is less than the<br />
wave frequency seen by a fixed structure, and if the vessel’s speed<br />
is great enough it may be overrunning some of the shorter waves which<br />
will give the appearance that these shorter waves are coming from<br />
ahead instead of from behind.<br />
The encounter frequency for a regular wave is given by the following<br />
relationship.<br />
‘e = u + VuZ/g<br />
where u is the wave frequen~y in radians per second as seen<br />
from a stationary observer,<br />
v is the<br />
velocity component parallel to and opposite in<br />
direction to the wave direction, and<br />
%-9
9 is the acceleration of gravity in units compatible<br />
with .thevelocity units. .’<br />
The energy of, or area under the curve of the sea spectrum must<br />
remain constant.<br />
f$e(@*dwe = fS(w)*du<br />
Taking the derivative of the encounter frequency equation gives the<br />
following.<br />
dwe = [1 + 2Vw/g]*dw<br />
Substituting the derivative<br />
following.<br />
nto the area ntegral gives the<br />
f$e(ue)*[l +2Vw/g]*dw = ~S(w)*dw<br />
Therefore, equating the integrands gives the relationship between the<br />
encounter spectrum and the stationary sea spectrum.<br />
Se(we) = s(l.u)/[l+2vw/gl<br />
This equation is required to transform a stationary sea spectrum to<br />
an encounter spectrum for the purpose of intergrating the responses.<br />
However, if only the response statistics are desired, and not the<br />
actual response spectrum, then the same substitutions as above can be<br />
made.<br />
se = s/[1 +2vw/g]<br />
dwe = [1+2VU /g]*du<br />
B-10<br />
@
Jr 2*$ *dti= Jr 2*S*du<br />
eeee.<br />
Therefore, the encounter frequency need only be used to select the<br />
response amplitude operator and the integration is still over the<br />
stationary frequency, u.<br />
i.e.,<br />
re = r(we) = r(w +<br />
2<br />
VU/g)<br />
B.2.3.3 Short-Crested Seas<br />
The usual mathematical representation of a sea spectrum is onedimensional<br />
with the random waves traveling in a single direction<br />
with the crests and troughs of the waves extending to infinity on<br />
either side of the direction of wave travel. A one-dimensional<br />
irregular sea is also referred to as a long-crested irregular sea.<br />
In the real ocean the waves tend to be short-crested due to the<br />
interactionof waves from different directions.<br />
A two-dimensional spectrum (short-crested sea) is created from a<br />
standard one-dimensional mathematical spectrum by multiplying the<br />
spectrum by a “spreading function.” The most commonly used spreading<br />
function is the “cosine-squared”function.<br />
f(lp)= (2/Tr)*cos2$<br />
where * is the angle away from the general wave heading,<br />
(-lT/2q%/2)<br />
The cosine-squared spreading function spreads the sea spectrum over<br />
an angle +/- 90 degrees from the general wave heading.<br />
To incorporatemulti-directional or short-crested irregular seas into<br />
a spectral analysis, the RAOS for a range of wave headings must be<br />
obtained. A spectral analysis is performed for each heading using<br />
the one-dimensional sea spectrum. The results of the one-dimensional<br />
analyses are then multiplied by integration factors and summed.<br />
B-n
The following is..a. sample...of..a“.set. of...heading angles<br />
integration factors for a cosine squared spreading function.<br />
and the<br />
$ Factor<br />
~o<br />
0.2200<br />
*2O0 0.1945<br />
*4O0 0.1300<br />
~600 0.0567<br />
3800 0 ● 0088<br />
6.2.4 Force Spectrum<br />
For simple single-degree-of-freedomsystems,<br />
generated directly from the calculated or<br />
forces.<br />
a force spectrum can be<br />
measured regular wave<br />
The force on the structure is calculated by empirical or theoretical<br />
methods, or is derived by analyzing measured strain records from<br />
tests on the structure or on a model of the structure. This force is<br />
the right hand side of the equation of motion as described in Section<br />
B.2.2.1.<br />
The force itself has an in-phase and an out-of-phase component<br />
relative to the regular wave which generates the force. The force<br />
can be written in complex form,<br />
F(u) = A*[FI(u) + i*FO(m)]<br />
or in force RAO and phase form,<br />
F(w)<br />
= A*RAOf(w)*cos(wt+$(w))<br />
where<br />
RAOf = SQRT (F12<br />
+ F02), and<br />
@ ‘ATAN (FO/FI).<br />
B–12
The force..spectrum can be created by multiplying a selected wave<br />
spectrum times the force RAO squared.<br />
Sf(M) = S(W)*RAOf2(U)<br />
B.2.5<br />
White Noise Spectrum<br />
Most sea spectra have a well defined peak of energy and the energy<br />
trails off to near zero away from the peak. Other environmental<br />
inputs that are described by spectra, such as wind force, may not<br />
have a definite peak and may even appear constant over a wide range<br />
(broad band) of frequencies.<br />
Often the response RAO is narrow banded, that is, the structure tends<br />
to respond at a narrow range of frequencies, centered about a<br />
resonant frequency. When the combination of a broad banded<br />
excitation spectrum and a narrow banded RAO exist, the spectral<br />
analysis can be greatly simplified.<br />
A broad banded spectrum can be approximated by a “white noise<br />
spectrum” which has constant energy over the whole frequency range of<br />
the spectrum.<br />
—<br />
For a single degree of freedom system, the response can be defined in<br />
terms of a “dynamic amplification function” times an expected static<br />
displacement. The dynamic amplification function is as follows,<br />
where<br />
IH(w)I= l/[(l-w/un)2)2 + (2gu/un)2]%<br />
u) is radial frequency,<br />
u<br />
is the undamped “natural frequency”,<br />
% = (k/m)%, B-13
is the damping ratio,..the ratio of.the actual damping to<br />
the critical damping. g = c/(4km)$,<br />
k<br />
is the spring constant,<br />
m<br />
is the mass that is in motion, and<br />
c<br />
is the actual damping.<br />
The expected static displacement is simply force divided by the<br />
spring constant, or the expected static displacement spectrum is as<br />
follows,<br />
Sd(d ‘Sf(u)/k2<br />
From<br />
these equations, the response<br />
spectrum is found to be,<br />
R(u) = (1/k2)*lH(w)12*Sf(~),<br />
and the mean squared response<br />
is,<br />
y2(t) =O~m(l/k)2*lH(w)12*Sf(m)*dw.<br />
The (l/k)z is constant, and by approximating the force spectrum by a<br />
white noise spectrum with magnitude Sf(wn),the mean squared response<br />
is simplified to,<br />
Y2(t) = (Sf(u)n/k2)*O~m lH(w)12*dm.<br />
For lightly damped systems, (&
0.3 EXTREME RESPONSE<br />
The extreme response of an offshore structure may be determined in<br />
two ways. An extreme environmental event may be selected, and the<br />
responses to the extreme event then calculated. A set of<br />
environmental spectra can be selected; the response spectra to each<br />
environmental spectra calculated; and the extreme responses derived<br />
by statistical analysis of the response spectra. The first method is<br />
often called a “deterministic” method, and the second method is<br />
referred to as a “probabilistic”method. In actual design practice<br />
the two methods are often intermixed or combined in order to confirm<br />
that the extreme response has been found.<br />
B.3.1<br />
Maximum Wave Height Method<br />
In deterministic design, a set of extreme conditions is supplied by<br />
oceanographers or meteorologists. The extreme conditions are of<br />
course derived from statistical analyses of wave and weather records,<br />
but the design engineer is usually not involved in that stage of the<br />
calculations.<br />
,..,,<br />
The given extreme conditions are applied to the offshore structure to<br />
determine the various responses. Unfortunately, the given extreme<br />
conditions may not always produce the extreme responses. For<br />
example, the prying and racking loads governing the design of many<br />
structural members of semisubmersibles are typically maximized in<br />
waves with lower heights and shorter lengths then the maximum height<br />
wave. Tendon loads on tension leg platforms (TLPs) are also often<br />
maximized in waves that are lower and shorter than the maximum wave.<br />
Since the oceanographer or meteorologistwho produced the set of<br />
extreme conditions does not have information about the<br />
characteristics of the offshore structure, he/she is unable to select<br />
an extreme or near extreme condition that will produce the greatest<br />
response. Conversely, the design engineer usually has little or no<br />
information about the wave and weather data that was used to derive<br />
B-15
the set of extreme conditions, and thus, he/she is unable to create<br />
alternate conditions to.check for greater response.<br />
The design engineer may request a range of extreme conditions, such<br />
as: the maximum height wave with a period of 9 see, the maximum<br />
height wave with a period of 10 see, etc. The increased number of<br />
conditions increases the number of analyses required, but allows the<br />
design engineer to confirm which conditions produce the extreme<br />
responses.<br />
The maximum wave height method is best used when<br />
highly nonlinear and the spectral analysis method<br />
appropriate.<br />
the response is<br />
is therefore not<br />
B.3.2 Wave Spectrum Method<br />
A full probabilistic analysis involves calculating responses to the<br />
entire suite of possible environmental conditions. Statistical<br />
analysis of these responses is then performed In order to predict a<br />
suitable extreme for each response. This requires far fewer<br />
assumptions on the part of those who supply environmental criteria,<br />
but a much more extensive set of environmental data.<br />
/ -.<br />
With the wave spectrum method, a set of wave spectra are provided by<br />
oceanographers or meteorologists. The RAOS for the response of<br />
interest are squared and multiplied by the wave spectrum. A wave<br />
spectrum is assumed to represent a Gaussian random distribution.<br />
Since the response spectrum is created by a linear multiplication,<br />
the response spectrum also represents a Gaussian random<br />
distribution.<br />
be calculated<br />
maximum, etc. wave heights.<br />
The significant response, maximum response, etc. can<br />
using the equations for calculating the significant,<br />
The equations<br />
response:<br />
for maximum wave height are summarized here in terms of<br />
B-16
Significant response, (DA):<br />
R5 = 4.00*(mO)~<br />
Maximum response in 1000 cycles, (DA):<br />
R1/looo =<br />
7.43*(mO)~ = 1.86*Rs<br />
Maximum response is t hours, (DA):<br />
R max<br />
= 2*[2*mo*ln(3600*t/Tz)]$<br />
where m. is the area under the response<br />
spectrum,<br />
Tz is the zero-up-crossing period of<br />
from the equation,<br />
the response as found<br />
Tz = 2~*(mo/m2)4, and<br />
m2<br />
is the second radial frequency moment of<br />
the response spectrum.<br />
B.4 OPERATIONAL RESPONSE<br />
In order to determine the normal day-to-day motions and stresses to<br />
assess motion related downtime and fatigue damage, the distribution<br />
of wave heights versus wave periods are considered. A wave scatter<br />
diagram condenses and summarizes wave height and wave period<br />
statistics. It is a two-parameter probability density function.<br />
Typically a wave scatter diagram is presented as a grid of boxes,<br />
with one axis of the grid being average zero-up-crossing periods and<br />
the other axis being significant wave heights. Within the boxes of<br />
the wave scatter diagram are numbers which represent the percentage<br />
of the sea records having the corresponding characteristics of Hs and<br />
Tz see Figure A-2.<br />
B-17
A response scatter diagram could be generated by taking the wave<br />
spectrum.for each sample used to.create the.wave scatter diagram and<br />
performing a spectral analysis for the response. The computed<br />
characteristics are then used to assign the percentage of occurrence<br />
to the appropriate box in the response scatter diagram. This entails<br />
considerablework and can be simplified by reducing the number of sea<br />
spectra considered, as described below.<br />
B.4.1<br />
Special Family Method<br />
All of the original sea spectra used to define the wave scatter<br />
diagram must be available, in order to select a special family of sea<br />
spectra to represent the whole population.<br />
The sea spectra are first grouped by wave height bands, such as O to<br />
2 ft significant wave height, 2 ft to 4 ft H~, etc. The average<br />
properties of the spectra within a group are computed. Within each<br />
group, which may contain thousands of sample sea spectra, a small set<br />
of sea spectra are selected to represent all of the spectra in the<br />
group. The small set will typically contain 4 to 10 spectra.<br />
The spectra of a representative set are selected by a Monte Carlo<br />
(Shotgun) process which randomly picks, say 8, spectra from the<br />
group. The mean spectrum and the standard deviation of the spectral<br />
ordinates about the mean spectrum are computed for the 8 spectra. A<br />
weighted sum of differences in properties between the 8 spectra and<br />
the total population of the group represent the “goodness of fit” of<br />
that set of 8 spectra.<br />
A second representative set of 8 spectra is then selected from the<br />
group, and the “goodness of fit” of the second set is computed. The<br />
better set (first or second) is retained and compared to a third<br />
sample of 8, etc. The process is repeated many times, say 1000,<br />
within each wave group.<br />
R-18
From this process, the original number of sea spectra, which may have<br />
been thousands, is reduced to the number of wave height bands times<br />
the number of spectra in each representative set.<br />
B.4.2<br />
Wave Spectrum Method<br />
A reduced set of sea spectra can be generated to represent the<br />
variation of Hs and Tz as given in a wave scatter diagram.<br />
If the original sea spectra are not available, a set of sea spectra<br />
can be created directly from the wave scatter diagram. In this case<br />
the shape of the spectrum must be assumed. For various areas of the<br />
world’s oceans, preferred mathematical spectrum equations exist. For<br />
the,North Sea, the mean JONSWAP spectrum is preferred. For open<br />
ocean, the Bretschneider (ISSC) spectrum is preferred. For the Gulf<br />
of Mexico, the Scott spectrum has been recommended.<br />
Using the Hs and Tz for each populated box in the wave scatter<br />
diagram, and the selected sea spectrum equation, a set of wave<br />
spectra are defined. With this method the number of sea spectra is<br />
reduced to the number of populated boxes in the wave scatter diagram,<br />
but no more than the number of wave height bands times the number of<br />
wave period band.<br />
B–19
2.4<br />
2.2<br />
2<br />
1<br />
‘Sen &-”Swull$p~ctru<br />
1.8<br />
SEA<br />
1.6<br />
1,4 M’<br />
1-<br />
SWELL<br />
1.2-<br />
0.8-<br />
0.6-<br />
I<br />
/<br />
0.4- #----<br />
0,2-<br />
0 I 1 # i 1 I i 1<br />
0.2 0,8 1 1.4 1.8 2,2 2.5 3 3.4<br />
RodialFrequency(md~see)<br />
z<br />
I<br />
E<br />
1.2<br />
I<br />
1.1 -<br />
1-<br />
0.9-<br />
1<br />
Iicspmse A-nplitudcOpcrotur<br />
0.8-<br />
0.7-<br />
0.6-<br />
