FATIGUE ASSESSMENT OF SHIP STRUCTURES

Transcription

1 CLSSIFICTION NOTES No FTIGUE SSESSMENT OF SHIP STRUCTURES FEBRURY 003 Veritasveien 1, N-13 Høvik, Noray Tel.: Fax:

2 FOREWORD is an autonomous and independent Foundation ith the objective of safeuardin life, property and the environment at sea and ashore. S is a fully oned subsidiary Society of the Foundation. It undertakes classification and certification of ships, mobile offshore units, fixed offshore structures, facilities and systems for shippin and other industries. The Society also carries out research and development associated ith these functions. operates a orldide netork of survey stations and is authorised by more than 10 national administrations to carry out surveys and, in most cases, issue certificates on their behalf. Classification Notes Classification Notes are publications hich ive practical information on classification of ships and other objects. Examples of desin solutions, calculation methods, specifications of test procedures, as ell as acceptable repair methods for some components are iven as interpretations of the more eneral rule requirements. n updated list of Classification Notes is available on request. The list is also iven in the latest edition of the Introductionbooklets to the "Rules for Classification of Ships and the "Rules for Classification of Hih Speed, Liht Craft and Naval Surface Craft". In "Rules for Classification of Fixed Offshore Installations", only those Classification Notes hich are relevant for this type of structure have been listed. This edition of Classification Note No replaces the issue from January 001. Main chanes Fiure.5 amended Chapter 8..1 Table nomenclature amended Chapter Table nomenclature amended Chapter nomenclature of the load condition updated Det Norske Veritas 003 Data processed and typeset by Det Norske Veritas Printed in Noray 5/0/003 3:16 PM - CN30.7.doc If any person suffers loss or damae hich is proved to have been caused by any nelient act or omission of Det Norske Veritas, then Det Norske Veritas shall pay compensation to such person for his proved direct loss or damae. Hoever, the compensation shall not exceed an amount equal to ten times the fee chared for the service in question, provided that the maximum compensation shall never exceed USD million. In this provision Det Norske Veritas shall mean the Foundation Det Norske Veritas as ell as all its subsidiaries, directors, officers, employees, aents and any other actin on behalf of Det Norske Veritas.

3 CONTENTS 1. General Introduction Scope, limitations, and validity Methods for fatiue analysis Fatiue accumulation Definitions Fatiue nalysis Cumulative damae Stresses to be considered S-N curves Corrosion Maximum alloable notch stress ranes Uncertainties in fatiue life prediction Simplified Stress nalysis General Lon term distribution of stresses Definition of stress components Combination of stresses Calculation of stress components Simplified Calculation of Loads General Wave induced hull irder bendin moments External pressure loads Internal pressure loads due to ship motions Ship accelerations and motions Wave Loadin by Direct Computation General The lon-term distribution Transfer functions Combination of transfer functions Desin ave approach Stress component based stochastic analysis Full stochastic analyses Finite Element nalysis Finite element models Global hull analysis Caro hold analysis Frame and irder models Local structure models Stress concentration models Fatiue nalysis of Oil Tankers Where to analyse Fatiue analysis Example of application (Simplified methods example) Fatiue analysis of bulk carriers Where to analyse Fatiue analysis Example of pplication, Bulk Carrier (Simplified methods example) Fatiue nalysis of Containerships and Other Ship Types Where to analyse Fatiue analysis Simplified Calculation of the Combined Lonitudinal Stress in Ships ith Lare Hatch Openins ppendix Stress Concentration Factors General Examples of K-factors for typical details in ships K-factors for holes ith ede reinforcement Workmanship ppendix B ssessment of Secondary Bendin Stresses Objective ssessment of secondary bendin stresses in double hulls ssessment of secondary panel stresses in sinle skin vessels ppendix C Simplified Loads for Direct Strenth nalysis ppendix D To-Slope S-N Curve Fatiue Damae Expression Weibull distributed stress rane Rayleih distributed stress rane ppendix E Backround for the S-N curves Introduction S-N curves for elded connections S-N curves for the base material References References...19