0,5-<br />
0.4-<br />
0.3-<br />
0.2-<br />
0.1 -<br />
0<br />
0.2 0.8 1 1.4 1.8 2.2 2,8 3 3.4<br />
Radial Frcqurwmy<br />
(md/see)<br />
0.9<br />
I<br />
Rtspcmse<br />
Spectrum<br />
k<br />
0.8-<br />
0.7-<br />
0.6-<br />
0.5-<br />
0.4-<br />
0.3-<br />
0.2-<br />
0.1 -<br />
0 # 1 1 # 1 1 [ 1 I 1 i !<br />
0.2 0.8 1 1.4 1,8 2.2 2.s 3 3.4<br />
Figure<br />
Rsdbl Frsqusney (rod/%w)<br />
B-1 Sea Spectra, Response Amplitude<br />
and Response Spectrum<br />
Operator (RAO)<br />
&
APPENDIX C<br />
STRESS CONCENTRATIONFACTORS<br />
CONTENTS<br />
c. STRESS CONCENTRATION FACTORS<br />
C*1 OVERVIEW<br />
C.1.l Objectives and Scope<br />
C*l.2 Current Technology<br />
C.2 STRESS CONCENTRATION FACTOR EQUATIONS<br />
C*2.1<br />
C.2.2<br />
Kuang with Marshall Reduction<br />
Smedley-Wordsworth<br />
C.3 PARAMETRIC STUDY RESULTS<br />
C.3.1<br />
C.3.2<br />
Ffigures<br />
Tables<br />
C.4 FINITE ELEMENT ANALYSES RESULTS<br />
C.4.1<br />
Column-GirderConnection<br />
C.5 REFERENCES
L ‘2
c. STRESSCONCENTRATIONFACTORS.<br />
C.1 OVERVIEW<br />
C.1.l<br />
Objectives and Scope<br />
A comprehensive document on stress Concentration factors (SCF) would<br />
include assessment of test results, detailed review of empirical<br />
equations, evaluation of finite element studies, and presentation of<br />
parametric studies showing the sensitivities of parameters affecting<br />
SCFS.<br />
The objective of this appendix is limited. Following a brief<br />
discussion of empirical equations, parametric study results are<br />
presented to assist the engineer in avoiding undesirable joint<br />
details. The sensitivity and interaction of variables shown in<br />
tables and figures also allow quick assessment of steps necessary to<br />
improve other geometries.<br />
Empirical formulations are applicable to a limited range of simple<br />
joint geometries. A complex joint often requires carrying out of a<br />
finite element analyses (FEA) to determine the SCFS. The results of<br />
a FEA is also presented to illustratethe applicable SCFS for a given<br />
geometry.<br />
—<br />
C1.2 Current Technology<br />
The SCF values can be computed through the use of a number of<br />
alternative equations. These equations have been mostly based on<br />
analytical (finite element) and small-scale experimental (acrylic<br />
model test) work. The tests carried out on joints that reflect those<br />
in-service (i.e. both in size and fabrication methods) are few and<br />
limited to several simple joint configurations. Thus, the equations<br />
available should be reviewed carefully to ascertain their range of<br />
validity and overall reliability prior to their use in design.<br />
Considering the simple joint configurations of T, Y, DT, K and X, the<br />
equations available for use in design are:<br />
c-l
o Kuang (ReferenceCl) “<br />
o Wordsworth (ReferencesC.2, C.3)<br />
o Gibstein (ReferencesC.4, C.5)<br />
o Efthymiou (ReferenceC.6)<br />
o Marshall (ReferenceC.7)<br />
o UEG (ReferenceC.8)<br />
There are significant differences in the validity ranges of these<br />
equations. The SCFS computed based on different equations also often<br />
vary considerably. The Kuang equations are applicable to T, Y, and K<br />
joints for various load types. Wordsworth and Wordsworth/Smedley<br />
equations are applicable to all simple joints. Gibstein equations<br />
are applicable to T joints while the Efthymiou equations cover T/Y<br />
joints and simple/overlappingK/YT joints. The equations proposed by<br />
Marshall are applicable to simple joints, based on those equations by<br />
Kellogg (Reference C.9), and were incorporatedinto API RP 2A.<br />
Substantial work has been carried out to validate the applicability<br />
of various SCF equations. Although some of the work carried out by<br />
major oil companies are unpublished, such work still influence ongoing<br />
analytical and experimental research. Delft von O.R.V. et al.<br />
(Reference C-10) indicate that the UEG equations offer a good<br />
combination of accuracy and conservatism while the Efthymiou (i.e.,<br />
Shell-SIPM) equations show a good comparison with experimental<br />
data.<br />
Ma and Tebbet (Reference C.11) report that there Is no consensus on<br />
whether a design SCF should represent a mean, lower bound or some<br />
other level of confidence. Tebbett and Lalani’s (Reference C.12)<br />
work on reliability aspects of SCF equations indicate that SCF<br />
equations underpredicting the 5CF values in less than 16% of the<br />
cases can be considered reliable. Thus, when presenting the findings<br />
of 45 elastic tests carried out on 15 tubular joints representing<br />
typical construction, Ma and Tebbet report that Wordsworth, UEG and<br />
Efthymiou equations meet this criteria and offer the best<br />
reliability.<br />
c-2
Ma and Tebbett also state that while both UEG and Wordsworth<br />
equations overpredict X joint SCFS, none of the equations overpredict<br />
the K joint SCFS. The comparative data indicate that the SCFS<br />
computed using Kuang and Gibstein equations for T/Y joints subjected<br />
to axial loading under predict the measured data in more than 16% of<br />
the cases. (See Figure Cl-l).<br />
Tolloczko and Lalani (ReferenceC.13) have reviewed all available new<br />
test data and conclude that reliability trends described earlier for<br />
simple joints remain valid and also state that Efthymiou equations<br />
accurately predict the SCFS for overlapping joints.<br />
c-3
C.2 STRESS CONCENTRATION FACTOR EQUATIONS<br />
C.2.1<br />
Kuang with Marshall Reduction<br />
The Kuang stress Concentration factor equations for simple<br />
unstiffened joints are shown on the following page. The brace stress<br />
Concentration factor equations include Marshall reduction factor,<br />
Qr. The validity ranges for the Kuang stress Concentration factor<br />
equations are:<br />
Term<br />
d/D<br />
T/l)<br />
t/T<br />
9/D<br />
D/L<br />
6<br />
Validity Range<br />
0.13 - 1.0<br />
0.015 - 0.06<br />
0.20 - 0.80<br />
0.04- 1.0<br />
0.05 - 0.3<br />
25-90<br />
.<br />
where,<br />
D = chord diameter<br />
T = chord thickness<br />
d = brace diameter<br />
t = brace thickness<br />
9 = gap between adjacent braces<br />
L = chord length<br />
e = angle between brace and chord<br />
c-4
C.2.2<br />
Smedley-Wordsworth<br />
The Smedley-Wordsworth stress Concentration factor equations for<br />
simple unstiffened joints are shown on the following pages. The<br />
notes for the equations shown on the following pages include the<br />
Shell d/D limitation of 0.95. This interpretation is open to a<br />
project-by-projectreview.<br />
The validity ranges for the $medley-Wordsworthequations are:<br />
Term Validity Range<br />
d/D 0.13 - 1.0<br />
D/2T 12.0 - 32.0<br />
t/T 0.25 - 1.0<br />
91D 0.05- 1.0<br />
2L/D 8.0 - 40<br />
30 -90<br />
where,<br />
D = chord diameter<br />
T = chord thickness<br />
d = brace diameter<br />
t = brace thickness<br />
9 = gap between adjacent braces<br />
L = chord length<br />
= angle between brace and chord<br />
c-5
C.3 PARAMETRIC STUDY RESULTS .<br />
C.3.1<br />
=<br />
The Kuang and Smedley-Wordsworth chord stress Concentration factors<br />
for T joints are shown in Section C.3.l(a) and C.3.l(b),<br />
respectively. The Kuang and Smedley-Wordsworth chord stress<br />
Concentration factors for K joints are shown in Section C.3.l(C) and<br />
C.3.l(d), respectively. The Smedley-Wordsworth chord stress<br />
Concentration factors for X joints are shown in Section C.3.l(e).<br />
Since the chord side of the weld stress Concentration factor is<br />
generally higher than the brace side of the weld stress Concentration<br />
factor, only the chord side of the weld stress Concentration factors<br />
are shown.<br />
C-6
C.3.l(a) Kuang Chord SCF’S for T-Joints<br />
The Kuang chord SCF’S for T-joints are shown on the following<br />
pages. The following parameters are assumed for the Kuang figures:<br />
1) y = D/2T = 12.0<br />
2) Q = = 30.0 degrees<br />
3) = = D/L = 0.0571<br />
c-7
30.<br />
2s ,<br />
20.<br />
15.<br />
10.<br />
5. &<br />
wORDSWORTH<br />
o<br />
05 IO IS 20 25 30 35<br />
Prod$ctOd SCF<br />
35-<br />
30. ●<br />
25,<br />
20-<br />
1s .<br />
10.<br />
5,<br />
KUANG<br />
o<br />
05 10 15 20 25 30 :<br />
Prmdmlod SCF<br />
359<br />
30.<br />
25.<br />
20.<br />
15.<br />
10.<br />
LEG<br />
o<br />
05 10 15 20 25 30 35<br />
Prmdxlod 5CF<br />
05 10 15 20 25 30 35<br />
PfOUctod<br />
Scf<br />
wJoInl ln$fd~ Oulsldo<br />
Tyoa V41#d,ly Vmllalty<br />
TIY ● ●<br />
K A b<br />
x o ●<br />
05 10 15 20 2s 30 :<br />
Prod)cltd EiCF<br />
Fig. 6-R~sulta o? compmnson—axial loading.
ChordSide<br />
BraceSide<br />
K-Joinlx<br />
SCFCX = 0.949‘y 4.666~4.059~1.IWq0.067sin1.521d SCFbX = 0s25y4“157g4.44170550~0.05S~1.44sin<br />
SCFW “ lao~<br />
4.3 ~0.06 ~o.94 sin 0.9 e<br />
SCFbV = 2.827 B a-m To-% sin050<br />
—.<br />
for O“
ChordSide Brace Side<br />
K-Joints<br />
SCFU = 1.8(rsinflfi) SCFbx = 1.0+ 0.6 Qr [1.0 +{;. SCFCX] >1.8<br />
SCFCY = 1.2(7 sin6fi) SCFby = 1.0 +0.6 ~ [1.0+%; SCFCY] >1.8<br />
SCFU = 2.7(rsin6fi ) SCFbz = 1.0+0.6Qr [1.0+~j. SCFU] >1.8<br />
Y-BranchJoints<br />
SCF Kuang = 2.06yoS808e“l“2~ (sin 6) 1 “694 71.333<br />
SCFAWS = 14rsind fory 25<br />
SCF cxmod<br />
= SCFCX +TcOS0<br />
SCFTC = SCFKuang s SCFAWS<br />
scFTb = 1.0+0.6 Qr [1.0 +/#. SCFTC] >1.8<br />
>SCF cxmod<br />
> SCFbX<br />
SCFY<br />
= sameasfm K<br />
SCFY = same as for<br />
K<br />
SCFZ = same asforK<br />
SCFZ = sameasfor<br />
K<br />
UnreinforcedCrossJoin=<br />
SCFX = 1.333(SCFTC) ~ branch+ ~<br />
.<br />
SCFbx = 1.0+ 0.6Qr [1. O+@ SCFXI >1.8<br />
SCFY<br />
= 1.333 (SCFCY)<br />
scFby = 1.0 +0.6 Qr [1.0 +3. SCFY]>1.8<br />
SCFZ = 1.333(SCFCZ)<br />
SCFbz = 1.0+0.6Qr[l.O+~. SCFz] >1.8<br />
angle betweenbraceandchord<br />
candiameter<br />
canthickness<br />
nominalchord thickness<br />
,,<br />
brace diameter<br />
stub thickness<br />
tA-<br />
(D–T)/2T<br />
dlD<br />
exp [-{0.5 T + t)~~t]<br />
MarshallFormulas Used for computingStressConcentration Factors<br />
-7 ‘3
Chord<br />
Sidg<br />
SCFCX= 1.7yT~(2.42 - 2.2S#2”2)sir$2(15 - 14”4P)6<br />
SCFCY=0.75y”*6r0”8(I .6#;14- 0.7~2)sin( 1.5– 1.66)6<br />
SCF~ =T@{l.56– 1.46~5)sh#2f15– 14.~)o<br />
Brace<br />
Side<br />
SCFbx= 1 + 0.63<br />
ScFby= 1+ 0.63<br />
SCFbz= 1+ 0.63<br />
SCFCX<br />
SCFCY<br />
SCFW<br />
Where<br />
~= BraceDiameter/ChordDiameter<br />
7=ChordRadius/ChordThicknes<br />
T= BraceThickness/Chord Thicknes<br />
O= AcuteAngleBetweenBraceandChord<br />
—<br />
SmedleyFormulasUsed for Computing Stress Concentration Factors<br />
for Unreinforced Cro= Joints
Definition of Parameters, Validitv Ranges and No:es OKI Tables<br />
o<br />
Definition of Tubular Joint Parameters<br />
~= 2L/D where D = chord outside diameter<br />
P = d/D T= chord vail thickness<br />
t = D12T L = chord length (distance between points of<br />
concraflexure)<br />
T = c}T d ● bkace outside diarnecer<br />
t ~ brace vail thickness<br />
~ ‘s/D g = gap between adjacenc braces<br />
Validitv Ranqes Eor Parametric !Zcuacions<br />
a ~4~&o<br />
O.lj:p:l.o<br />
12
(3) Table 2 only<br />
..<br />
(l)-~f~ ~ 0.95 for out-of-plane bending then use ~ = 0.95.<br />
(2) The equations indicated for K and KT join~s apply only to<br />
loading on,all braces in che same direction for ouc-of-piane<br />
bending.<br />
(&) TabLe 3 only .<br />
(1) FOC K joints in ouc-of-<br />
(~ ~ l+t&’e”::: ; > 0<br />
by che carin i - . -“<br />
ding replace the conscant 0.9<br />
(2) For KT joints in out-oi-ulane bending replace the conscant 0.8<br />
by the carm (1 - 0.1 1+4!J2 when < ~ O.<br />
....
-,<br />
l—<br />
I<br />
+<br />
?1<br />
-—<br />
I<br />
.<br />
4><br />
a ,-<br />
I<br />
I<br />
+<br />
x<br />
N<br />
—4 ----<br />
m<br />
/k -<br />
. .<br />
-c<br />
ii<br />
C.1<br />
I<br />
m<br />
Q<br />
;<br />
c-v<br />
:-:<br />
x<br />
w<br />
b<br />
u-l<br />
w<br />
o<br />
P<br />
-.<br />
“1<br />
+<br />
(-<br />
a<br />
-—<br />
I<br />
I<br />
1<br />
I<br />
I<br />
!<br />
!<br />
1<br />
---
‘1<br />
——. -.— ——. . ..-—_____ .<br />
--<br />
ICIIJT TYPE ~!ltI LO ADI:J(;<br />
,-<br />
Id.1-11 Ir t<br />
...- —..- .-.<br />
F1.,<br />
.—.<br />
-:llnrJ<br />
..<br />
—-. ,<br />
-& ,. . :.:L<br />
‘L_ . _<br />
— ..—.—-— .—<br />
+.3<br />
::-r<br />
Dr, X d::r,~$ .—. _<br />
-...----- —- ——-— .- ——- -— ---- —-<br />
I: HORD, !i.lDDLE SCF<br />
— —-. — --<br />
CHORD, CROWt4 SCF<br />
0.757 ‘6 ,0”5 (t.6&”25= 0.7f12) S,j’”5-J’6dh<br />
7Tfl(l.56-<br />
l.41j~5)S,,,<br />
(fJ7(15-14.4f3 ))0<br />
SCF=O<br />
K J() Ilfi.<br />
- .-- —__ —___<br />
;(”I . ;IL15<br />
-.<br />
/ \<br />
L-<br />
c-. ~<br />
/q “<br />
..-.’<br />
—._ __ . .._- ,____ .- —---- .<br />
—-.. -—<br />
.<br />
-d<br />
%?<br />
-. . .. -. . .. --- ————— _______ -—. ..-__ ____ --- -----—-——--..---—— — __. — —,
Kuang SCF Computdion<br />
1<br />
L<br />
u<br />
UI<br />
\<br />
f-<br />
1-<br />
0-<br />
0.2 0.4 0.6<br />
0.8<br />
Beta = d/D
5<br />
4<br />
L<br />
u<br />
w 3“<br />
9<br />
t<br />
o<br />
-. ii<br />
Ic<br />
Kuang SCF Computation<br />
T<br />
2-<br />
1<br />
0<br />
0.2 0.4 0.6<br />
Beta = d/D<br />
O*8<br />
.<br />
1
.,<br />
I<br />
,<br />
1- out— FIano SCF<br />
m<br />
tg<br />
n<br />
II<br />
~<br />
J
C.3.l(b) Smedley-Wordsworth Chord SCF’S for T-Joints<br />
The Smedley-Wordsworth chord SCF’S for T-joints are shown on the<br />
following pages. The following parameters are assumed for the<br />
Smedley-Wordsworth figures:<br />
1) y = D/2T = 12.0<br />
2) Q = = 30.0 degrees<br />
3) = = 2L/D = 35.0<br />
The Shell d/D limitations have not been imposed for the SCF<br />
calculation.<br />
-----<br />
C-8
—<br />
6<br />
Smedley-Wordswoti<br />
SCF Computation<br />
c<br />
5“<br />
!)<br />
IL<br />
9<br />
4-<br />
\<br />
.,.. -,<br />
I<br />
I<br />
L<br />
u<br />
3“<br />
4<br />
\<br />
..><br />
.-. \<br />
u<br />
.<br />
“x<br />
<<br />
0.2 0.4 0.6 0.8 - 1
—<br />
&.<br />
15<br />
Smedley-Wordsworth SCF Computation<br />
14<br />
13<br />
12<br />
11<br />
1t<br />
10<br />
9<br />
8<br />
{<br />
I ,,... 5’ I<br />
7<br />
.<br />
,,-..<br />
“;<br />
(/l<br />
u<br />
“x<br />
<<br />
1-<br />
6- — —<br />
5<br />
4<br />
3<br />
2<br />
1<br />
0<br />
+<br />
0.2 0.4 0.6 0.8 1<br />
Beta = d/D
T ln— Plan- SCF CrOwn POsltlon<br />
o -<br />
-s=<br />
u-l<br />
f<br />
,’)
—<br />
-.<br />
....<br />
5<br />
Smedley-Wordswotih SCF Computation<br />
c<br />
o<br />
4<br />
3<br />
“-‘ :<br />
I<br />
I<br />
M<br />
2<br />
.<br />
L<br />
I<br />
4J<br />
i<br />
1<br />
0-<br />
0.2 0.4 0,6 0.8<br />
1 I i I<br />
1<br />
Beta = d/D
C.3.l(C) Kuang Chord SCF’S for T-Joints<br />
The Kuang chord SCF’S for K-joints are shown on the following<br />
pages. The following parameters are assumed for the Kuang figures:<br />
1) y = D/2T = 12.0<br />
2) o = = 30.0 degrees<br />
3) a = D/L = 0.0571<br />
c-9
—<br />
5<br />
Kuang SCF Computation<br />
4<br />
IL<br />
L)<br />
(/l<br />
3-<br />
I<br />
1-<br />
kJ<br />
0“<br />
0.2<br />
0.4 0.6 0.8<br />
Beta = d/D<br />
I
5<br />
Kuang SCF Computation<br />
4<br />
IL<br />
u<br />
M 3<br />
Q<br />
c<br />
i!<br />
—.<br />
-1<br />
c<br />
2<br />
Y<br />
(<br />
4<br />
[<br />
1<br />
0- -44<br />
I<br />
0.2<br />
0.4<br />
i<br />
I<br />
0.6 0.8 1<br />
Beta = d/D
j<br />
K Out— ~lanq SCF<br />
‘?’<br />
+<br />
F<br />
.
C.3.l(d) Smedley-WordsworthChord SCF’S for<br />
K-Joints<br />
The Smedley-Wordsworth chord SCF’S for K-joints are shown on the<br />
following pages. The following parameters are assumed for the<br />
Smedley-Wordsworthfigures:<br />
1) y = D/2T = 12*O<br />
2). o = 02 = 30.0 degrees<br />
3) ~ = 2L/D = 35.0<br />
The Shell d/D limitations have not been imposed for the SCF<br />
calculation.<br />
—<br />
c-lo
,<br />
5<br />
Smediey-Wordswotih SCF Computation<br />
4<<br />
m<br />
3.<br />
2-<br />
1-<br />
Oj<br />
I i I<br />
0.2 0.4 0.6 0.8 1<br />
Beta = d/D
.<br />
,i<br />
5<br />
Smedley-Wordsworth SCF Computation<br />
s<br />
o<br />
4<br />
o<br />
u<br />
u<br />
u<br />
u)<br />
3<br />
I ,._.., I<br />
. ,-<br />
IL<br />
o<br />
u)<br />
2<br />
u<br />
!t<br />
1<br />
0<br />
1 1 I I I I I<br />
0.2 0.4 0.6 0.8 1<br />
Beta = d/D
Smedley-Wordswotih SCFComputation<br />
*<br />
i<br />
o<br />
L<br />
.—.<br />
1<br />
I<br />
IL<br />
u<br />
m<br />
/<br />
/<br />
I<br />
-’l<br />
0<br />
c<br />
u<br />
L<br />
I<br />
c<br />
0.2<br />
I--l<br />
0.4 0.6 0.8 1<br />
[
I<br />
I<br />
K Out— Plan= SCF —— Saddle Pos Itlon<br />
o<br />
L<br />
o - r=<br />
I<br />
t<br />
G-4<br />
+<br />
m .<br />
.
C.3.l(e) Smedley-Wordsworth Chord SCF’S for X-Joints<br />
The Smedley-Wordsworth chord SCF’S for X-joints are shown on the<br />
following pages. The following parameters are assumed for the<br />
Smedley-Wordsworth figures:<br />
1) y = D/2T = 12.0<br />
2) @ = = 30.0 degrees<br />
3) m = D/L = 35.0<br />
The Shell d/D limitations have not been imposed for the SCF<br />
calculation.<br />
C-n
-<br />
10<br />
Smedley-Wordsworth SCFComputation<br />
9<br />
t<br />
o<br />
“n<br />
o<br />
n<br />
7<br />
8<br />
I<br />
I<br />
IL<br />
u<br />
ul<br />
u<br />
5<br />
4<br />
3<br />
–x<br />
2-— n<br />
1<br />
0<br />
O*2 0.4 0.6 0.8 1<br />
Beta= d/D
\<br />
Smedley-Wordswoti<br />
SCF Computation<br />
4<br />
3<br />
2<br />
“1<br />
..<br />
:<br />
0<br />
ii<br />
Ic<br />
x<br />
1“<br />
o-<br />
Beta = d/D
‘\,<br />
?v<br />
x out— Plan- SCF —— Saddl= Position<br />
II<br />
I<br />
T 1 T I I<br />
/
C.3.2<br />
Tables<br />
The Kuang and Smedley-Wordsworth chord stress Concentration factors<br />
for T joints are shown in Section C.3.2(a) and C.3.2(b),<br />
respectively. Since the chord side of the weld stress Concentration<br />
factor is generally higher than the brace side of the weld stress<br />
Concentration factor, only the chord side of the weld stress<br />
Concentration factors are shown.<br />
.- .<br />
C-12
C.3.2(a)<br />
Kuang Chord SCF’S for T-Joints<br />
The Kuang chord SCF’S for T-joints are shown on the following<br />
pages. The following parameters are assumed for the Kuang figures:<br />
1) = = D/L = 0.0571<br />
C-13
I@q W ~tim<br />
T-juint Axial W<br />
Owd Side114hld<br />
1<br />
I 1 1 ~. l~o : Sml = 15*O I h= 20.0 : -=.=0 i<br />
: 1<br />
l— 1<br />
1 I I I !<br />
: Theta= :Tau=; Ma am : Ma 4m : Ha=d/R ! Ha =dIfi :<br />
I<br />
1 It/T: 0.3: 0.5: 0.7: 0.9!0. 3:0.5:0.7:0.9 !0,3:0.5; 0.7: 0,9: 0.3: 0,s1 om7:o.q:<br />
!1 — 1—1—l—!—i—l—l—l—l—l—;—l—l—!—!—!—:—!<br />
1<br />
:0.20 !O.bn OAK 0.432o,2n:o.7570.6730,5MO,ab:o.% 0.849o.b540.411:1.1441.0170.7830.492:<br />
#<br />
1 i— !<br />
i I ! I<br />
! 30.OdegSO.40~1.593 1,416I,w o,~;~.~ i,~ lm~ 0,~~zM7 ~1~ Iaw I,~&~ 2.W 1,~ lm241~<br />
! 0m524rad;-! I I I :<br />
1<br />
I /0.tiZ73h2.W 1A72l,17i3;3.Zb Z?1322421.411;4,1343.k75ZG?91.7EO;4SW4.4013.= ZIQ;<br />
i ;—; 1 ! 1 I<br />
1<br />
I0.~14.0143.%92.747l,rn;4.OMl 4,2743.~ Z070!6.W5.3934,151L612;7.2Mb.450 4.972 3.lZl;<br />
I<br />
i<br />
I 0:20 !1.137 1.011 0,~ 0.4W!1.U2 1.211 0.932 O,WH.71q l.= 1.176 0.740 :2ME 1.H30 1.W O,Mbl<br />
I _l 1<br />
i 45.Odeg; O.U&fjbb 2.549 1.96j L23&m 3.@ Zw 1.47614,331 3,Ei0 2.9b4 1.fh5;5.167 4.611 3.550 2.~<br />
i O,~radi-! I I 1<br />
: :0. bO 14,921 4,375 WE 2,119 ;5.~ 5.Z9 4.033 MZ4;7.~ 6.bll 5.W 3.202 ;8.91H 7.917 b.M4 3.0S<br />
1<br />
t ;—; I 1 I<br />
.-, ! !O.~17,221 6.4~ 4.942 3.11O;U49 7.&E 5.919 3.7Z4;1O,91 9,7M 7.4b7 4.699 ;13,0b 11.61 H.943 5,b27<br />
i :m :1*W41,4%1.097 O,m!l.m1.7071.314<br />
O.m?:zm21541.b5E1.043:2.902 2.5E) 1.W Lao!<br />
I 1 I—1 1 ; I 1<br />
I<br />
1 bO.Odqi 0.40 !4,041 3.~ ~7fi l,740;4,~ 4.302 3.312 2.~ lb.1~ 5m4~ 4,179 2.~;7,312 hm~l S,w 3,149;<br />
! L047radj-! 1 I I I<br />
1<br />
! O.M ib.9U b.ltl 4.74H 2,9Wi3.XEI 7.3E7 uw 3.ai10.4E 9.320 7,174 4m51&i5 11.16 E.592 5.Mb;<br />
I<br />
1<br />
;—~<br />
: i<br />
i<br />
I<br />
I 1<br />
I 0.&l IIO,lH 9.til 6.967 4.W4 :1219 10.83 8.344 $.ao 115.a 13967 10.Q 6,&L42 16.37 12.KI 7.933;<br />
1<br />
: 0.20 :2.OM 1.819 L4W O.ml KL451 2.in 1.677 1.055:3.092 Z74? 2116 1.33113.703 3.292 2.534 1.595:<br />
i ;—: 1 1<br />
I<br />
1<br />
I<br />
I<br />
1<br />
1<br />
90.Odegi 0.4 ;5.j5j 4,~ 3.= ~~;&.1~ S,#j 4,% 2&~7a790 6.% ~,~ 3m~~q,~ g,~ b.= 4m019<br />
i 1.571 radl-l 1<br />
I 1<br />
1<br />
! 0.bOM52 7470 ME 3.612 ;1O.M ‘?.4= 7.= 4.%5 !13,37 11.E9 9.154 5.7&4&Ol 14.24 10.96 6.6%<br />
1 1 l— i i<br />
1<br />
;<br />
1<br />
!o.81ilz9a 11.54 H.m 5.594 !15.5 13*U lo.b4 6.II W;19.U 17.44 13.43 B.m H3.54 m.w ltl.lm 10.12
KuwqSE _im<br />
T-joint In%neW<br />
UrrdSickofkld<br />
1<br />
I<br />
I I1 a = 12.0 ! a= 1s0 I ~. ~,o : h= ao !<br />
1 I l— !<br />
I ! ! I<br />
I llwta=!Tw=; W#D ~ Ma=d/D : W=dll : *W i<br />
I :tfl:o.3: 0.5:0.7:0.?!0.3:0.5:0,7:0.9!0.3:0.510.7:0.9!0.3: 0.5:0.7:0.9!<br />
~—;—~ — 1 —I—i—!—!—f—{—!—I—! —] —l—~—]—l—j<br />
: !0,~;0.551 ‘0.540 O.~ 0.S28:0.63;0.6180.410o.~ ;0.7490.7340.7240.717!0S70,640O.= O.E?O!<br />
) I_l<br />
: I 1 f<br />
! I 30*0dq;0,40h.ml 0.9Mom9M0.95E!1.145Lln 1.107i.m ;1.3?61 l.m 1.3151.302;1.5561.=41s041.469;<br />
! o.n4rdi—! : i i 1<br />
~<br />
:0.60:1.419 1.391,3721.358!1.6231.901.5b~l.= !1,9291.8901.864l.Mb12203MO 2.1312J1O;<br />
1 ~—~ I I 1 1<br />
I<br />
:O.@!1.S181.781.7571.740ELOn2,036Zm 1.589;Z470Z4B 2.= 2Jb4;U24 Z767 2.7?4 2703 ;<br />
1<br />
:0.20 :o.&n O.= 0.650 0.643 !0.766 0.7s5 0,743 0.75:0.913 0.89s O.m 0.674:1.044 1.OZ 1.009 0,999:<br />
I_ I I 8 I<br />
i<br />
45.0 dq; 0,40 bII 1,155 1.179 1.16H;1,3% 1.<strong>367</strong> 1.349 l.= ;l.H 1.624 1.603 l.~ ;l.1%% l.~ 1.832 1.814!<br />
i 0,7WIrdi—i : # 1<br />
i<br />
1<br />
I :0.60 H.729 Lb94 1.672 1.655:1.977 LW I.fll i.~ ;~~ Lx 2,272 29249b ~b32 2.597 2S71 :<br />
! I~—; : 1 1 1<br />
I i O,m R215217021412120KLm 2.481 2AM 2A24 ;3.0102.?49L9W Zw A 3.3713.Q6 3*293i<br />
1<br />
! Oa !0.754 0.739 Omm 0.722:0.863 0.B45 0,E34 O,EH:1.025 i.w O*W1O.ml !Li72 L148 Li33 1.12<br />
i .~—~ 1<br />
:<br />
#<br />
! 60.0 deg! 0.40:1.370 1.342 1.Z!4 1.311;l.X4 1,S4 1.514 1.49911.861 1.623 1.799 L7Sl ;2.U3 2sW 2.057 2.036<br />
i 1,047radl-1 1 I<br />
i<br />
#<br />
1<br />
! O.M ;1.941 1.W 1.877 1.EH !’2.ZO Z175 2.146 2.124 ;2.t211 2.= 2.5W 2.Z4 :3.016 2.955 2.715 2.N<br />
1 :—j : 1 I<br />
1<br />
i O.M :2.486 2AM 2404 2sW HJ43 Z7H L748 2.721 ;3.~ 3.310 3.264 3.2S ;3.0S 3.704 3.734 3A9b<br />
1<br />
1 I 0.20:0.819 O.~ 0.792 0.7B4 10.936 0.917 0,9M 0.89611.113 1,090 l,07b 1.065M72 1.247 1.230 1.21B<br />
( I l— i :<br />
i<br />
I<br />
1<br />
90.0 degi 0.40 !1.487 1.457 1.437 1.423 ;1.700 l,b& 1.M 1.627 ;Zm 1,979 1,?53 1,~ H.31O 2.263 2.~ 2.210 !<br />
1 1.571 radl-! 1 I , 1 , 1<br />
1<br />
! 0.64 !2J07 2.M 2037 2.o17 ;zW 2.361 z.~ 234 !zE63 2.KU 2.764 2.740 ;3.z74 3.B7 3.164 3.133i<br />
i I —1 I 1 I I<br />
1<br />
; O.W i2.6W 2.644 2.M9 2.W3 ;3.U36 3.023 2.%3 2~ ;3,647 3.593 3.545 3.W9 i4.193 4.lM 4.053 4,012~
-- -- -- .- -- -- -- -- -- -- .- -- -- . . -- -- -- -- -- -- -- -- .- -- -- -- -- --<br />
“. -- -- -.<br />
as ------- -.-.-<br />
-.4’K..<br />
..--.--..-k d=lw .-<br />
iniiii<br />
i?<br />
u<br />
.- -- .- .- -- -.. . .<br />
niini<br />
-- -- -- -- ..- -- -.<br />
iiiiiiik<br />
-- -- .- - -- . . -.<br />
glij$g<br />
-- -- -- .- -. -- --<br />
iiiiiii<br />
ii<br />
-- -- .- .- -- -. -.<br />
NH<br />
-- -. -. -- -- -- --<br />
Uinil<br />
iiiiiii<br />
Him<br />
ii<br />
.- - -. . . -. .- .-<br />
inib<br />
Uinii<br />
ii Fd iii<br />
uik ii<br />
. . -- -. -. -. .- . .<br />
USE<br />
.- -- .- -- . . -- .-<br />
Iii Uii<br />
iiviiii<br />
. 04<br />
-. -- -- -- -- -- .-<br />
-- -- -- -- -. -. --<br />
.- -- -- -- -- -- --<br />
.- -- -- -.. -- -- --
C.3.2(b) Smedley-HordsworthChord SCF’S for T-Joints<br />
The Smedley-Wordsworth chord SCF’S for T-joints are shown on the<br />
following pages. The following parameters are assumed for the<br />
Smedley-Wordsworthfigures:<br />
1) = = 2L/D = 35.0<br />
The Shell d/D limitations have not been imposed for the SCF<br />
calculation.<br />
C-14
—<br />
T-joint hid<br />
S5 Crm i%itim<br />
1 1<br />
1 i -= 120 : - = 15.0 ! ~= 20.0 j w=. a.o :<br />
1 ;—l I 1<br />
1 I t<br />
I<br />
: kh= ;Tw=l Ha 4/D ! b din : Ma*/o : h =dm<br />
I<br />
1<br />
I IWT : O*3!0.5: 0.710,9: 0,3! O,s: 0.7: 0.9: 0.3: 0.5: 0.7: 0.9: 0.3:0,5:0.7: 0.9;<br />
j—~— !— 1<br />
—<br />
1<br />
—1—l—!—l<br />
—l—l—l—l—!—l—l—l—l—!<br />
1<br />
I O.a :2447‘3.OSI ‘Mu 3.2H3 :2.929 3*N13.1773.21E :3J693*14a3.l&l3.166:3,207 3*m 3.* 3.145 !<br />
I :—; I ! : I<br />
: W.o fkj 0.54 W23 5.7W Lm b.(m ;5.365.7195,9075*W1;sm~g~,~ 5,7935,~:5,716 5m~ 5,~ 5,=;<br />
I 0.S?4rad:-1 I 1 t 1<br />
: ! 0,25:8.15 9.361 %793 ?.~ k 140 %012 ?,ZJ1 9,015 ;8.276 E,W 8.891 R.= ;8.516 8.W7 8.7M MM ;<br />
{<br />
i— r I ~ I 1<br />
f<br />
! 1*Mh 13.5 14.49 13.71ill.% 13.(U 13,44 1275 :11.4E 12.46 12.57 11.Q h 1M6 12J5 ilma ;<br />
....<br />
..-,<br />
1<br />
1 i 0S 13.2U 3m&14.344 4.UO S294 3.W 4,320 4,7iZ :3.4~ 3.W 4,304 4.6W :3.E57 3.954<br />
4,311 4.651:<br />
I<br />
I —i !<br />
I 1<br />
!<br />
1 M.O deq; 0,50 :5.74? 7.026 0.1~ 9.@ :5.06? 7,MI 7,% &EM ;kM4 7.(U3 7,905 MN k45 7.1~7,903 0,S1 !<br />
I 1.047radi-! t 1 # 1<br />
)<br />
:0.75 :a.m10.M 12s27 13,52:s.643 10,47 11,95 13.10km 10.46 11,72 1273 k 10.57 11.66 i2.54 ;<br />
1<br />
1<br />
~—~ I<br />
i : :<br />
! lm~ :11.57 14.63 16.M 1H,33;11.65 14.~ 16.Z 17.ti :11.94 ]4.14 15.83 17.03:12.32 14.19 15.65 16.71;<br />
1<br />
i 0.3 i3.E14 3AM 4.6UI 5.W !3.1S 3.W 4S47 5.1?6 13.240 3.BW 4,511 5.CQ7!3S43 3.741 4.549 5.046i<br />
1 ;—{ t t 1<br />
i<br />
i W.Odeqi O.~ !5.%9 7,X5 H.6W 10.02;5.6!II 7.15# E.= %779;5.M2 7,184 8.422 9,= ;b.03S 7.2K 9.397 9.443i<br />
I 1.571radl-! I<br />
! i I<br />
I<br />
:0.75 :H.~ 10.81 13.ti 14.93;E.315 10,67 12.73 14,51:9.%4 10.64 12SI 14.13IH.E1410.71 12.41 13.93:<br />
I :—~ I I<br />
i<br />
i<br />
1<br />
: 1.M !!1.06 14.76 17.76 ZII,1O;11.14 14,45 17.22 !9.45 ;11.39 14.XI 16.7E 18.86:11.70 14,?3 MO 18.54!