4 4 Classification Notes No General 1.1 Introduction Fatiue cracks and fatiue damaes have been knon to ship desiners for several decades. Initially the obvious remedy as to improve detail desin. With the introduction of hiher tensile steels (HTS-steels) in hull structures, at first in deck and bottom to increase hull irder strenth, and later on in local structures, the fatiue problem became more imminent In the DNV Rules for Classification of Ships, the material factor f 1, hich ive the ratio of increase in alloable stresses as a function of the material yield point as initially introduced in The factor is varyin ith the yield point at a loer than linear rate, this to ive some (but insufficient) contribution to the eneral safety aainst fatiue fracture of hiher tensile steels. Hoever, durin recent years a roin number of fatiue crack incidents in local tank structures in HTS steels have demonstrated that a more direct control of fatiue is needed This Classification Note is intended to ive a eneral backround for the rule requirements for fatiue control of ship structures, and to provide detailed recommendations for such a control. The aim of the fatiue control is to ensure that all parts of the hull structure subjected to fatiue (dynamic) loadin have an adequate fatiue life. Calculated fatiue lives, calibrated ith the relevant fatiue damae data, may ive the basis for the structural desin (steel selection, scantlins and local details). Furthermore, they can form the basis for efficient inspection prorams durin fabrication and throuhout the life of the structure To ensure that the structure ill fulfil its intended function, fatiue assessment, supported appropriate by a detailed fatiue analysis, should be carried out for each individual type of structural detail hich is subjected to extensive dynamic loadin. It should be noted that every elded joint and attachment or other form of stress concentration is potentially a source of fatiue crackin and should be individually considered. 1. Scope, limitations, and validity This Classification Note includes procedures for evaluation of fatiue strenth, but not limited to, for the folloin: Steel ship structures excludin hih speed liht crafts. Foundations elded to hull structures. ny other areas desinated primary structures on the drains of ship structures ttachment by eldin to primary ship structures, such as double plates, etc. This Classification Note may be adapted for modification to existin ship structures, subject to the limitations imposed by the oriinal material and fabrication techniques. This Classification Note is valid for steel material ith yield stress less than 500 MPa. 1.3 Methods for fatiue analysis Fatiue desin may be carried out by methods based on fatiue tests (S-N data) and estimation of cumulative damae (Palmrens - Miner rule) The lon term stress rane distribution is a fundamental requirement for fatiue analysis. This may be determined in various ays. This Classification Note outlines to methods for stress rane calculation: 1) postulated form of the lon-term stress rane distribution ith a stress rane based on dynamic loadin as specified in the rules. ) Spectral method for the estimation of lon-term stress rane In the first method a Weibull distribution is assumed for the lon term stress ranes, leadin to a simple formula for calculation of fatiue damae. The load effects can be derived directly from the ship rules. The nominal stresses have to be multiplied by relevant stress concentration factors for calculation of local notch stresses before enterin the S-N curve. The second method implies that the lon-term stress rane distribution is calculated from a iven (or assumed) ave climate. This can be combined ith different levels of refinement of structural analysis. Thus a fatiue analysis can be performed based on simplified analytical expressions for fatiue lives or on a more refined analysis the loadin and the load effects are calculated by numerical analysis. The fatiue analysis may also be performed based on a combination of simplified and refined techniques as indicated by the diaonal arros in Fiure The requirement to analysis refinement should be areed upon based on experience ith similar methods on existin ships and structural details ith respect to fatiue

5 Classification Notes No consequences of a fatiue damae in terms of service problems and possible repairs In eneral, the simplified method for fatiue life calculation is assumed to ive a ood indication as to hether fatiue is a sinificant criterion for desin or not. The reliability of the calculated fatiue lives is, hoever, assumed to be improved by refinement in the desin analysis It should further be kept in mind that real fatiue lives are a function of orkmanship related to fabrication and corrosion protection. Therefore, to achieve the necessary link beteen the calculated and the actual fatiue lives for ships, the fabrication has to be performed accordin to ood shipbuildin practice ith acceptance criteria as assumed in the calculation. 1.4 Fatiue accumulation The fatiue life under varyin loadin is calculated based on the S-N fatiue approach under the assumption of linear cumulative damae (Palmrens-Miner rule). The total damae that the structure is experiencin may be expressed as the accumulated damae from each load cycle at different stress levels, independent of the sequence in hich the stress cycles occur. The desin life assumed in the fatiue assessment of ships is normally not to be taken less than 0 years. The accumulated fatiue damae is not to exceed a usae factor of 1.0. The acceptance criteria is related to desin S-N curves based on mean- minus-to-standard-deviations curves for relevant experimental data. 1.5 Definitions The folloin eneral symbols are used in this Classification Note: B C B C D D F σ ( σ) H(ω) H s I I a Cross sectional area Greatest moulded breadth of ship in m measured at the summer aterline Block coefficient = /1.05LBT RULE Wave coefficient as iven in DNV Rules for Ships Pt.3, Ch.1. Moulded depth of ship, cfr, Rules Pt.3 Ch.1 Sec.1 Fatiue damae Weibull distribution Transfer function Sinificant ave heiht Moment of inertia Moment of inertia for the transverse frame I b Moment of inertia for the lonitudinal striner/ irder K Stress concentration factor K Geometric stress concentration factor K n Un-symmetrical stiffeners ith lateral loadin stress concentration factor K te Eccentric tolerance stress concentration factor (normally plate connections) K t nular mismatch stress concentration factor (normally plate connections) K Weld eometry stress concentration factor L Rule lenth of ship in m, cfr. Rules Pt.3 Ch.1 Sec.1. L pp Lenth beteen perpendiculars M Moment M o Wave induced vertical moment M H Wave induced horizontal moment N S Number of cross ties Q ( σ) Probability level for exceedance of stress rane σ S η ( ω) Wave spectrum S σ ( ω) Stress response spectrum T d Desin life T act Drauht actual T vessel mean moulded summer drauht in m T z Zero crossin period Z Section modulus a S-N fatiue parameter a Local / lobal load combination factor b Local / lobal load combination factor b f Flane idth a i cceleration in direction i f 1 Material factor as specified in the Rules f e Environmental reduction factor f m Mean stress reduction factor f r Factor for calculation of load effects at 10-4 probability level cceleration of ravity (= 9.81 m/s ) h Weibull shape parameter h o Basic Weibull shape parameter h Web heiht i a I a / S i b I b / l s l Stiffener lenth lo( ) 10th loarithm ln( ) Natural loarithm m S-N fatiue parameter m n Spectral moment of order n p Lateral pressure p ij Occurrence probability of sea condition i and headin j p s Sailin rate = fraction of desin life at sea