T-joint Axial SF Saddle Posiiim
.<br />
T-joint<br />
In-PlaneSE UM P~itim<br />
.,<br />
1<br />
~0.3 :0.M5 1.0)9 1.079 1,066loom 1.153 1.233 1.21? ;1,175 1.370 1.4M 1.449!1.343 1.567 1.h7b l,b7 :<br />
1 ;—~ : 1 1 I<br />
1<br />
45.0 dq; O,W ;I.w 1.7% 1,879 I.W 11,721 ?.~ 2.lW z.1~3 ;2,046 2.224 2.553 2.523:2.339 2.728 2,918 2.W5 (<br />
I .0.785rad~—1 1 1 1 1<br />
i’ ~ 0.75 !2.083 2.429 2.599 2.5$9;2.331 2.778 2.971 2,931 ;2.830 3.301 3.ECfl3.490 ;3.25 3.774 4.037 3.’?91 i<br />
1 :—; I 1 1 !<br />
n<br />
iLoo i2ab22 3.059 3.271 3.234;2.%113.497 3.740 3.W7 !3.%2 4.155 4,445 4.34 ;4,073 4.751 Lou<br />
5*OZ:<br />
1<br />
:0.25:1.063 1.162 L165 l.om !1.214 1.329 1.32 1.234;1,445 1.579 low 1.447:1.652 1.W 1*E1O1.577:<br />
I :—~ 1<br />
;<br />
1 1<br />
1<br />
60.0 deql 0,50 il.852 ?.024 2.029 I.w 12,117 2.314 2.~ 2.149 !2.516 2.755 2,757 ?.554 ;Z87k 5.144 3,152 2.9?0 ;<br />
I<br />
!<br />
i,C-47rad:—: 1 [ 1<br />
1<br />
1 ! 0.75 :2.5Li 2.W 2.907 2.6W :2.92j 3.201 3,.W 2.973 ;3.W3 3,!304LU4 :.533 h9 4.3XI 4,360 4.039;<br />
~—~ I 1 1 I<br />
: I.W !3.224 3.Z5 3,573 3,273!3.&4 4.030 4.04433.742 !4.381 4,7E9 4,801 4,440:5.009 5.475 5,449 5m’OK!<br />
1<br />
1<br />
: 0.25:1.231 1.%<br />
:—;<br />
1.231 1.OW!l.4c4 1.470 1.407 1.245!1.673 1,747 1,672 i.480 :1.913 i.W7 1.912 1.b92!<br />
1 1 1 1<br />
I<br />
90.0 dq.1 0,50 ;2.144 2.239 2.143 LW&;2.452 2.5? 1.450 I.IM ~Z.9143,~2 1.912 2.576 K3ZS 3.47H 3.329 2,9% ;<br />
I 1.571radl—1 I<br />
i<br />
1 ! 0.75 il.% 3,097 2.W5 2.h23 !3.391 3.540 Z.W 2.9W !4.034 4,207 +.02B 3.’%4:4.M9 4.010 Lb05 4.074;<br />
! {—; 1 z 1 1<br />
1<br />
: 1.00 !3.734 5.WR 3.TL 3.302!4.26’?4.457 4.267 :,775 !5.073 5.296 5.070 4,4S6;5.W 6.055 5.7W 5.129 ;
-—<br />
T-joint CtkuHlane ELFSaddlePcisitim<br />
;—~—j _~_~_ ~ —l— l—~— :— 1— I —;— ;—1— ,—, 1<br />
— j_<br />
1<br />
I ! 0,5 !0,527 0.773 1).m O.w ;(),W ‘%%7 1.031 ().6% ;Omm ‘imm 1.375 O*9ZI;1s103 1.612 1,719 1.15’?<br />
I ;—: 1 1 1<br />
t<br />
?%0daq~0.50 !I.OEN1.547 l,b50 1.112;1.324 1.934 2.W 1,390!l.7& 2.579 2,755 1,E54:2.207 3.224 3.439 2.31E<br />
I 0,524rad:—; I 1 1<br />
I<br />
! 0.75 !1.58’+2.321 2.475 1,6M k 2.902 3.094 2.wb ;2,b47 3,@M4,125 ?.7E1;3,311 4,W7 5,157 3,477<br />
!<br />
:—! I t I x<br />
! ! 1,00:1.119 3,095 3.W<br />
1<br />
2,23 ;?aMq 3,~94.1~2,711 :3,532 ~=j5q5,5M3.709:4.415 b,~q~,g~b~a~:<br />
!O,fi;O.E72 1,347L5bl1,176 ;1,0%) 1.6841.9sl1,4711:1.454<br />
;—~ 1 1<br />
2,2+5 2.b02 1.%4 !lAIE 2.EJ17:.252 2.45fJ:<br />
# 1<br />
1<br />
45.0 deq! 0,50 !1.745 2,h74 3.122 2.352 ;2.1!313.W 3,9Q3 2,740 :2,90B 4,471 5,204 3.920!3.b3b 5.b14 b.~ 4,9@ :<br />
!<br />
f 0.7fi rad;—; 1 [<br />
;<br />
1<br />
$<br />
~0.75 ;2.610 4,042 Lb03 3,520k.2R s.0535.K44,410 ;4.3Hb.7377.00b5.M%4540.4219,757<br />
7.351 !<br />
I ;—/ 1 1 ;<br />
1<br />
1<br />
! 1.C4!3.490 5.399 6.245 4,704 ;4.%3 6,737 7,W 5.W ;5.817 8.?03 10.40 7,B41;7.272 11Z2 13.01 ‘?.601;<br />
1<br />
: O.fi 11.169 1.EM 2.2b7 1.82 !l, %0 2,329 2.K3 2.2j% !1,?47 3,10h ;.7i0 3.037 !2.434 3,= 4,723 3.7% i<br />
! :—1<br />
;<br />
1 [<br />
[<br />
1<br />
60.0 dey 0.50 !2.237 3.7274,n4;.M4:2.’%?1 4,L595.6474,5%;3.m b.2127,557!5.074 :4.063 7.X5q.w 7.593:<br />
! 1.047 radi—i # ! 1<br />
1<br />
1<br />
:9.75 !3.505 5.591 4,MII 5.4b7 14,X2 .5.%Q E1.ml .5,e3.4 ;5.W2 ‘q,31E 11.33 ~,112 ;7,303 11.64 14.16 11.391<br />
1 :—/ f 1 I I<br />
1<br />
: I.@ !4.h74 7.%5 ?.W 7.LW%.842 q.31B 11.33 7,11217,790 12.42 15.11 12.14h7 15.= 1S,69 15.16I<br />
1<br />
! 0.25:1.437 2.344 2.9FA 2.4E4 ;1.7% 2,732 3.h92 3,10a :2,~q 3.710 4,923 4,144 !2.994 4.~ b.154 5.lM ;<br />
- :—; 1 1<br />
i<br />
!.<br />
1<br />
?0.0 dqi 0.54:2,874 4.692 5,X4 4,W3 ;3,593 5.W 7.= A.21b14.7?1 7.U 7,047 a.= ;5.989 7.775 12.W 10.3b;<br />
; 1.571racll-! I t 1 1<br />
i %75 :4.312 7,03E S.M2 7.457 ~5.390 0,797 11.b7 q,E4 ~7.107 11.73 14.77 12.43h.9B4 14.bb lB.4b 15.54;<br />
;—; 1<br />
i<br />
1 I<br />
1<br />
: 1.C4:%74q 7.X4 11.81 7.W :7,10] 11.73 14,77 12.43:q.~ 15.& 17.h9 16.57;11.’77 Iq.Zj 24.M X.72 :
—<br />
i : m ! 0.3: O*5: 0.7: o*?: 0.3: 0.5: 0.7: 0,9: 0,3; 0,5: 0.7: 0.9! 0.3! 0.5! 0.7: 0.9<br />
:—; —i—:—i— — — — — — —! —!—l—!—!—<br />
I ! 0.20:0.427 0.414 0.406 ‘0.40) ;0.4% ‘o*Ml 10.471‘0.464 ;0,600 ~’m;o.w !0.6% 0.676 o.bb2 O.m<br />
t<br />
1 :—l I 1 I<br />
1 30.0 I@; 0.40 10.91E0.6?! 0.E73 0,860;1.065 1.034 1.o13 0,99Eh?tl 1.252 LW l.~ h 1.453 1.424 1.403<br />
I 0.524rail-l I t i :<br />
t<br />
i0.60!1.4371.3941“371.347iLbb71.6181.3 1.%2:20191.9591.921.89223432.2132.2292.1?6i<br />
# I 1—1 : :<br />
I 1<br />
1<br />
!O.W!1.9741.9151.878l.~ KLm 2g~ 2.1792147!2,774Lb%?2A9 Zw ;3.2193.lZ3.M23’.017 ;<br />
!<br />
IO.~!0,7240.702Om~ 0.b7E:0.840 0,M5 0.7W 0,7B7:1.017 0.9H7 0,967 0.953:1.10 1.145 1.122 l.iti !<br />
: :—l I 1 1<br />
1<br />
I 45.0 degl 0.40 :L~ 1.510 1.W 1.4% ;L8E 1,751 1.717 1.69212.186 2J21 2.W 2J49 k 2.461 2.413 2.37Ei<br />
! 0.7% rti:-1 i i<br />
1 1<br />
: i 0.60MM 2.3k231b2.202i2.~21412A872647!3,4213.3J03,Z43.W ;3,9bq 3.~ 3.77b3.720;<br />
1<br />
I —1 i I : 1<br />
! I ;O.~&345 3.2453.lD3.1S13.~3.7653.b913.637;4.700 4,%1 4.4714AM !%4535.2915.1B75.111i<br />
1<br />
1 !0.20!0.9%0.9560.937O.m :1.1431.lUI1.W 1.0711.W 1.3431.317l.m :1.646I.= l*52al.m I<br />
I<br />
:— 1 t<br />
i<br />
1< i<br />
1 60.0d~l0.40;Z1182.052.015l.~ ;2.4Z2.W Zm 2303%97bZm 2.KH2.7W;3.45J 3.3513.= 3.Zb!<br />
I 1.047 rxll-1 : i<br />
1 1<br />
i i 0.60!3.3143.2153.lD 3.106;3,B453.7313.6S73,604{4.6574,5194.4Y 4,365;5.4035.2435.1405.064;<br />
1 :—: I 1 t t<br />
1<br />
: O.~ :4.= 4.418 4.Q1 4,267 ;5.2W S.126 5.M 4.%! ;bm~ b,~ 6.(M6 5.996;7.4237,2037.061b.557 ;<br />
1<br />
:0.20 :1.z?b 1.190 1.1661.149!1.423 l,m l.m 1,353:1.723 1.672 1.639 1.615:1.999 1.940 1.902 1474:<br />
:<br />
I—1<br />
:<br />
1 1 1<br />
i 90.0 deg; 0.40 ;2.634 HE 2A07 2.470!3.MB 2.9L4 2.W 2.EM :3.704 3.94 3.24 3.472i4.2Xi 4.170 4.OW 4.02Ei<br />
; 1.571rdi—1 1 t I 1<br />
I<br />
i 0.60 !4.124 4.W 3.~ 3.EM;4.E 4,643 4,5s? 4,M ;5,Rb 5.b24 5.513 5.M ;b.Z?5 b.= b.397 b.W :<br />
i<br />
I —1 1 I : !<br />
I<br />
! O.~ ;5.M45.4q5.~ 5.311 ;6.574 b.m b.m b.lK ;7.963 7,726 7,575 7.443 :9.Z9 8.965 8,7W %.bW;
I 1 1<br />
1 1 ~. l~o ~ ha= 1s0 : ~. ~J ; a=ao i<br />
i ;—{ I I<br />
1 \ 1 i<br />
! W= !Tw=i ti#D i ti=d/o : M=d/u : *=IVD :<br />
i ! tll : 0.3! 0.5: 0.7: o.? ! 0.3: 0.5I 0.7: 0.9! 0.3; 0,5: 0.7: 0.9: 0.3: 0.5: 0.7: 0.9:<br />
:—~— 1.— 1 I —i—I—!—I—!—!—!—i—I—: —l—i—!—l<br />
1<br />
I ! 0.20;0.s14‘0.S0 ‘0.5410.54910.559 0.5770,5890,598:0.6240,6440.6570.647!0.679o.7m0,7150.726:<br />
! i —t I I I<br />
:<br />
t<br />
30.Odtqi O.40;0.m 1.017l.~ M54;L074i.107l.1~1.147;Ll~1.ZSl.~ 1.279;1.304 1.3441.372l.~!<br />
: 0.!E4rtij-! 1 1<br />
: 1<br />
I<br />
!0.60:1.444 1.4891.S201.543;l.m 1.b21,6s4M&54 1*W 1,8451.873:1*WI.w 2.W 2.039;<br />
: 1,—1 1 I 1 I<br />
#<br />
;O.Ol&f l.~ 1.992.022;2.060 2.1242.16$2.201&l 210 2.41SZ~;2.S02La 2A32LbR;<br />
I<br />
i 0m3~0.702 0.724 0.739 0.Z4 :0.764 0,7M OM4 0,816 :0.~ 0.879 0,897 0.91110.928 0.957 0,977 O.Wl {<br />
i :—: I I [ I<br />
1<br />
# 45.0 deqi0,40U48 1.390 1.410 1.439;1,467 1.513 1,543 1,5b7;i.b3b 1.W 1.722 1.74&81 1.W 1.874 l.~;<br />
I<br />
! O.~raU—l<br />
I 1<br />
:<br />
i 10. M:l.w3 2&4 2m07b2.107;M8 U14 NM 2.Z4;U9b L470 2S21 2.S9hCtl 2.hW 2.744 L7M:<br />
1<br />
I<br />
1<br />
1 —t t :<br />
I #<br />
i 10mKk!3b 2M L7212.762;2s815L902 2.761 3.0u 13.140 3.m 3.304 3.354:394103.5243.36 3.651i<br />
#<br />
! 0.20:0.843 0.8M O.w O.m !o.917 0.946 0.965 O.w!l.oz 1.0551.077 1.093:1.114 1.149 1.172 1.190:<br />
1<br />
I i,—~ 1 t I<br />
!<br />
I<br />
bO,Odeqi0.40 !1.617 1.664 1.702 1.7ZI;I.761 1.815 1,~ i.mi ;1,?64 2JE 2,066 2.C49!2.138 2.204 2.249 2,284!<br />
! L047rad:-1 t 1 t t<br />
1<br />
: o.bo :2.3b8 2.442 2492 2.519 ;Z57112.H 2.712 2,~;ZE75 2.9~ 3.&5 3.071;S,130 3.W 3.293 3.343;<br />
1 f—~ i i<br />
i<br />
1<br />
1<br />
: 0.kl!3.103 3.~ 3.2LS”3.315:L~ S.m 3.54 3.LuJ:3.7M 3.W 3,9B 4.u:4.j~ 4,~ 4.31b 4.3E2;
Ku&W<br />
Cr@atim<br />
K-jrnntiht++laae SF<br />
M Sik 114Md<br />
t 1 1 ~. 1~~ : & ❑ 15.0 I -= 20.0 : bm=ao I<br />
I<br />
[ i —1 1 1<br />
I I<br />
: TMiI= :Tu=; mta#D ; Ma =dm i w =dm ! maw :<br />
1<br />
:UT ! 0.3! 0.5: 0.56 !0.75; o,3:o.55! o,5b; o.75: o*3:o.s: o.%: o.nio.310*5:om% :0.75:<br />
:—~_ I —i—<br />
I —i— l— i—i—i—l—:—l—l—i—l—;—l—!<br />
1<br />
:0.20 ;0.400 0.645 ‘o,ku 0.557 :0.w2 0.w9 o.~ o,hw ~o.672 Low w2 0.936!O.E43tJ59 1.407 1.174:<br />
i : —1 i i<br />
1 1<br />
t<br />
24).0dql 0,40 ;o,742 1.196 i,= LOS :o.~ 1.4w Im5521,~ :I,2M 2.007 2,0n 1.734!L5M L517 2.iM 2.174;<br />
! 0.=4 radl-! i i i i<br />
: : O*M:1.064 1.715 1.77s L48i :1.334 2*1W 2.m 1.= !1,786 2878 Zm 2.487EL240 3AM 3.737 3.118:<br />
1<br />
!—1 1<br />
i I 1 I<br />
I<br />
! O.MIM74 2.214 22?2 1.?13 ;1.723 L777 2675 2.399 ;%3)7 3.717 3s848 3.12 h93 4.662 4.026 4,027:<br />
1<br />
i 0.20 iOA07 1.107 1.146 0.957:Omlk?1.39 1.43a1,~ !1.154LK19 1,925l.bM ;1,447 2231 2.413 2014 i<br />
1<br />
1<br />
:—:<br />
1<br />
! 0.20 !0.942 1.519 1.512 1,312!l.IU 1.904 1,971 1.645:l,!E? 2S49 2A? 2.X? !l.9E4 3.197 3.309 2.7&2:<br />
I<br />
;<br />
i<br />
i<br />
1<br />
KMdq; 0.40:1.745 2.013 2.912 2.430 !~jw 3.527 3,M1 3,047 ;2.730 4,722 4,m 4,079 !3.b74 5.~ 6.12? 5.115:<br />
! 1.047rad:-: i i<br />
I 1<br />
i<br />
: OJJ) :2.~ 4nN 4,175 3.~ :3.139 5.059 5mZb 4,370 !4.202 b.771 7.(N9 5X4 !5,269 8.45V B.7B9?,= ;<br />
i I,—; 1<br />
1<br />
;<br />
i<br />
1<br />
I O.~ :3.~ 5.2W 5.392 4.5M ;4.054 b.Sj2 b.7b2 543 :S,4W 0.744 9,0Q 7,555;b.8)5 10.9b 11.35 9.473:<br />
1<br />
t ! 0.20 il.179 1.9M L9b7 1.M1 :1.478 2.~ 2.4bb 2.o5E:1.979 3,1W 3.302 2.E5 :2.M2 3.W 4.140 3.455:<br />
1 I,—j 1 1 I<br />
!<br />
t<br />
90.0 dqi 0.40 ;2.184 3.519 3.643 3.040 ;2.~ 4.412 4.% 3,812 ;3,W 5.907 kl15 5,1K3 :4.597 7.407 ].U b.3W :<br />
! 1.571rd:—! : I 1 1<br />
! I ~O.ti :3.131 5.046 5.224 4.359 !3,927 b.Q7 b,~ 5.447 ;5.257 S,471 9.769 7.318 ;b,5W 10.b2 10.W 9.lV” :<br />
I ;—; I 1 I 1<br />
1<br />
! O.B) :L044 6.517 6.746 5.bJl ;5.071 8.171 .9.459 7.Obo;b.7W 10.?3 11.32 %451;8.513 13.71 14.20 11.5 ;
1<br />
!<br />
! O.fi ;0.762 %841 0,769 O,= !0.881 0.973 o,m o,673 ~1,~ 1,1~ I,071 u,B1l !!.n 1,35$ i,2~ 0,9~<br />
I I—; 1 1<br />
1<br />
1<br />
1<br />
I 30.0dq; 0,50:1,525 1,M3 1.537 1.164;1,7L3 1,?46 1.777 1,36 :2,M 2.347 2,143 1.b2212,4~ 2.7:3 2.477 1.876<br />
{ 0,524radl—!<br />
i<br />
1 ! 1<br />
>,<br />
I<br />
1<br />
~—; 1 I<br />
I<br />
1<br />
45;0 dql O,M :1,B!4 2.M2 l,&2B 1,JE4;2,097 2.315 2,1i4 l,W) ;2.5M 2,7?1 2.548 ~.Y~ :2.923:.~ ~.?~ 2,231<br />
! 0.765radi-; 1 1 1<br />
1<br />
45,0 dq: 0,75 !2,721 3,M)3 2,742 2.076 ;3,:46 3.472 3,!71 2.401 ;3,793 4.1B6 3.B23 2,W4 ;4,3!%I41~ 4.419 3.346<br />
! 0.7!35radl—! I 1 1<br />
1 I 1.00 ;3mb~ 4,005 LL7 2.7t9 ;4nlfi MJO 4,220 3,201 ;5.057 %= 5.097 5,SS9k.%? b.455 Z.N3 4,462
!WsA-MIEy<br />
SF bqutaticm<br />
K-joint<br />
Axial SK Saddlei%iticn<br />
1<br />
I<br />
: kfi<br />
~—:<br />
M.53i 0.57? MM 0,239 :0,61! 0.670 0.5?4 0.27b ;0,723 0.7B9 0,b24 0,3ZI \fh7W 0,671 0.690 0,’UO:<br />
I 1 1 I<br />
1<br />
:0,0 daqi 0.3 !1,062 1,15E 0,917 CI.47E;1,ZZ3 1,340 1.061 0,553 ;1.446 i.J~ 1,24? 0,651 :1.5~ 1s743 l.~ 0072f:<br />
: 0.524radi—1<br />
1 1 1<br />
!<br />
$ 30.0 dq; 9.75 !:.593 1.TM 1,37A 0.7!8 ;LE42 2.010 1,592 0,830 ;2,169 2.366 1,E74 0,?77 ;2,397 2.A15 2.571 :Sw :<br />
! 0,~4 rad;—~ I<br />
1<br />
1 r 1<br />
1.00 !2.124 2.317 1,635 0.937 !2,457 2.MO 2.122 1.107{2.~i 3.155 L4W !*X3 !:.1?6 3*M7 2.761 ;,440 [<br />
1<br />
1 { 0.23 !0.?63 1,076 O,we 0,515:1.114 1,245 1,039 0.5% :1,3!1 1,465 1.223 0’701!1.449 1.619 !,32 0.774 !<br />
I ~—~ 1 r ! 1<br />
! 45,1) rjq; O,m /Iu~~ ~.j~ 1,~ l,OM ~z,~ ~,490 2a079 1,191~2,b~ 2,931 2,447 1,402~2m5_99 3,239 ~,704 IE39 ~<br />
1).7Wrad;—1 I t 1 I<br />
I<br />
t 45.0 deg! O.~ :2.~ 3,229 2,696 1.545;3,343 :,~ 3,119 1,~ ;:,935 4.397 3,b71 2,103;4,349 4,E59 4.055 2.324;<br />
! O.~ racil-! I 1 I t<br />
I<br />
1 ! 1.00 !3,B54 4,X)6 3,595 2.OM:4,458 4,9B0 4.159 2.3S3 ;5.247 5,243 4.355 2,M5 ;5,797 6.478 5.409 LW9 ;
Ikrdwth-kdlw<br />
SF l%quiaiifm<br />
K-joint IwPM<br />
ST Crew Pmitim<br />
I,M91.0791.066!0,9W1.1S 1,2S 1,219;1,175 1.370 1.466 L449!1,3431.%71.b7b 1.657:<br />
f 0<br />
I<br />
1<br />
1.7’55LB79 l,&57;1,721 2.008 2,14H 1,123;2.04S 2,386 2.Z3 2.523!2,33~ 2.72E 2.910 M35 :<br />
I I I 1<br />
2.429 2.99 2.M9 ;2,331 2,770 2,971 2.931 ;2,U.O 3.301 3.531 $4$0:3.235 2.774 4,037 Z.Ri ;<br />
1 I 1 8<br />
~.~ 3.271 :.2:4 ;2.99E Z,497 3,740 3.b97;3.562 4,155 4,445 4.394:4.073 4.751 5.062 5.(U3i<br />
...
-- ..—--- ------ -- ------ ------<br />
in-3cl.<br />
2LLxi%.3<br />
------- ------ --<br />
-1 ~<br />
&<br />
ma<br />
----- ----- ---- --<br />
-- -- -- ,.-<br />
S<br />
?#<br />
w<br />
------ ------ --<br />
iiiwiii<br />
-- -- -- .- -- -- --<br />
;;5s<br />
------<br />
.- .-<br />
------ ---<br />
.<br />
~<br />
* %. $=<br />
------ ------ --<br />
*.<br />
s En<br />
-- -- .- -. -- -- --<br />
------ ------ ---
hdsmthqwdhy W Camputatim<br />
l-joint Axial$# %ddleh51tl~<br />
I<br />
1<br />
:—~<br />
I 1 1<br />
1<br />
:0.0dql 0.50’:3,550 2.5391.&i2!,7M;4,437 3,1742,3152,232;5.9~64.2323.C%72.?77:7,3S LHO S,Hi? ;,721<br />
2.:24 radl—; 1<br />
) 1<br />
~0975:ZazzZBOB 2.7782.H7!6.6564,761?.4733,249!8,075L14B 4*M 4,465;11.097,?355,7895,32<br />
I ;—l<br />
1 1 I<br />
,<br />
1<br />
45;0dqi 0.50 :4.55 4,991 4,271 Ljt7 16s1916.29 5,339 3,959 ;8,~ %,:1? 7,:1? 5.278k31 10.39 9,BW 5.598<br />
! 0.7M radl—1 1 I 1<br />
I ! 0.75 !7.430 7,487 k407 4,73 ;9.237 7,3E4 E.(W?5.9?4 :12,38 12,47 10.67 ?,?10 :15.47 !5.59 15.34 9.E97<br />
) ~—~ 1 1 I<br />
1<br />