6 6 Classification Notes No q Weibull scale parameter s Stiffener spacin t Plate thickness t p Plate thickness t f Flane thickness t Web thickness t n Net plate thickness δ Deformation v ij Zero crossin frequency in short-term condition i, j ω Wave frequency v o Lon-term averae zero frequency ρ Correlation coefficient σ Stress amplitude σ Secondary stress amplitude resultin from bendin of irder system σ 3 Tertiary stress amplitude produced by bendin of plate elements beteen lonitudinal and transverse frames/stiffeners σ nominal Nominal stress amplitude, e.. stress derived from beam element or finite element analysis η Fatiue usae factor moulded displacement in t in salt ater (density 1.05 t/m 3 ) on drauht T σ Stress rane σ Global stress rane σ l Local stress rane σ h Nominal stress rane due to horizontal bendin σ v Nominal stress rane due to vertical bendin Γ( ) Gamma function

7 Classification Notes No Simplified nalysis Direct nalysis Load Response Sec Load Response, Sec. 5. Load Transfer Function. Sec Stress Components Interchaneable Results FE Model of Ship, Ch. 6 SCF: K-factors, Ch. 10 Interchaneable Results FE Model of detail, Sec Combination of Stresses, Sec Desin Wave pproach, Sec. 5.5 Local Stress Transfer Functions for stress components Sec. 5.4 Lon Term Stress Distribution, Sec.3. Stress Component based Stochastic Fatiue nalysis. Sec. 5.6 Full Stochastic Fatiue nalysis, Sec. 5.7 Equivalent Lon Term Stress Distribution (Weibull param.), Sec. 5. Fatiue Damae Calculation, Sec..1 Fatiue Damae Summation: Summation of damae contributions from each ave period/ship headin combination for each sea state in the ave scatter diaram Fiure 1.1 Flo diaram over possible fatiue analysis procedures

8 8 Classification Notes No Fatiue nalysis.1 Cumulative damae.1.1 The fatiue life may be calculated based on the S-N fatiue approach under the assumption of linear cumulative damae (Palmrens-Miner rule). When the lon-term stress rane distribution is expressed by a stress historam, consistin of a convenient number of constant amplitude stress rane blocks σ i each ith a number of stress repetitions n i the fatiue criterion reads D a, m k n i N i η k k ni 1 D = = n i ( σ i) η N a i= 1 i i= 1 = accumulated fatiue damae = S-N fatiue parameters = number of stress blocks m = number of stress cycles in stress block i = number of cycles to failure at constant stress rane σ i = usae factor. ccepted usae factor is defined as η = 1.0 pplyin a historam to express the stress distribution, the number of stress blocks, k, is to be lare enouh to ensure reasonable numerical accuracy, and should not be less than 0. Due consideration should be iven to selection of interation method as the position of the interation points may have a sinificant influence on the calculated fatiue life dependent on interation method..1. When the lon-term stress rane distribution is defined applyin Weibull distributions for the different load conditions, and a one-slope S-N curves is used, the fatiue damae is iven by, N load p n N ν T load 0 d m m D = pq n n Γ( 1+ ) η a h n= 1 n = total number load conditions considered = fraction of desin life in load condition n, Σp n 1, but normally not less than 0.85 = secs. ) h n = Weibull stress rane shape distribution parameter for load condition n, see Section 3. q n = Weibull stress rane scale distribution parameter for load condition n v o = lon-term averae response zerocrossin frequency Γ( 1+ m = amma function. Values of the amma ) h function are listed in Table.1 n h m = 3.0 h m= Table.1 Numerical values for Г(1+m/h) The Weibull scale parameter is defined from the stress rane level, σ o, as q n σ 0 = 1 (ln n 0 ) / n o is the number of cycles over the time period for hich the stress rane level σ o is defined. ( σ o includes mean stress effect) h n In combination ith calculation of stress rane σ o by the simplified method iven in Chapters 3, 4 and 10, the zerocrossin-frequency may be taken as, 1 = lo ( L) ν T d = desin life of ship in seconds ( 0 years L is the ship Rule lenth in meters.

9 Classification Notes No lternatively, in combination ith calculation of stress rane σ o by direct analyses, the averae zero-crossin-frequency can be derived as iven in Chapter When the lon term stress rane distribution is defined throuh a short term Rayleih distribution ithin each short term period for the different loadin conditions, and a oneslope S-N curve is used, the fatiue criterion reads, N ν T m load 0 d m D = Γ( 1+ ) pn rijn ( m0ijn ) η a r ij v o m oij n= 1 all seastates all headins i= 1, j= 1 = the relative number of stress cycles in short-term condition i, j, see also 5..6 = lon-term averae response zero-crossinfrequency, see 5..6 = zero spectral moment of stress response process The Gamma function, Γ( 1+ m ) is equal to 1.33 for m = 3.0. Expressions for fatiue damae applyin bi-linear S-N curves are iven in ppendix D. Stresses to be considered..1 The procedure for the fatiue analysis is based on the assumption that it is only necessary to consider the ranes of cyclic principal stresses in determinin the fatiue endurance. Hoever, some reduction in the fatiue damae accumulation can be credited hen parts of the stress cycle rane are in compression. It should be noted that in elded joints, there may be several locations at hich fatiue cracks can develop, e.. eld toe and eld root of fillet joints. It is recommended to check potential hot spot locations for fatiue cracks by comparin the principal stress level ith the maximum alloable notch stress ranes in Tables.8 and.9... When the potential fatiue crack is located in the parent material at the eld toe, the relevant local notch stress is the rane of maximum principal stress adjacent to the potential crack location ith stress concentrations bein taken into account. This stress concentration is due to the ross shape of the structure and the local eometry of the eld. s an example, for the eld shon in Fiure.1 a), the relevant local notch stress for fatiue desin ould be the tensile stress, σ, multiplied by the stress concentration factor due to the eld K. For the eld shon in Fiure.1 b), the stress concentration factor for the local eometry must in addition be accounted for, ivin the relevant local notch stress equal to K K σ, K is the stress concentration factor due to the hole. For butt-elds ith the eld surface dressed flush and small local bendin stress across the plate thickness, K of 1.0 is to be used. Otherise K of 1.5 is to be used. The maximum principal stress rane ithin 45 of the normal to the eld toe should be used for the analysis. 45 de 45 de Max principal stress in this sector to be used in analyses a) b) Fiure.1 Explanation of local notch stresses..3 For fatiue analysis of reions in the base material not sinificantly effected by residual stresses due to eldin, the stress rane may be reduced dependent on hether mean cyclin stress is tension or compression. This reduction may e.. be carried out for cut-outs in the base material. Mean stress means that the static notch stress includin relevant stress concentration factors. The calculated stress rane obtained may be multiplied by the reduction factor f m as obtained from Fiure. before enterin the S-N curve. For variable amplitude stresses, σ can be taken as the stress rane at the 10-4 probability level of exceedance. Fiure. Stress rane reduction factor to be used ith the S-N curve for base material