! 1.00 !9.906 9.%2 E.543 6,FA ;:2.Z !2.47 10A7 7,91BIIA.51 16.6.3 !4.23 10.55;20.LJ 23.79 17,79 !3,19
kbrdwmii%dl~y<br />
SCFComputation<br />
,..-,
Ejrnnt flkf+lane<br />
SF SaddlePmitim<br />
~,<br />
#<br />
45,0 d~j 0.50 :2.M7 2,:!: 2.:94 2,125 !2.509 2.W 2.7%2 2,656:3,346 5,852 Z,WO Z.542 ;4.MI 4.M5 4,7W 4,43<br />
~ 0.785 rad;—;<br />
1 1 1<br />
1<br />
i 0.7513.011 3.4% 3.591 3,W k,764 4,333 4.489 3,965k.!U95,7785.% 5.313 ;6.274 7.D27,4%?b.b42<br />
1 I ,—~ 1 1 1<br />
1<br />
!<br />
! 1,0014.015 4.6224.7E8 4.251 ;5.019 5.778 %985 5.313 k.t/72 7.704 ?,~l 7.0f5 ;8,U6 ‘?.634 f’,77k 0.256
C.4 FINITE ELEMENT ANALYSES RESULTS<br />
Finite element analyses were performed on the connections between the<br />
column tops and upper hull girders and between the corner columns and<br />
tubu1ar bracing of the column-stabilized, twin-hulled<br />
semisubmersible. The overall geometry and locations of the two<br />
connections are indicated in Figures C.4-1 and C.4-2. Longitudinal<br />
and transverse girders (8.2 m deep) coincide with the column faces.<br />
The columns are 10.6 by 10.6 m in cross section.<br />
Some dimensions of interest are:<br />
Overall length<br />
Overall width<br />
96.0 m<br />
65.0 m<br />
.,—<br />
Lower Hulls (two)<br />
Length<br />
Width<br />
Depth<br />
96.0 m<br />
16.5 m<br />
8.0 m<br />
Stability Columns (six)<br />
Size (square w/ rounded columns) 10.6x1O.6m<br />
Transverse spacing (center-to-center) 54.o m<br />
Longitudinal spacing 33.0 m<br />
Upper Hull<br />
Length<br />
Width<br />
Depth<br />
77.0 m<br />
65.0 m<br />
8.2 m<br />
C.4.1<br />
Column-Girder Connection<br />
The location of the connection is shown in Figure C.4-1. The joint<br />
dimensions are given in Figure C.4-3. The loading analyzed was a<br />
combined axial, shear and moment load. The .SCFis defined as:<br />
C-15
MAXIMUM PRINCIPAL STRESS<br />
SCF = --------------------------------------<br />
NOMINAL STRESS IN GIRDER<br />
(P/A+M/S)<br />
The moment M Is due to a combination of moment and shear load.<br />
The maximum SCF was found in the gusset plate connecting the<br />
transverse girder and column top, at the edge of the gusset plate in<br />
the weld between the gusset web and flange. It was equal to 1.66. The<br />
SCF in the longitudinal girder at the windlass cutouts reached a<br />
value of 1.87. Figure C.4-4 shows the equivalent stress variation<br />
over the entire connection. The maximum stress, as already nokl,<br />
occurs in the crotch region. Figure C.4-5 shows an equivalent stress<br />
contour plot of the windlass holes. Table C.4-1 summarizes the SCFS.<br />
,.—.<br />
C-16
,. ..<br />
..-. __ _________ ._ _ _______ _______ ______<br />
------ ________ _ _________ __________<br />
I MEMBER LOCATION DIRECTION SCF I<br />
I<br />
TO WELD I<br />
I<br />
_____________________<br />
-----_______________ __ _______________________ _--_+-+__________________<br />
1<br />
I Center Column Middle of Parallel 1.66 I<br />
I Transverse Gusset<br />
I<br />
I<br />
I<br />
I Girder-Column Gusset-Girder Perpendicular 1.37 I<br />
I Connectlon Connection<br />
I<br />
I<br />
I<br />
I 2.3x2.3x1.1 m Gusset-Column Perpendicular 1.10 I<br />
I Gusset Connection<br />
I<br />
I<br />
I<br />
I<br />
Girder Flange Parallel 1.05 I<br />
I<br />
I -------------------_-...-------------------------------<br />
I<br />
I Longitudinal Middle of Parallel 1.52 I<br />
I Girder-Column Gusset<br />
I<br />
I Connection Gusset-Girder Perpendicular 1.14 I<br />
I<br />
Connection<br />
I<br />
I<br />
I 1.lxl.lx.55m Gusset-Column Perpendicular 1.05 I<br />
i<br />
Connection<br />
I<br />
I<br />
I<br />
I Girder Flange Parallel 1.01 I<br />
I<br />
I<br />
I------------------------ ----- .s.-.-+.- -------- ---------- i<br />
I<br />
I<br />
I EXteri or Bottom Right Parallel 1.87 I<br />
I Longitudinal Corner of<br />
I Girder Exterior Hole I<br />
i<br />
I<br />
] @ Windlass Upper Left Parallel 1.73 I<br />
I Holes Corner of<br />
I<br />
Interior Hole<br />
I<br />
I<br />
===-==============------—--------------------------------========<br />
Table C.4-1<br />
Summary of SCFS for a Column - Girder Connection
‘-W,<br />
.\ I Ill I<br />
: II m<br />
-,,<br />
!)<br />
.—. --, — .’<br />
.— --- .<br />
1<br />
—..<br />
—..—-<br />
LOCATION OF<br />
FINITE ELEMENT<br />
MODEL<br />
=-mL<br />
-.— ---— .<br />
-4<br />
Figure C.4-1<br />
Overall Geometry of Vessel and Lacation of<br />
Column-to-Girder Connection<br />
— --, —
—..—.—<br />
-.. — 7<br />
a<br />
.<br />
LOWER HUL AN(I TuBULAR ERhCING PLAN B-Al<br />
o<br />
Hl<br />
.---------- — +----- . -..t-- . 5=1===-<br />
‘:&<br />
.. . . - ! .. ------- . . . . . . . . . . . . . . . ---------<br />
. . . . . . ..- . . . . . . . ------- . . . . . . . -------
’ \./<br />
““’ w~”mm<br />
m ““-<br />
Y<br />
--+\<br />
/<br />
Figure C.4-3<br />
Finite Element Model and Dimensions of<br />
Col~-to-Girder Connection
.<br />
Loading ~<br />
Figure C.4-4<br />
Equivalent Stress Contour Plot for the<br />
Column-to-Girder Connection
. .<br />
I<br />
Column<br />
L’<br />
Longitudinal<br />
Girder><br />
Figure C.4-5 Equivalent Stress Contour Plot Of Windlass Holes
C.5 REFERENCES<br />
C.1<br />
Kuang, J.G. et al., “stress Concentrations in Tubular<br />
Joints,” Proceedingsof Offshore Technology Conference, OTC<br />
Paper No. 2205, Houston, TX. 1975 (Revisions introduced in<br />
SPE Journal, pp. 287-299, August 1977).<br />
C.2<br />
Wordsworth, A.C., “Stress ConcentrationFactors at K and KT<br />
Tubular Joints,” Fatigue in Offshore Structural Steels,<br />
Institutionof Civil Engineers, London, February 1981.<br />
C.3<br />
Wordsworth, A.C., “Aspects of Stress Concentration Factors<br />
at Tubu1ar Joints,” TS8, Proceedings of the 3rd<br />
International Offshore Conference on Steel in Marine<br />
<strong>Structure</strong>s, (SIMS 87), Delft, The Netherlands, June 1987.<br />
C*4<br />
Gibstein, M.B., “Parametric Stress Analyses of T Joints,”<br />
Paper No. 26, European Offshore Steels Research Seminar,<br />
Cambridge, November 1978.<br />
..—.,<br />
C.5<br />
Gibstein, M.B., “Stress Concentration in Tubular K Joints<br />
with Diameter Ratio Equal to One,” TS1O, International<br />
Offshore Conference on Steel in Marine <strong>Structure</strong>s (SIMS<br />
87), Delft, The Netherlands, 1985.<br />
C.6<br />
Efthymiou, M., et al. “stress concentrations in T/Y and<br />
Gap/Overlap K Joints,” The 4th InternationalConference on<br />
Behavior of Offshore <strong>Structure</strong>s (BOSS 1985), Amsterdam, The<br />
Netherlands, 1985.<br />
C.7<br />
Marshall, P.W. and Luyties, W.H., “Allowable stresses for<br />
Fatigue Design,” International Conference on Behavior of<br />
Offshore <strong>Structure</strong>s (BOSS 1982), Cambridge, Massachusetts,<br />
August 1982.<br />
C-17
C.8 Underwater Engineering Group, “Design of Tubular Joints for<br />
Offshore <strong>Structure</strong>s,” UEG Publication, UR33, 1985.<br />
C.9 Kellogg, M.W., “Design of Piping Systems”, Second Edition,<br />
Wiley, 1956.<br />
C.lo<br />
Delft, van D.R.V. et al. “The Results of the European<br />
Fatigue Tests on Welded Tubular Joints Compared with SCF<br />
Formulas and Design Lives,” Paper No. TS 24, Proceedings of<br />
the 3rd International Conference on Steel in Marine<br />
<strong>Structure</strong>s (SIMS 87), Delft, The Netherlands, June 1987.<br />
C*11 Ma, S.Y., Tebbett, I.E., “Estimations of Stress<br />
Concentration Factors for Fatigue Design of Welded Tubular<br />
Connections,” Proceedings of the 20th Annual Offshore<br />
Technology Conference, OTC Paper No. 5666, Houston, TX.,<br />
May 1988.<br />
C.12 Tebbett, I.E., and Lalani, M., “A New Approach to stress<br />
Concentration Factors for Tubular Joint Design,”<br />
Proceedings of 16th Annual Offshore Technology Conference,<br />
ITC Paper No. 4825, Houston, TX, May 1984.<br />
C.13 Tolloczko, J.A., and<br />
Data on the Fatigue<br />
Proceedings of the<br />
Conference, OTC Paper<br />
Lalani, M., “The Implications of New<br />
Life Assessment of Tubular Joints,”<br />
20th Annual Offshore Technology<br />
No. 5662, Houston, TX, May 1988.<br />
C-18
(THIS PAGE INTENTIONALLY LE~ BLANK)
APPENDIX D<br />
VORTEX SHEDDING AVOIDANCE AND FATIGUE DAMAGE COMPUTATION<br />
CONTENTS<br />
NOMENCLATURE<br />
D.<br />
VORTEX SHEDDING<br />
D.1<br />
D*2<br />
D.3<br />
D.4<br />
D.5<br />
D*6<br />
D.7<br />
D*8<br />
D.9<br />
INTRODUCTION<br />
VORTEXSHEDDINGPARAMETERS<br />
SUSCEPTIBILITY TO VORTEXSHEDDING<br />
D.3.1 In-Line Vortex Shedding<br />
0.3.2 Cross-Flow Vortex Shedding<br />
D.3.3 Critical Flow Velocities<br />
AMPLITUDES OF VIBRATION<br />
0.4.1 In-Line Vortex Shedding Amplitudes<br />
0.4.2 Cross-Flow Vortex Shedding Amplitudes<br />
STRESSES DUE TO VORTEX SHEDDING<br />
FATIGUE LIFE EVALUATION<br />
EXAMPLE PROBLEMS<br />
D*7.1 Avoidance of Wind-Induced Cross-Flow Vortex Shedding<br />
D.7.2 Analysis for Wind-Induced Cross-Flow Vortex Shedding<br />
METHODS OF MINIMIZING VORTEX SHEDDING OSCILLATIONS<br />
0.8.1 Control of Structural Design<br />
D.8.2 Mass and Damping<br />
0.8.3 Devices and Spoilers<br />
REFERENCES
NOMENCLATURE<br />
co<br />
CLj<br />
CLO<br />
o<br />
‘tot<br />
DI<br />
D2<br />
E<br />
H<br />
I<br />
I<br />
Io<br />
K<br />
KS<br />
L<br />
N<br />
Re<br />
s<br />
s<br />
SCF<br />
‘t<br />
s~<br />
T<br />
Te<br />
v<br />
Vm<br />
v max<br />
v min<br />
Vr<br />
Y<br />
Ym<br />
Y~<br />
Coefficient of drag<br />
Design lift coefficient<br />
Base lift coefficient<br />
Fatigue damage<br />
Total fatigue damage<br />
Fatigue damage due to vortex shedding<br />
Fatigue damage due to storm<br />
Modulus of elasticity<br />
Submerged length of member<br />
Member moment of inertia<br />
Turbulence parameter<br />
Turbulence parameter<br />
Constant representingmember fixity<br />
Stability parameter<br />
Span between member supports<br />
Number of cycles to failure at hot spot stress range<br />
Reynolds number<br />
Hot spot stress range<br />
Member section modulus<br />
Stress concentration factor<br />
Strouhal number<br />
Correspondinghot spot stress range<br />
Wave period<br />
Time for which Vmin is exceeded<br />
Flow velocity normal to member axis<br />
Maximum orbital velocity due to wavemotion<br />
Maximum water particle velocity<br />
Minimum Vr, required for motion<br />
Reduced velocity<br />
Member midspan deflection<br />
Maximum member midspan deflection<br />
Refined maximum member midspan deflection
a<br />
an<br />
b<br />
d<br />
f“<br />
f ar<br />
f~<br />
‘bmax<br />
f~<br />
fn<br />
f~<br />
m<br />
z<br />
‘i<br />
‘j<br />
n<br />
‘e<br />
,--- no<br />
‘s<br />
‘w<br />
t<br />
v<br />
‘cr<br />
Vr<br />
w<br />
‘o<br />
WI<br />
y(x)<br />
y’(x)<br />
6<br />
E<br />
v<br />
‘N<br />
P<br />
Maximum modal amplitude<br />
Natural frequency coefficient<br />
Pit depth<br />
Member diameter<br />
Member vortex shedding frequency<br />
Turbulence parameter<br />
Member bending stress<br />
Member maximum bending stress<br />
Member maximum hot spot stress<br />
Member natural frequency<br />
Member vortex shedding frequency<br />
Mass of member per unit length excluding marine growth<br />
Effective mass per unit length<br />
Mass of member per unit length includingmarine growth<br />
Generalized mass per unit length for mode j<br />
Mode of vibration<br />
Member end condition coefficient<br />
Total number of occurrences per year<br />
Actual number of cycles at hot spot stress range<br />
Number of oscillations during one wave cycle<br />
Nominal caisson thickness<br />
Applied velocity<br />
Critical wind velocity<br />
Reduced velocity<br />
Load per unit length<br />
Weight per unit length of member<br />
Weight per unit length of supported Items<br />
Fundamentalmode shape<br />
Equivalent fundamental mode shape<br />
Logarithmic decrement of damping<br />
Damping ratio<br />
Kinematic viscosity<br />
Ratio of midspan deflection to member diameter (Y/d)<br />
Mass density of fluid
x<br />
(THIS PAGE INTENTIONALLY LEFT BIANK)
D. VORTEX SHEDDING<br />
0.1. INTRODUCTION<br />
When a fluid flows about a stationary cylinder, the flow separates,<br />
vortices are shed, and a periodic wake is formed. Each time a vortex<br />
is shed from the cylinder, the local pressure distribution is<br />
altered, and the cylinder experiences a time-varying force at the<br />
frequency of vortex shedding.<br />
In steady flows, vortices are shed alternately from either side of<br />
the cylinder producing an oscillating lift force transverse to the<br />
flow direction at a frequency equal to that at which pairs of<br />
vortices are shed. In the flow direction, in addition to the steady<br />
drag force, there is a small fluctuating drag force associated with<br />
the shedding of individual vortices at a frequency twice that of the<br />
lift force.<br />
As the flow velocity increases, the vortex shedding frequency<br />
increases. Thus, provided the flow velocity is high enough, a<br />
condition will be reached where the vortex shedding frequency<br />
coincides with the natural frequency of the flexible element.<br />
—<br />
In general, marine members and appurtenant pipework are of a diameter<br />
and length that preclude the occurrence of in-line vibrations induced<br />
by vortex shedding. However, all susceptible members must be<br />
analyzed to ensure that the stresses due to in-line vibrations and<br />
possible synchronized oscillations are small and do not result in a<br />
fatigue failure.<br />
Response to vortex shedding cannot be predicted using conventional<br />