10 10 Classification Notes No Residual stresses due to eldin and construction are reduced over time as the ship is subjected to loadin. If a hot spot reion is subjected to a tension force implyin local yieldin at the considered reion, the effective stress rane for fatiue analysis can be reduced due to the mean stress effect also for reions effected by residual stresses from eldin. Mean stress means that the static notch stress includin relevant stress concentration factors. The folloin reduction factor on the derived stress rane may be applied, see Fiure.3. Table.. The backround for the S-N curves are described in ppendix E..3.4 The use of bi-linear S-N curves complicates the expression for fatiue damae, ref. ppendix D. In order to reduce the computational effort, simplified one-slope S-N curves have been derived for typical lon term stress rane distributions in ship structures. Use of the one-slope S-N curves, defined by parameters as iven in Table.3, leads to results on the safe side for calculated fatiue lives exceedin 0 years. f m = reduction factor due to mean stress effects = 1.0 for tension over the hole stress cycle = 0.85 for mean stress equal to zero = 0.7 for compression over the hole stress cycle.3.5 The basic desin S-N curve is iven as, lon = loa mlo σ ith S-N curve parameters iven in Tables. and.3. For parts of the structure bein exposed to both compressive and tensile mean stress dependin on the loadin situation, the reduction factor f m = 0.85 may be applied on the lonterm stress rane distribution. For variable amplitude stresses, σ can be taken as the stress rane at the 10-4 probability level of exceedance. N σ m loa = predicted number of cycles to failure for stress rane σ = stress rane = neative inverse slope of S-N curve = intercept of lon-axis by S-N curve lo a = lo a - s Fiure.3 Stress rane reduction factor that may be used ith S-N curve for elded structures.3 S-N curves.3.1 The fatiue desin is based on use of S-N curves hich are obtained from fatiue tests. The desin S-N curves hich follo are based on the mean-minus-to-standard-deviation curves for relevant experimental data. The S-N curves are thus associated ith a 97.6% probability of survival..3. The S-N curves are applicable for normal and hih strenth steels used in construction of hull structures..3.3 The S-N curves are presented as straiht lines in a lo-lo scale. S-N curves in air are often presented as bi-linear ith a chane in slope beyond 10 7 cycles. S-N curves for elded joints and base material in air/cathodic protected environment and for corrosive environment are iven in a = is constant relatin to mean S-N curve s = standard deviation of lo N; s = 0.0 In combination ith the fatiue damae criteria iven in Section.1. and.1.3, curves Ib, II, IIIb and IV should be used. S-N Curve Material N 10 7 N > 10 7 lo a m lo a m I Welded joint III Base Material a) ir or ith cathodic protection: S-N Curve Material lo a m II Welded joint IV Base material b) Corrosive Environment: Table. S-N parameters

11 Classification Notes No S-N Curve Material lo a m Ib Welded joint IIIb Base material Table.3 lternative one-slope S-N parameters S-N CURVES (Table.) 1000 lll Stress rane (Mpa) 100 l lv ll 10 1,00E+04 1,00E+05 1,00E+06 1,00E+07 1,00E+08 1,00E+09 Stress cycles Fiure.4 Desin S-N Curves.3.6 The desin S-N curves iven here correspond to test results from smooth specimens havin a stress concentration factor K = 1.0. The curves for base material assume that flame cutor otherise rouh surfaces are round. For base material the surface and ede corners are machined or round smooth and protected aainst corrosion, the fatiue life may be increased by a factor of.0. The curves for elded joints assume elds proven free from sinificant defects. For fatiue analysis of details ith K 1.0, the stress rane must incorporate the stress concentration factors, see Chapter The fatiue strenth of elded joints is to some extent dependent on plate thickness and on the stress radient over the thickness. Thus for thickness larer than 5 mm, the S-N curve in air reads lon = loα m 4 lo t m lo σ 5 t is thickness (mm) throuh hich the potential fatiue crack ill ro. This S-N curve in eneral applies to all types of elds. For fatiue analysis of details the stress concentration factor is less than 1.3, the thickness effect can be nelected and the basic S-N curve can be used. Such stress concentration factors are normally only achieved throuh rindin or machinin of the eld/base material transition. See 10.4 for a description of normal orkmanship associated ith use of these S-N curves..4 Corrosion.4.1 It is reconised that the fatiue life of steel structures is considerably shorter in freely corrodin condition submered in sea ater than in air, i.e. in dry indoor atmosphere such as common laboratory air. For steel submered in sea ater and fully cathodically protected, approximately the same fatiue life as in dry air is obtained. n intact coatin system ill also protect the steel surface from the corrosive environment, so that the steel can be considered to be as in dry air condition.