dynamic analysis techniques since the problem is non-linear. The<br />
motion of the structure affects the strength of the shedding which,<br />
in turn affects the motion of the structure. This feedback mechanism<br />
causes the response to be either significantly large or negligibly<br />
smal1. Once excited, there is also a tendency for the vortex<br />
D-1
shedding frequency to synchronize with the natural frequency of the<br />
structure. This results in sustained resonant vibration even if the<br />
flow velocity moves away from the critical velocity.<br />
Oscillations can be predominantly in-line with the flow direction or<br />
transverse to it. In-line motion occurs at lower flow velocities<br />
than transverse or cross-flow motion, but the latter is invariably<br />
more severe and can lead to catastrophic failure due to a small<br />
number of cycles of oscillation.<br />
Response to vortex shedding is further complicated as the<br />
excitational force is not necessarily uniform along the length of the<br />
members and the actual amplitude of oscillation depends to a large<br />
extent on the degree of structural damping.<br />
0.2. VORTEX SHEDDING PARAMETERS<br />
A number of parameters are common to this phenomenon:<br />
Reduced velocity (Vr)<br />
Vr = V/fnd<br />
where:<br />
v = flow velocity normal to the member axis<br />
fn = fundamental frequency of the member (Hz)<br />
d = diameter of the member<br />
Reynolds number (Re)<br />
where:<br />
Re = Vdi v<br />
w = kinematic viscosity of the fluid<br />
The Strouhal number (St) is a function of the Reynolds number for<br />
circular members. The Reynolds number for typical cylindrical<br />
members under storm current ranges from 3.5 x 105 to 1.0 x 106. The<br />
Strouhal number is reasonably approximated as 0.21 for this range of<br />
Reynolds numbers.<br />
D-2
Vortex Shedding Frequency<br />
(f”)<br />
fv=~=<br />
St v<br />
vortex<br />
shedding frequency of the member<br />
If<br />
the vortex shedding frequency of the member coincides with<br />
natural frequency of the member, resonance will occur.<br />
the<br />
Stability parameter<br />
(Ks)<br />
21iI& / P d 2<br />
21tE = logarithmic decrement<br />
damping ratio<br />
/--<br />
mass density of the f’ uid<br />
Iii=<br />
effective mass per un” t length<br />
~~(m)[y(x)]2dx<br />
$ [Y’ (X)]%<br />
L<br />
span between member supports<br />
m<br />
mass of member per unit length<br />
y(x),<br />
&y’(x) = fundamental mode shapes as a function of the<br />
ordinate x measured from the lower support<br />
along the longitudinalaxis of the member<br />
As<br />
given in References 0.1 and 0.2, the effective mass is used to<br />
equate the real structure with an equivalent structure for which<br />
o-3
deflection and stability parameters are known. The deflected form of<br />
this equivalent structure is a cantilever, while typical structure<br />
members and appurtenancesdeflect as a simply supported beam. Hence,<br />
the equivalent structure has a mode shape given by:<br />
y’(x) =<br />
a- a cos(~)<br />
while the real structure has a mode shape given by:<br />
y(x) =<br />
a sin(~)<br />
Substituting into the effective mass formulation,we obtain:<br />
m=<br />
f~[m][a sin ~]2dx<br />
J: [a - a cos ~]2dx<br />
where:<br />
a = maximum modal amplitude<br />
Integration of the above equation leads to the relationship:<br />
i= 2.205 m for simple supported span<br />
i= 1.654 m for fixed supports<br />
ii=<br />
m for cantilever span<br />
Damping Ratio<br />
Welded marine structures exhibit very low values of structural<br />
damping. Vibratory energy is typically dissipated by material and<br />
aerodynamic (radiation) damping. Individual members subjected to<br />
large vibratory motions dissipate energy through the connections to<br />
the main structure largely as dispersive bending and compression<br />
D-4
waves. When only isolated members undergo large vibration response,<br />
energy dispersion exceeds reflected energy and represents a major<br />
source of damping.<br />
Structural members may be grouped Into two classes, depending on the<br />
fixity of their supports. Tubular braces welded on to regions of<br />
high rigidity, such as structure columns or legs, are defined as<br />
Class 1 members. Tubular braces welded on to regions of low<br />
rigidity, such as other braces, are defined as Class 2 members. The<br />
damping ratio applicable for structural members are:<br />
Structural Member - Class<br />
Structural Member - Class<br />
1 Damping ratio E = 0.0035<br />
2 Damping ratio E = 0.0015<br />
Although the recommended damping ratios are for vibrations in air,<br />
they may be conservativelyused for vibrations in water.<br />
Non-structural continuous members, such as tubulars supported by<br />
multiple guides, have both structural and hydrodynamic damping. The<br />
hydrodynamic damping occurs due to sympathetic vibration of spans<br />
adjacent to the span being evaluated for shedding. Recent work by<br />
Vandiver and Chung (Reference 0.3) supports the effectiveness of<br />
hydrodynamic damping mechanism. The lower bound structural damping<br />
ratio for continuous tubulars supported by loose guides is given as<br />
0.009 by Blevins (Reference0.4). The applicable damping ratios are<br />
assumed to be:<br />
Non-StructuralMembers - Continuous Spans<br />
Damping ratio E = 0.009 in air<br />
Damping ratio E = ().02 in k@t@r<br />
Natural Frequency<br />
The fundamental natural frequency (in Hz) for uniform beams may be<br />
calculated from:<br />
D-5
a<br />
fn = ~ (EI/mi L4)%<br />
where:<br />
the<br />
moment of inertia of the beam<br />
3.52 for a beam with fix-free ends (cantilever)<br />
9.87 for a beam with pin-pin ends<br />
15.4 for a beam with fix-pin ends<br />
22.4 for a beam with fix-fix ends<br />
1ength<br />
mode of vibration<br />
mass per unit length<br />
The amount of member fixity assumed in the analysis has a large<br />
effect on vortex shedding results, because of its impact on member<br />
stiffness, natural period, amplitude of displacement, and member<br />
stress. Hence, careful consideration should be given to member end<br />
conditions. Members framing into relatively stiff members can<br />
usually be assumed to be fixed. Other members, such as caissons and<br />
risers, may act as pinned members if supports are detailed to allow<br />
member rotation.<br />
For members with non-uniform spans, complex support arrangements or<br />
non-uniform mass distribution, the natural frequency should be<br />
determined from either a dynamic analysis or from Tables provided in<br />
References D.5 and D.6. Reid (Reference D.7) provides a discussion<br />
and a model to predict the response of variable geometry cylinders<br />
subjected to a varying flow velocities.<br />
The natural frequency of a member is a function of the member’s<br />
stiffness and mass. For the purposes of vortex shedding analysis and<br />
design, the member’s stiffness properties are computed from the<br />
D-6
member’s nominal diameter and thickness. The member mass per unit<br />
length m is taken to include the mass of the member steel including<br />
sacrificial corrosion allowance, anodes, and contained fluid. For<br />
the submerged portion of the member, the added mass of the<br />
surrounding water is also included. This added mass is the mass of<br />
water that would be displaced by a closed cylinder with a diameter<br />
equal to the nominal member outside diameter plus two times the<br />
appropriatemarine growth thickness.<br />
Because of insufficient knowledge of the effect of marine growth on<br />
vortex shedding, the member diameter “d” in vortex-shedding<br />
parameters Vr, Re, KS, and the member effective mass = in parameter<br />
Ks do not incluc(e any allowance for the presence of marine growth.<br />
D.3.<br />
SLISCEPTIBILITYTOVORTEX SHEDDING<br />
,/—.<br />
The vortex shedding phenomena may occur either in water or in air.<br />
The susceptibility discussed and the design guidelines presented are<br />
applicable for steady current and wind. Wave induced vortex<br />
shedding has not been investigated in depth. Since the water<br />
particle velocities in waves continually change both in magnitude and<br />
direction (i.e. restricting resonant oscillation build-up), it may be<br />
reasonable to investigate current-induced vortex shedding and<br />
overlook wave actions.<br />
To determine susceptibility of a member to wind- or current-induced<br />
vortex shedding vibrations, the reduced velocity (Vr) is computed<br />
first. For submerged members, the stability parameter (Ks) is also<br />
calculated. Vortex shedding susceptibility defined here is based<br />
upon the method given in Reference D.8, with a modified lower bound<br />
for current-induced shedding to reflect present thinking on this<br />
subject (ReferenceD.9).<br />
D-7
0.3.1 In-Line Vortex Shedding .<br />
In-1ine vibrations in wind and current environments may occur when:<br />
Current Environment<br />
Wind Environment<br />
1.2 ~ Vr< 3.5 1.7 < Vr< 3.2<br />
and KS 51.8<br />
The value of Vr may be more accurately defined for low KS values from<br />
Figure O-1, which gives the reduced velocity necessary for the onset<br />
of in-line motion as a function of combined mass and damping<br />
parameter (i.e. stability parameter). Corresponding amplitude of<br />
motion as a function of K5 is given on Figure D-2. As illustratedon<br />
this Figure, in-line motion is completely supressed for Ks values<br />
greater than 1.8.<br />
Typical marine structure members (i.e. braces and caissons on a<br />
platform) generally have values of Ks greater than 1.8 in air but<br />
less than 1.8 in water. Hence, in-line vibrations with significant<br />
amplitudes are often likely in steady current but unlikely in wind.<br />
0.3.2 Cross-Flow Vortex Shedding<br />
The reduced velocity necessary for the onset of cross-flow vibrations<br />
in either air or in water is shown on Figure D-3 as a function of<br />
Reynold’s number, Re, cross flow vibrations in water and in air may<br />
occur when:<br />
Current Environment<br />
Wind Environment<br />
3.95vr59<br />
and Ks s 16<br />
4.7
\Lf-!<br />
of cycles. Thus, the reduced velocity necessary for the onset of<br />
cross-flow vibrations in steady current should be avoided.<br />
0.3.3 Critical Flow Velocities<br />
The criteria for determining the critical flow velocities for the<br />
onset of VW can be expressed in terms of the reduced velocity<br />
(Section D.2):<br />
v cr = (Vr)cr (fn* d)<br />
where:<br />
(Vr) Cr = 1.2 for in-line oscillations in water<br />
= 1.7 for in-line oscillations in air<br />
= 3.9 for cross-flow oscillations in water<br />
=<br />
4.7 for cross-flow oscillations in air<br />
0.4. AMPLITUDES OF VIBRATION<br />
Amplitudes of vibrations can be determined by several methods. A DnV<br />
proposed procedure (Reference 0.8) is simple to apply and allows<br />
determination of member natural frequencies, critical velocities and<br />
maximum amplitudes of vortex-shedding induced oscillations. The<br />
procedure yields consistent results, comparable to the results<br />
obtained by other methods, except for oscillation amplitudes. The<br />
DnV calculation of oscillation amplitudes is based on a dynamic load<br />
factor of a resonant, damped, single-degree-of-freedomsystem. this<br />
approach is not valid unless the nonlinear relationship between the<br />
response and damping ratio is known and accounted for. Consequently,<br />
in-line and cross-flow vortex shedding amplitudes are assessed<br />
separately.<br />
D-9
0.4.1 In-LineVortex Sheddinq Amplitudes<br />
The reduced velocity and the amplitude of vibrations shown on Figures<br />
D-1 and D-2, respectively, as functions of stability parameter are<br />
based on experimental data. The experimental data obtained are for<br />
the cantilever mode of deflection for in-line and cross-flow<br />
vibrations.<br />
Sarpkaya (Reference D.1O) carried out tests on both oscillatory flow<br />
and uniform flow and observed smaller amplitudes of vibration for the<br />
oscillatory flow than for the uniform flow. It is also suggested by<br />
King (Reference 0.1) that the maximum amplitude for an oscillatory<br />
flow is likely to occur at a Vr value in excess of 1.5 (as opposed to<br />
1.0 assumed by DnV) and that an oscillation build-up of about 15<br />
cycles is required before “lock-in” maximum-amplitude vibration<br />
occurs. In light of this evidence, the amplitude of vibrations shown<br />
in Figure D-2 is based on Hallam et al (Reference D.2) rather than<br />
the DnV (ReferenceD.8).<br />
Since typical marine structure members have stability parameters (Ks)<br />
in excess of 1.8, in-line vibrations of these members in air are<br />
unlikely.<br />
D.4.2<br />
Cross-flow Vortex Shedding Amplitudes<br />
The amplitude of the induced vibrations that accompanies cross-flow<br />
vibration are generally large and creates very high stresses.<br />
Therefore, it is desirable to preclude cross-flow induced<br />
vibrations. Figure D-4 illustrates a curve defining the amplitude of<br />
response for cross-flow vibrations due to current flow and based on a<br />
cantilevermode of deflection.<br />
Cross-flowoscillations in air may not be always avoidable, requiring<br />
the members to have sufficient resistance. The DnV procedure<br />
(Reference 0.8) to determine the oscillation amplitudes is derived<br />