12 1 Classification Notes No The basic S-N curve for elded reions in air is only to be applied for joints situated in dry spaces or joints effectively protected aainst corrosion. For joints efficiently protected only a part of the desin life and exposed to corrosive environment the remainin part, the fatiue damae may be calculated as a sum of partial damaes accordin to.4.3. Estimatin the efficient life time of coatin- and cathodic protection systems, due consideration is to be iven to specification, application and maintenance of the systems. uideline for effective life times for common corrosion protection systems is listed in Table.4. Full cathodic protection is defined as: Steel surfaces submered in normal, aerated sea ater havin a potential of volts measured ith a silver/silver chloride reference cell ( or - 0,79 volts versus a calomel cell, etc.). The potential limit for cathodic overprotection ill vary ith the deree of steel strenth. For normal strenth carbon mananese steel (yield strenth min. 35 MPa) it ill be - 1,1 volts versus silver/silver chloride. Cathodic overprotection, especially in hydroen sulphide H S containin environment, can lead to hydroen embrittlement of hih strenth steels. Coatin system Epoxy based (liht coloured ) Epoxy coal tar (Coal Tar Epoxy) Other reconised coatin systems 1) 5 years Prep. I ) Coats x mic. 1 x 00 1 x years Prep II ) Coats x mic. s above x years Prep III ) Coats x mic. s above x 175 (to x 5 ) 1. Other-than-epoxy based ith ell documented performance. Prep. I: Steel plates shop primed on blast cleaned surface to Sa - Sa,5. Welds and burns mechanically cleaned to min. St 3. To obtain a coatin durability 5 years the steel surface preparation for shop primin should be Sa,5 Prep. II: Zinc rich shop primer on surface blast cleaned to Sa,5 or better. Sharp edes broken. Welds and burns cleaned to min. St 3. Dry conditions: ir humidity 85 % and steel temperature 3 C above the de point durin blast cleanin and coatin application. Prep. III: Zinc rich shop primer on surface blast cleaned to Sa,5 or better. Sharp edes broken. Welds, burns and broken edes blast cleaned to Sa,5 or better. Clean conditions: ny oil, rease, dust, eld smoke or salt contamination on shop primed or other surface to be coated, removed by cleanin before final blastin operations. Dry conditions: ir humidity 85 % and steel temperature 3 C above the de point durin blast cleanin and coatin application. Table.4 Duration of corrosion protection.4. Global stress components may be calculated based on ross scantlins. Local stress components should be calculated based on reduced scantlins, i.e. ross scantlins minus corrosion addition t k as iven in Table.5. (The corrosion addition specified belo is similar to that specified in the Rules [1] ) Tank/hold reion Internal members and plate boundary beteen spaces of the iven cateory Location Within 1.5 m belo eather deck tank or hold top Else Ballast tank 1) Caro oil tank only (0) ) Hold of dry bulk caro carriers 4) Plate boundary beteen iven space cateories Ballast tank 1) / Caro oil tank only Ballast tank 1) / Hold of dry bulk caro carrier 4) Ballast tank 1) / Other cateory space 3) Caro oil tank only / Other cateory space 3) (3) 5) Within 1.5 m Else belo eather deck tank or hold top (1.0) ) (0) ) Hold of dry bulk carrier 4 ) / Other cateory space 3) ) The term ballast tank includes also combined ballast and caro oil tanks, but not caro oil tanks hich may carry ater ballast accordin to Reulation 13 (3), of MRPOL 73/78, see Rules ) The fiure in bracket refers to non-horizontal surfaces. 3) Other cateory space denotes the hull exterior and all spaces other than ater ballast and caro oil tanks and holds of dry bulk caro carriers. 4) Hold of dry bulk caro carriers refers to the caro holds of vessels ith class notations Bulk Carrier and Ore Carrier 5) The fiure in bracket refers to loer part of main frames in bulk carrier holds. Table.5 Corrosion addition tk in mm.4.3 Fatiue strenth may normally be assessed ith the S-N curve in air for the effective corrosion protection period, i.e. the corrosion protection period of coatin plus 5 years. The S-N curve in corrosive environment is to be used for the remainin time. Carriae of sour crude oil caroes may reduce the fatiue strenth of structures in caro tanks, hile the seet oil caroes may not reduce the fatiue life. s type of oil caroes and actual corrosion protection period of the coatin system are uncertain at the desin stae, for ne buildin of oil tankers it is assumed that caro tanks are exposed to seet caroes for the loaded condition. Thus, for caro tanks, the S-N curve in air may normally be used for ne buildin of oil tankers. If sour caroes and the other caroes causin corrosion are to be carried out frequently in caro tanks, the S-N curve in corrosive environment may normally be used.

13 Classification Notes No For the ballast condition, as there is no caro in the caro tanks, the S-N curve in air may normally be used for ne buildin of vessels. For ballast tanks, the effective corrosion protection period of 15 years may normally be used for ne buildin of vessels. For dry caro holds, fuel oil tanks, void spaces, cofferdam, and hull external surfaces, the S-N curve in air may normally be used..5 Maximum alloable notch stress ranes.5.1 Dependin on the required accuracy of the fatiue evaluation it may be recommended to divide the desin life into a number of time intervals due to different loadin conditions and limitations of durability of the corrosion protection. For example, the desin life may be divided into one interval ith ood corrosion protection and one interval the corrosion protection is more questionable for hich different S-N data should be used, see Section.3. Each of these intervals should be divided into that of loaded and ballast conditions. s a simplification therefore, the fatiue damae may be calculated applyin S-N curves for air and then multiply the fatiue damae by the factor χ iven in Table.6. The procedure can be described as: 1. Calculate the fatiue life accordin to I or III (In ir). If the calculated fatiue life is reater than the effective corrosion protection period, i.e. T c +5 years, the corrected life is calculated as Tdesin DInir T = Tc Tc 5 D Inir DCorrosive 1 D Inir correspond to the calculated fatiue life (In ir), T c = coatin life time, T desin = desin life time in years. The factor χ as iven in the table belo is determined Dcorrosive assumin. 3 DInir The effect of varyin deree of corrosion protection of ballast tanks durin desin life may, as a simplification, be accounted for by multiplyin the calculated fatiue damae by a factor χ from Table.6. It is then assumed that the fatiue damae has been calculated for the total life time usin only one of the S-N curves I or III. Guidance on expected effective corrosion protection period of coatin is iven in.4 and in Guidelines No 8. For ballast tanks protected by anodes only, it is recommended to apply S-N curves II or IV. Ballast tanks Tanks for caro oil S-N curve used for calculation of fatiue damae Effective corrosion protection period (Yrs) Desin life (years) I or III Table.6 Factor χ environment on fatiue damae to account for corrosion protection of structure exposed to a corrosive Example: ssume that a fatiue damae of ballast tank structure D air has been calculated for a desin life equal to 5 years based on S-N curve I. ssume further that the considered detail is exposed to a corrosive environment but ith an effective corrosion protection period lastin 15 years. Then the resultin fatiue damae is obtained as D = 1.5 D air