from a simplified approach applicable to vortex shedding due to<br />
D-lo
s ;<br />
J<br />
steady current, by substituting the mass density of air for the mass<br />
density of water. Hence, the oscillation amplitude is not linked<br />
with the velocity that causes vortex-induced motion. The resulting<br />
predicted amplitudes are substantially higher than amplitudes<br />
predicted based on an ESDU (Reference D.11) procedure that accounts<br />
for interactionbetween vortices shed and the forces induced.<br />
The iterative ESDU procedure to determine the amplitudes can be<br />
simplified by approximating selected variables. The peak amp1itude<br />
is represented in Equation 9 of the ESDU report by.<br />
●<br />
_=nN=— Y 0.00633 ~ —— d2 1 ~L. _ 0“0795 cLj<br />
d E ITlj<br />
KS $:<br />
Using this formulation, a corresponding equation can be established<br />
for a structure, while making assumptions about the individual<br />
parameters. Following step 3 of the procedure, the parameters may be<br />
set as:<br />
‘j =<br />
generalized mass/unit length for mode j<br />
= 2.205 m for pinned structure,<br />
= 1.654 m for fixed structure<br />
KS = stability parameter = y<br />
pd<br />
P<br />
= mass density of air = 1.024 kg/m3<br />
6 = decrement of damping = 21rE<br />
E = damping parameter = 0.002 for wind<br />
% =<br />
Strouhal Number = 0.2<br />
Q-11
CLO = base lift coefficient = 0.29 high Reynolds number<br />
= 0.42 low Reynolds number<br />
CLj = design lift coefficient = CLO X farX 10 X+ x 1.2<br />
f ar =<br />
turbulence parameter = 1.0<br />
~=<br />
turbulence parameter = 1.0<br />
10 = turbulence parameter = 0.45<br />
Evaluating the equation based on the high Reynolds number (Re ><br />
500,000) leads to:<br />
nN = 0.0795 (0.29 )(1.0 )(0.45 )(1.0 )(1.2 )/[ K~(0.2)2]<br />
or<br />
ON =<br />
~ (high Reynolds number, Re > 500,000)<br />
s<br />
0.4510<br />
~N ‘~<br />
(low Reynolds number, Re < 500,000)<br />
The amplitude can also be determined iteratively by utilizing the<br />
ESDU recommended turbulence parameter and following steps 1 through<br />
5.<br />
Step 1: Determine correlation length factor, l.. Depending cm the<br />
end fixity, 10<br />
is:<br />
10 = 0.66 for fixed and free (cantilever)<br />
= 0.63 for pin and pin (simple beam)<br />
= 0.58 for fixed and pin<br />
= 0.52 for fixed and fixed<br />
step 2: Assume 1/10 = 1.0 and calculate the amplitude.<br />
D-12
Step 3: Obtain a new valueof 1/10 based on initial<br />
amplitude.<br />
Step 4: Recompute the amplitude based on<br />
the new value of l/l..<br />
Step 5: Repeat Steps 3 and 4 until convergence.<br />
0.5. STRESSES DUE TO VORTEXSHEDDING<br />
Once the amplitude of vibration has been calculated, stresses can be<br />
computed according to the support conditions. For a simply supported<br />
beam with a uniform load w, the midspan deflection Y, and the midspan<br />
bending stress fb are given as follows:<br />
w<br />
Y<br />
a<br />
= 5 WL41<br />
m “T*T<br />
=<br />
384 EIY<br />
T“ ~<br />
~ax<br />
=<br />
WL2= 384 EIY<br />
T m— ~2<br />
Md EDY<br />
‘bmax = ~*~=4”8~ at midspan<br />
Expressing fbmax =<br />
K . EDY<br />
~<br />
The K value varies with support conditions and location as shown on<br />
Table D-1.<br />
Fixity Mid-Span Ends<br />
Fix Fix 8.0 16.0<br />
Fix Pin 6.5 11.6<br />
Pin Pin 4.8 0<br />
Fix Free N.A. 2.0<br />
Table D-1 K Values Based on Fixity and Location<br />
D-13
The vortex shedding bending stress is combined with the member axial<br />
and bending stresses due to global deformation of the marine<br />
structure.<br />
0.6. FATIGUE LIFE EVALUATION<br />
The fatigue life evaluation can be carried out in a conservative twostep<br />
process. First, the fatigue damage due to the vortex-induced<br />
oscillations is calculated as D1. Second, a deterministic fatigue<br />
analysis is performed by computer analysis. Hot spot stress range vs<br />
wave height (or wind velocity) for the loading directions considered<br />
is determined from the computer analysis. The critical direction is<br />
determined and a plot is made. From the plot of hot spot stress<br />
range vs wave height (or wind velocity), the stress ranges for the<br />
fatigue waves are determined. The maximum vortex-induced stress<br />
ranges for the fatigue environment are added to the deterministic<br />
fatigue stress ranges. Then, the standard deterministic fatigue<br />
analysis is performed using the increased stress range. The fatigue<br />
damage calculated in this second step is D2. Therefore the total<br />
fattgue damage is equal to the sum of D1 and D2, or Dtot = 01 + 02.<br />
The fatigue life fin Years is therefore calculated as l/Btot.<br />
A typical fatigue life evaluation procedure is given below:<br />
Step 1:<br />
a.<br />
Calculate the natural frequency fn (Hz) of the member.<br />
b.<br />
Calculate the stability parameter of the member.<br />
K5=~<br />
pd2<br />
c.<br />
Determine the minimum Vr required for vibrations based on Ks in<br />
Figure D-1.<br />
d.<br />
Calculate Vmin, the minimum velocity at which current- or wind-<br />
0-14
vortex shedding will occur, i.e., Vmin = ‘r(req’d) x ‘n x ‘-<br />
e.<br />
Check the applied velocity profile to see if Vmax is greater<br />
than Vmin. If Vmax is less than Vmin, then no vortex<br />
oscillations can occur.<br />
f.<br />
For Vmax greater than Vmin, vortex oscillations can occur. The<br />
displacement amplitude is based on stability parameter Ks, and<br />
is determined from Figure O-2 for in-line vibration. A<br />
conservative approach is used to determine Y/d vs Ks. For Ks <<br />
0.6 the first instability region curve is used. For Ks > 0.6<br />
the second instability region curve is used. This<br />
conservatively represents an envelope of maximum values of Y/d<br />
vs Ks from Figure O-2. Displacement amplitude is normalized to<br />
Y/d.<br />
9*<br />
Given (Y/d), calculate the bending stress, fb.<br />
h.<br />
Multiply bending stress fb by an SCF of 1.S to produce hOt SpOt<br />
stress fH. A larger<br />
SCF will be used where necessary.<br />
i.<br />
From the maximum hot spot stress, the hot spot stress range is<br />
calculated as 2fH.<br />
j.<br />
Allowable number of cycles to failure (N) should be calculated<br />
using an applicable S-N curve (based on weld type and<br />
environment).<br />
k.<br />
Assume conditions conducive to resonant vortex shedding occur<br />
for a total time of T (seconds) per annum (based on current or<br />
wind data relevant to applicable loading condition).<br />
1.<br />
Hence, in time T, number of cycles n = fnT and the cumulative<br />
damage D1 = n/N = fnT/N in one year.<br />
Step 2:<br />
D-15
a. Depending on marine structure in service conditions (i.e.<br />
structure in water or in air) run an applicable loading<br />
analysis. Assuming a marine environment, run a storm wave<br />
deterministic fatigue analysis and obtain the results of hot<br />
spot stress range vs wave height for the wave directions<br />
considered and as many hot spots as are needed.<br />
b. Determine the critical hot spot and wave direction and draw the<br />
hot spot stress range vs wave height graph.<br />
c. Determine the hot spot stress range for each of the fatigue<br />
waves.<br />
d. For the larger fatigue waves in which vortex-induced<br />
oscillations occur, add the increase in stress range due to<br />
vortex-induced oscillations to the stress range from the<br />
deterministic fatigue analysis.<br />
e. Calculate the fatigue damage 02 over a 1 yr period for the full<br />
range of wave heights:<br />
f. Calculate the total fatigue damage:<br />
Dtot = DI + D2<br />
9= Calculate the fatigue life in years as:<br />
Life =~<br />
‘tot<br />
D-16<br />
, ,
h. The fatigue life may -be modified to include the effects of<br />
corrosion pitting in caissons. Corrosion pitting produces an<br />
SCF at the location of the pit. The SCF is calculated as:<br />
SCF=~+—<br />
3 (:)<br />
(1-# (1-$2<br />
where:<br />
b = plt depth<br />
t<br />
= nominal caisson thickness<br />
The new life including corrosion damage is calculated as:<br />
Old Life<br />
New Life =—<br />
(SCF)3<br />
This estimate of fatigue damage can, if necessary, be refined by<br />
consideration of the number of wave occurrences for different<br />
directions and evaluation of the damage at a number of points<br />
around the circumference of the member.<br />
0.7. EXAMPLE PROBLEMS<br />
0.7.1 Avoidance of Wind-Induced Cross-Flow Vortex Sheddinq<br />
It<br />
can be shown that for a steel beam of circular cross section, the<br />
following relationship holds:<br />
where:<br />
c. Vrn~ W.<br />
v r— ]4<br />
cr = ‘W+WJ<br />
(L/d)2 0 1<br />
v cr =<br />
critical wind<br />
for the onset<br />
shedding<br />
velocity of the tubular necessary<br />
of cross-flow wind-induced vortex<br />
D-17
c<br />
constant (See<br />
Vr<br />
reduced velocity<br />
‘e<br />
member end efficiency<br />
1.5 fixed ends<br />
1.0 pinned ends<br />
‘o<br />
weight per unit length of tubular<br />
WI<br />
weight per unit length of supported item (e.g.,<br />
anodes)<br />
L<br />
beam length<br />
d<br />
tubular mean diameter<br />
For Vr = 4.7, ne =<br />
1.5 (fixed condition), and WI = O, this reduces<br />
to:<br />
v = 97240/(L/d)~ ft/sec<br />
Cr = 29610/(L/d) m/see<br />
Hence, if maximum<br />
setting all brace<br />
cross-flow vortex<br />
are required.<br />
expected wind speed is 65.6 ft/s (20 m/s), then<br />
L/d ratios at 38 or less precludes wind-induced<br />
shedding, and no further analyses or precautions<br />
However, maximum wind speeds may be so high that the above approach<br />
may be uneconomical. In this case, either precautionary measures<br />
must be taken or additional analyses considering strength and fatigue<br />
must be undertaken.<br />
D-18<br />
./<br />
‘-. -7<br />
[ J
NOTE:<br />
The relationship given is based on:<br />
44<br />
v cr<br />
=Vrfnd= Vr(~[EI/MiLl )d<br />
substituting<br />
an = (ne m)2<br />
ITli= (W. + Wi)/g<br />
I = T d3t/8<br />
E = 4176 X<br />
106 lbs/ft2 (200,000 MN/m2)<br />
9 = 32.2 ft/sec2 (9.806m/sec2)<br />
‘o = Ys ~dt<br />
= weight density of steel, 490 lbs/ft3 (0.077MN/m3)<br />
‘s<br />
2 .2<br />
Vcr= Vr (+) [ E (~d3t/8) ~]. + d<br />
(w. + Wl) L<br />
2,<br />
‘e E (wo/Ys) d2 9 ~<br />
Vcr. Vr (~) [ ].d<br />
8 (WO+-, WI) L4<br />
Substituting for E,<br />
s, and g<br />
v<br />
cr<br />
where<br />
=Vr.<br />
C ne2<br />
(L/d)2<br />
constant C = 9195 for Vcr as ft/sec<br />
= 2800 for V~r as m/see<br />
D-19
D.7.2 Analysis for Wind-Induced Cross-flow Vortex Shedding<br />
Using procedures discussed in Section D.4 a flare structure bracing<br />
members are analyzed for crossflow oscillationsproduced by vortex<br />
shedding. The analysis is performed using a Lotus spreadsheet. The<br />
general procedure is as follows:<br />
(a)<br />
Member and environmental parameters are input.<br />
(b) Critical velocity, peak amplitudes of oscillation and<br />
corresponding stress amplitudes are computed.<br />
(c) The time (in hours) of crossflow oscillation required to cause<br />
fatigue failure is computed.<br />
Analysis Description<br />
The following is a detailed description of the spread sheet input and<br />
calculation.<br />
(a) Spread Sheet Terminology<br />
Columns are labeled alphabetically while rows are labeled<br />
numerically. A “cell” is identified by referring to a specific<br />
row and column.<br />
(b) General Parameters<br />
The following are parameters common to all members analyzed as<br />
given at the top of the spread sheet.<br />
CELL C5: DAMPING RATIO = E<br />
CELL C6: AIR MASS DENSITY = ~<br />
0-20
CELL C7: KINEMATIC VISCOSITY = U<br />
CELL C8: STRESS CONCENTRATION FACTOR = SCF<br />
CELL C9: MODULUS OF ELASTICITY= E<br />
CELL K5: RATIO OF GENERALIZE MASS TO EFFECTIVE MASS = ( # )<br />
e<br />
CELL K6: FIXITY PARAMETER INFORMUIA FOR CRITICAL VELOCITY =<br />
‘e<br />
CELL K7: FIXITY PARAMETER INFORMULA FOR STRESS AMPLITUDE = C<br />
CELL K8: FIXITY PARAMETER INFORMULA FOR MEMBER FREQUENCY = a n<br />
(c) Specific Member Analysis<br />
The following describes the content of each column in analyzing<br />
a specific member. Entries and formulas for vortex shedding<br />
analysis of member group HI on line 16 are also provided.<br />
Formula coding is described in the LOTUS 1-2-3 Users Manual.<br />
COLUMN A:<br />
ENTER THE MEMBER GROUP IDENTIFIER<br />
COLUMN B:<br />
ENTER THE EFFECTIVE SPAN OF THE MEMBER = L (m)<br />
COLUMN C:<br />
ENTER THE OUTSIOE OIAMETER OF THE-TUBULAR = d (mm)<br />
COLUMN D:<br />
ENTER THE TOTAL OUTSIOE DIAMETER = D (mm) INCLUDINGAS<br />
APPLICABLE, MARINE GROWTH, FIRE PROTECTION, ETC.<br />
COLUMN E:<br />
ENTER THE TUBULAR WALL THICKNESS = t (mm)<br />
COLUMN F:<br />
ENTER ADOED MASS (kg/m), IF APPLICABLE<br />
COLUMN G:<br />
THE MOMENT OF INERTIA OF THE TUBULAR = I (cm4) IS<br />
COMPUTED.<br />
D-21
I . + [ (+)-L (+ - t)4] (cm4)<br />
COLUMN H: THE TOTAL EFFECTIVE MASS IS COMPUTED<br />
me = , [(+)~.( + - t)2] (0.785) +ma (kg/m)<br />
COLUMNI:<br />
THE CRITICAL VELOCITYFOR CROSSFLOW OSCILLATION IS<br />
COMPUTED.<br />
COLUMNJ:<br />
13160 n:<br />
v= cr<br />
(m/s)<br />
(L/D)2<br />
ENTER THE THRESHOLDWINDVELOCITY= ‘thr<br />
COLUMN K: THE STABILITY PARAMETER IS COMPUTED<br />
Ks =<br />
2me (2~E)<br />
PD2<br />
COLUMN L: THE REYNOLDS NUMBER IS COMPUTED<br />
Before performing calculation in the following columns, the<br />
velocity is compared with the threshold velocity. If the<br />