14 14 Classification Notes No It can be assumed as a simplification that the lon-term stress rane distribution over the desin life is a Weibull distribution. ny stress rane in this distribution is a particular quantile, i.e. it is associated ith a particular probability of exceedance. The maximum alloable stress rane at a particular probability of exceedance is a desin stress rane, defined as the stress rane that results from usin the Weibull distribution in conjunction ith a characteristic S-N curve and requirin the accumulated damae in the desin life to be equal to 1.0. The characteristic S-N curve is obtained as the mean S-N curve shifted to standard deviations of lon to the left. Under the current safety format, the characteristic S-N curve is used as the desin S-N curve. Different maximum alloable notch stress ranes result for different Weibull shape parameters h and for different characteristic S-N curves. (For backround see Chapter 10 of Ref. [8].) In Tables.8 and.9, the maximum alloable notch stress rane ( σ o ) at probabilities of exceedance 10 4 and 10 8, respectively, has been calculated for total desin lives of , and cycles. n example of use of the tables is shon in Fiure.5. The S-N parameters for S-N curves I, II, III, and IV are iven in Table.. The maximum alloable notch stress rane includes the stress concentration factors (K-factors), such that the maximum alloable nominal stress rane to be used for desin is obtained as Example: Weibull shape parameter h = 0.87 Total number of stress cycles n total = Welded joint, corrosive environment, S-N curve II It follos from Tables.9 that the maximum alloable notch stress rane at 10-8 probability level of exceedance is MPa and from Table.8 that maximum notch stress rane at 10-4 probability level of exceedance is MPa The maximum alloable nominal stress rane takin into account that the considered detail is exposed to a corrosive environment, but protected aainst corrosion for a fraction of the desin life, can be approximated as σ nominal = σ K χ σ 0 is taken from Tables.8 and.9 assumin S- N curve I or III and χ is iven in.5.1. Example: ssume a elded detail ith resultin stress concentration factor K = 3.1, h = 0.9, desin life 0 years and number of cycles From Table.9 e et σ 0 = 516. MPa. ssume effective corrosion protection 10 years, then it follos from.5.1 that χ = 1.7. Maximum alloable nominal stress rane (extreme value) is then σ nominal = 516. / 3, /3 = 140 MPa σ nominal = σ 0 K Fiure.5 Stress rane versus probability of exceedance

15 Classification Notes No Weibull Shapeparameter h cycles S-N Curve I S-N Curve II Welded joint S-N Curve III Base material S-N Curve IV Base material Welded joint ir/cathodic, σ 0 Corrosive, σ 0 ir/cathodic, σ 0 Corrosive, σ 0 Desin life cycles Desin life cycles Desin life cycles Desin life cycles cycles cycles cycles cycles cycles Table.7 Maximum alloable notch stress rane (MPa) at a probability of exceedance 10-4 to keep the fatiue damae less than 1.0 for different desin life cycles. ( Weibull distribution for the lon-term stress rane is assumed.) cycles cycles cycles cycles cycles cycles

16 16 Classification Notes No Weibull Shapeparameter h S-N Curve I S-N Curve II Welded joint S-N Curve III Base material S-N Curve IV Base material Welded joint ir/cathodic, σ 0 Corrosive, σ 0 ir/cathodic, σ 0 Corrosive, σ 0 Desin life cycles Desin life cycles Desin life cycles Desin life cycles cycles cycles cycles cycles cycles cycles cycles cycles cycles cycles cycles cycles Table.8 Maximum alloable notch stress rane (MPa) at a probability of exceedance 10 8 to keep the fatiue damae less than 1.0 for different desin life cycles. ( Weibull distribution for the lon-term stress rane is assumed.)