velocity is larger, crossflow oscillations will not occur<br />
computations are supressed. An “N.A.” is then inserted<br />
column.<br />
critical<br />
critical<br />
and the<br />
in each<br />
If the critical velocity is less than the threshold value, the<br />
following computations are performed.<br />
COLUMN M: THE AMPLITUDE OF VIBRATION IS COMPUTED<br />
Y=&<br />
(~) Ks<br />
Where a = 0.04925 for Re<br />
and a = 0.07178 for Re<br />
> 500,000<br />
< 500,000<br />
D-22
COLUMN N: THE STRESS AMPLITUDE IS COMPUTED<br />
fb=y<br />
(MPa)<br />
where C depends on beam end fixity (see Section D.4)<br />
COLUMN O: THE HOT SPOT STRESS RANGE IS COMPUTED<br />
S = 2 (SCF)fb<br />
(MPa)<br />
COLUMN P: THE NUMBER OF CYCLES TO FAILURE UNDER THE HOT SPOT<br />
STRESS RANGE IS COMPUTED.<br />
N<br />
=<br />
~o(14.57 - 4.1 LogloS) (cycles)<br />
COLUMN Q: THE MEMBER NATURAL FREQUENCY IS COMPUTED<br />
a<br />
fn=~ (— %?<br />
~E:4 )<br />
e<br />
(Hz)<br />
where an depends on beam end fixity (see Section D.2)<br />
COLUMN R: THE TIME IN HOURS TO FATIGUE FAILURE UNDER N CYCLES OF<br />
STRESS RANGE S IS COMPUTED<br />
T=+<br />
n<br />
D.8.<br />
METHODS OF MINIMIZING VORTEX SHEDDING OSCILLATIONS<br />
D.8.1<br />
Control of Structural Design<br />
The properties of the structure<br />
velocity values in steady<br />
oscillations.<br />
can be chosen to ensure that critical<br />
flow do not produce detrimental<br />
D-23
Experiments have shown- that for a constant mass<br />
parameter (m/pd2 = 2.0), the critical velocity depends mainly on the<br />
submerged length/diameter (L/d) ratio of the member.<br />
Thus, either high natural frequency or large diameter is required to<br />
avoid VIV’S in quickly flowing fluid. A higher frequency will be<br />
obtained by using larger diameter tubes, so a double benefit<br />
occurs. An alternative method of increasing the frequency is to<br />
brace the structurewith guy wires.<br />
0.8.2 Mass and Dampinq<br />
Increasing the mass parameter, m/PdZ, and/or the damping parameter<br />
reduces the amplitude of oscillations; if the increase is large<br />
enough, the motion is suppressed completely. While high mass and<br />
damping are the factors that prevent most existing structures from<br />
vibrating, no suitable design criteria are presently available for<br />
these factors, and their effects have not been studied in detail.<br />
Increasing the mass of a structure to reduce oscillatory effects may<br />
not be entirely beneficial. The increase may produce a reduction in<br />
the natural frequency (and hence the flow speeds at which oscillation<br />
will tend to occur). It is thus possible that the addition of mass<br />
may reduce the critical speed to within the actual speed range.<br />
However, if increased mass is chosen as a method of limiting the<br />
amplitude of oscillation, this mass should be under stress during the<br />
motion. If so, the mass will also contribute to the structural<br />
damping. An unstressedmass will not be so effective.<br />
If the structure is almost at the critical value of the combined<br />
mass/damping parameter for the suppression of motion, then a small<br />
additional amount of damping may be sufficient.<br />
D-24
D.8.3 Devices and Spoilers<br />
Devices that modify flow and reduce excitation can be fitted to<br />
tubular structures. These devices (see Figure D-5) work well for<br />
Isolated members but are less effective for an array of piles or<br />
cylinders. Unfortunately, there is no relevant information<br />
describing how the governing stability criteria are modified. The<br />
most widely used devices are described below.<br />
Guy Wires<br />
Appropriately placed guy wires may be used to increase member<br />
stiffness and preclude wind-induced oscillations. Guy wires should<br />
be of sufficient number and direction to adequately brace the tubular<br />
member; otherwise, oscillations may not be eljmlnatecl completely and<br />
additional oscillations of the guys themselves may occur.<br />
Strakes or Spoilers<br />
$trakes and spoilers consist of a number (usually three) of fins<br />
wound as a helix around the tubular. These have proven effective in<br />
preventing wind-induced cross-flow oscillations of structures, and<br />
there is no reason to doubt their ability to suppress in-line motion.<br />
provided that the optimum stroke design is used. This comprises a<br />
three-star helix, having a pitch equal to five times the member<br />
diameter. Typically each helix protrudes one-tenth of the member<br />
diameter from the cylinder surface. To prevent in-line motion,<br />
strakes need only be applied over approximately in the middle onethird<br />
of the length of the tubular with the greatest amplitude.<br />
Elimination of the much more violent cross-flow motion requires a<br />
longer strake, perhaps covering the complete length of tube. The<br />
main disadvantage of strakes, apart from construction difficulties<br />
and problems associated with erosion or marine growth, is that they<br />
increase the time-averaged drag force produced by the flow. The drag<br />
coefficient of the straked part of the tube is independent of the<br />
Reynold’s number and has a value of CD = 1.3 based on the tubular<br />
diameter.<br />
D-25
Shrouds<br />
Shrouds consist of an outer shell, separated from the tubular by a<br />
gap of about 0.10 diameter, with many small rectangular holes. The<br />
limited data available indicates that shrouds may not always be<br />
effective. The advantage of shrouds over strakes is that their drag<br />
penalty is not as great; for all Reynold’s numbers, CD= 0.9 based on<br />
the inner tubular diameter. Like strakes, shrouds can eliminate the<br />
in-line motion of the two low-speed peaks without covering the<br />
complete length of the tubular. However, any design that requires<br />
shrouds (or strakes) to prevent cross-flow motion should be<br />
considered with great caution. Their effectiveness can be minimized<br />
by marine growth.<br />
Offset Dorsal Fins<br />
This is the simplest device for the prevention of oscillations. It<br />
is probably the only device that can be relied upon to continue to<br />
work in the marine environment over a long period of time without<br />
being affected adversely by marine growth: It has some drag penalty,<br />
but this is not likely to be significant for most designs.<br />
The offset dorsal fin is limited to tubular structures that are<br />
subject to in-line motion due to flow from one direction only (or one<br />
direction and its reversal, as in tidal flow).<br />
This patented device comprises a small fin running down the length of<br />
the tubular. Along with the small drag increase there is a steady<br />
side force. This may be eliminated in the case of the total force on<br />
multi-tubular design by placing the fin alternately on opposite sides<br />
of the tubulars.<br />
D-26
D*9<br />
REFERENCES<br />
D.1<br />
King, R., “A Review of Vortex Shedding Research and Its<br />
Application,” Ocean Engineering, Vol. 4, pp. 141-171,<br />
1977.<br />
D.2<br />
Hallam, M. G., Heaf, N.J.,Wootton, L.R., Dynamics of Marine<br />
<strong>Structure</strong>s, CIRIA, 1977.<br />
D.3<br />
Vandiver, J.K., and Chung, T.Y., “Hydrodynamic Damping in<br />
Flexible Cylinders in Sheared Flow,” Proceedingsof Offshore<br />
Technology Conference, OTC Paper 5524, Houston, 1988.<br />
D.4<br />
Blevins, R.D., Flow Induced Vibrations, Van Nostrand<br />
Reinhold, 1977.<br />
,.—.><br />
D.5<br />
Blevins, R.D., Formulas for Natural Frequency and Mode<br />
Shape, Van Nostrand Reinhold, 1979.<br />
D.6<br />
German, D.I.,Free Vibration Analyses of Beams and Shafts,<br />
Wiley-Interscience, 1975.<br />
0.7<br />
Reid, D.L., “A Model for the Prediction of Inline Vortex<br />
Induced Vibrations of Cylindrical Elements in a Non-Uniform<br />
Steady Flow,” Journal of Ocean Engineering, August 1989.<br />
D.a<br />
Det Norske Veritas, “Rules for Submarine Pipeline Systems,”<br />
Oslo, Norway, 1981.<br />
0.9<br />
Griffin, O.M. and Ramberg, S.E., “Some Recent studies of<br />
Vortex Shedding with Application to Marine Tubulars, and<br />
Risers,” Proceedingsof the First Offshore Mechanics and<br />
Arctic Engineering/Deep Sea Systems Symposium, March 7 -<br />
10, 1982, pp. 33 - 63.<br />
D.lo<br />
$arpkaya, T., Hydroelastic Response of Cylinders in<br />
D-27
Harmonic Flow, Royal Institute of Naval Architects, 1979.<br />
D.11 Engineering Sciences Data Unit, Across Flow Response Due to<br />
Vortex Shedding, Publication No. 78006, London, England,<br />
Clctober, 1978.<br />
D-28
Fint Insmbility Rtgion I !%mmd InatabtiwRagim<br />
I<br />
I<br />
I<br />
Motion<br />
I<br />
I<br />
I<br />
I<br />
No Motion<br />
I<br />
I<br />
I<br />
I<br />
. . .- .-<br />
--i 0.5 1.0 1.2 1.5 2<br />
/<br />
I<br />
I<br />
o<br />
SmbiliPfParanww (KJ<br />
l.o
o.:<br />
Q.,<br />
0“..<br />
0.1<br />
0.1<br />
*<br />
REGION<br />
0<br />
9?* 1 ! 1 1 1$,,<br />
10’ 101 10]<br />
AND L4MINAR<br />
Of- TUR13ULENT VORTEX<br />
BOUNDARY IAYER<br />
CYLINDER<br />
TRAIL<br />
ON THE I /<br />
I ,’ :<br />
-“” “’ .:.. ..=. ...&. ~x---<br />
1d<br />
REYNOLDS<br />
NUMDER, l_tC<br />
10s<br />
I<br />
REGION WHERETIIE<br />
VORTEX SHEDDING<br />
FREQUENCY UN K<br />
DEFINED AS THE<br />
DOMINANT FREQUENCY<br />
IN A SPECTRUM<br />
u<br />
10E 10’<br />
Figure D-3 The Strou.halversus<br />
ReynoldIs Numbers<br />
for Cylinders<br />
..-.><br />
1.8<br />
1.6. \<br />
,’ .<br />
I # #’<br />
,’ *’9<br />
.,<br />
.-#-+-<br />
144-<br />
1,2<br />
1.0<br />
0.8<br />
0.6<br />
0.4<br />
0.2<br />
0 2.0 4.0<br />
StabilityPatwnemr<br />
(K$)<br />
Figure D-4 Amplitude of Response for cms$-FIOW Vibrations
.—____<br />
*U.S. E.P.O. :1993-343-273:80117<br />
~’q
(THIS PAGE INTENTIONALLY LEFT BIANK)
COMMllTEE ON MARINE STRUCTURES<br />
Commission on Engineering and Technical Systems<br />
National Academy of Sciences - National Research Council<br />
The COMMITTEE ON MARINE STRUCTURES has technical cognizance over the interagency<br />
<strong>Structure</strong> <strong>Committee</strong>’s research program.<br />
Peter M. Palermo Chairman, Alexandria, VA<br />
Mark Y. Berman, Amoco Production Company, Tulsa, OK<br />
Subrata K. Chakrabarti, Chicago Bridge and Iron, Plainfield, IL<br />
Rolf D. Glasfeld, General Dynamics Corporation, Groton, CT<br />
William H. Hartt, Florida Atlantic University, Boca Raton, FL<br />
Alexander B. Stavovy, National Research Council, Washington, DC<br />
Stephen E. Sharpe, <strong>Ship</strong> <strong>Structure</strong> <strong>Committee</strong>, Washington, DC<br />
LOADS WORK<br />
GROUP<br />
Subrata K. Chakrabarti Chairman, Chicago Bridge and Iron Company, Plainfield, IL<br />
Howard M. Bunch, University of Michigan, Ann Arbor, Ml<br />
Peter A. Gale, John J. McMullen Associates, Arlington, VA<br />
Hsien Yun Jan, Martech Incorporated, Neshanic Station, NJ<br />
Naresh Maniar, M. Rosenblatt & Son, Incorporated, New York, NY<br />
Solomon C. S. Yim, Oregon State University, Corvallis, OR<br />
MATERIALS WORK GROUP<br />
William H. Hartt Chairman, Florida Atlantic University, Boca Raton, FL<br />
Santiago Ibarra, Jr., Amoco Corporation, Naperville, IL<br />
John Landes, University of Tennessee, Knoxville, TN<br />
Barbara A. Shaw, Pennsylvania State University, University Park, PA<br />
James M. Sawhill, Jr., Newport News <strong>Ship</strong>building, Newport News, VA<br />
Bruce R. Somers, Lehigh University, Bethlehem, PA<br />
Jerry G. Williams, Conoco, Inc,, Ponca City, OK<br />
C-3<br />
*~117 LS7<br />
/
SHIP STRUCTURE COMMITTEE PUBLICATIONS<br />
SSC-351 An Introduction to Structural Reliability Theorv by Alaa E. Mansour<br />
1990<br />
SSC-352 Marine Structural Steel Touqhness Data Bank by J. G. Kaufman and<br />
M. Prager 1990<br />
SSC-353 Analysis of Wave Characteristics in Extreme Seas by William H. Buckley<br />
1989<br />
SSC-354<br />
SSC-355<br />
SSC-356<br />
SSC-357<br />
SSC-358<br />
SSC-359<br />
SSC-360<br />
SSC-361<br />
SSC-362<br />
SSC-363<br />
SSC-364<br />
SSC-365<br />
SSC-366<br />
None<br />
Structural Redundancy for Discrete and Continuous Svstems by P. K.<br />
Das and J. F. Garside 1990<br />
Relation of Inspection Findinqs to Fatique Reliability by M. Shirmzuka<br />
1989<br />
Fatique Performance Under Multiaxial Loai by Karl A. Stambaugh,<br />
Paul R. Van Mater, Jr., and William H. Munse 1990<br />
Carbon Equivalence and Weldability of Microalloyed Steels by C. D.<br />
Lundin, T. P.S. Gill, C. Y. P. Qiao, Y. Wang, and K. K. Kang 1990<br />
Structural Behavior After Fatique by Brian N. Leis 1987<br />
Hydrodynamic Hull Dampina (Phase 1~by V. Ankudinov 1987<br />
Use of Fiber Reinforced Plastic in Marine <strong>Structure</strong>s by Eric Greene<br />
1990<br />
Hull Strappinq of Shim by Nedret S. Basar and Roderick B. Hulls 1990<br />
<strong>Ship</strong>board Wave Heiqht Sensor by R. Atwater 1990<br />
Uncertainties in Stress Analysis on Marine <strong>Structure</strong>s by E. Nikolaidis<br />
and P. Kaplan 1991<br />
Inelastic Deformation of Plate Panels by Eric Jennings, Kim Grubbs,<br />
Charles Zanis, and Louis Raymond 1991<br />
Marine Structural Inteqrity Proqrams (MSIP) by Robert G. Bea 1992<br />
Threshold Corrosion Fatique of Welded <strong>Ship</strong>building Steels by G. H.<br />
Reynolds and J. A. Todd 1992<br />
<strong>Ship</strong> <strong>Structure</strong> <strong>Committee</strong> Publications - A Special Bibliography
4!! .<br />
,—.-.,