17 Classification Notes No Uncertainties in fatiue life prediction.6.1 There are a number of different uncertainties associated ith fatiue life predictions. The calculated loadin on the ship is uncertain due to uncertainties in ave heihts, periods and distribution of aves. The resultin stresses in the ship are uncertain due to uncertainties in the loadin, calculation of response and calculation of stress concentrations..6. Because of the sensitivity of calculated fatiue life to the accuracy of estimates of stresses, particular care must be taken to ensure that stresses are realistic. Fatiue damae is proportional to stress raised to the poer of the inverse slope of the S-N curve. I.e. small chanes in stress result in much reater chanes in fatiue life. Special attention should be iven to stress raisers like eccentricities and secondary deformations and stresses due to local restraints. Due considerations should, therefore, be iven to the fabrication tolerances durin fatiue desin..6.3 There is a rather lare uncertainty associated ith the determination of S-N curves. The scatter in the test results hich form the basis for the S-N curves is enerally accepted to relate to the normal variation of eld imperfections ithin normal orkmanship, indicated in Table The ratio beteen calculated fatiue lives based on the mean S-N curve and the mean minus to standard deviations S-N curve is sinificant as shon in Fiure There is also uncertainty associated ith the determination of stress concentration factor. The error introduced in the calculated fatiue life by ron selection of stress concentration factor is indicated in Fiure.7. Fatiue life, L (years) h = 0.90 = m + s m = mean (S-N curve) Fiure.6 Fatiue life influence of stress level and S-N data for elded connections Fatiue life, L (years) h = 0.80 h = 0.90 h = K-factor h = 1.00 : L = h = 0.90 : L = h = 0.80 : L = 93.0 K K K 3.0 Fiure.7 Fatiue life sensitivity to stress concentration factor K and Weibull shape factor h.6.5 It should be kept in mind that a hih fatiue life is an efficient means to reduce probability of fatiue failure, see Fiure.8. It also reduces the need for in-service inspection. In order to arrive at a cost optimal desin, effort should be made to desin details such that stress concentrations, includin that of fabrication tolerances, are reduced to a minimum. Stresses may also be reduced by increasin the thickness of parent metal or eld metal, hich ill improve fatiue life Miner sum Fiure.8 Relative probability of failure versus Miner sum. P fo is probability of failure in lifetime for a Miner sum equal 1.0

18 18 Classification Notes No Simplified Stress nalysis 3.1 General This section outlines a simplified approach to determine the distributions of lon-term stress ranes for closed or semiclosed hull cross sections, expressed as Weibull distributions. Simple formats for combination of lobal and local stress components are iven, and alternative models for simplified calculations of stress response in ship structures are iven Stress response may be calculated by different levels of accuracy: Calculation of hull irder stresses is the simplest ay of ettin reasonable approximations to the stress level in lonitudinal hull irder elements and connections and can be used for quick evaluation of stress levels in important details. Dynamic pressure load analysis combined ith manual calculation of stiffener bendin response is the simplest ay for determinin the stress response of lonitudinal and transverse frames due to external and internal pressure loads. The member end restraints/moments must, hoever be evaluated ith reat care in order to arrive at reliable results. frame analysis is enerally more reliable and should be performed if it is uncertainty about used restraints/moments. Frame nalysis should be applied in order to assess stresses in hull elements like transverse frames, floors and irders. Both -D and 3-D models can be used (Section 6.4). In order to obtain a more precise stress estimation of the response in the hull, Finite Element nalysis should be applied (Section 6). 3. Lon term distribution of stresses 3..1 The lon term distribution of stress ranes at local details may be described by Weibull distribution : Q h q h σ Q( σ ) = exp q = probability of exceedance of the stress rane σ = Weibull shape parameter = Weibull scale parameter, defined as q = σ 0 ( lnn ) 1/ h 0 The stress rane distribution may also be expressed as σ o n o lnn σ = σ 0 lnn0 1/ h = reference stress rane value at the local detail exceeded once out of n o cycles = total number of cycles associated ith the stress rane level σ o When the lon term stress rane follos a Weibull distribution ith shape parameters in the rane , the main contribution to the cumulative fatiue damae comes from the smaller aves, see Fiure 3.1. The lon term stress rane should enerally be based on a reference stress rane σ o bein the hihest stress rane out of 10 4 stress cycles. Main contribution to accumulated fatiue damae Lo n Fiure 3.1 Contribution to fatiue damae from different stress blocks If the hihest stress rane out of 10 8 stress cycles are used to describe the lon-term stress rane distributions, the calculated fatiue damae is very sensitive to the estimate of the Weibull shape parameter h. 3.. The Weibull shape parameter may be established from lonterm ave load analysis. In lieu of more accurate calculations, the shape parameter may be taken as h = h o h = h o + h a ( D - z ) / ( D - T act ) h = h o + h a h = h o + h a z/t act (T act -z ) h = h o T act For deck lonitudinals For ship side above the aterline For ship side at the aterline For ship side belo the aterline For bottom lonitudinals T act <z <D z = T act z < T act