Marine_Structural_Design Episode 11 doc

378 Part 111 Fatigue and Fracture maximum value with nearly zero probability of occurrence. The calculated stress ranges are used to evaluate the integral in Eq. (20.20). For each sea-state, the fatigue damage associated with each current velocity is multiplied by the probability of occurrence of the current velocity. When stress ranges for all sea-states are obtained through the wave force model, the fatigue damage is calculated using Eq. (20.20). The advantage of using the time-domain fatigue for pipeline ad riser assessment is to account for the non-linearity in the drag forces and structural dynamic response. The other benefit is to reduce the conservatism introduced in the boundary condition for spectral fatigue analysis. An engineering practice is to derive the ratio of the predicted fatigue life from these two approaches for a few well-selected and performed analyses, and then to apply this ratio to similar fatigue scenarios. 20.3.3 Analysis Methodology for Time-Domain Fatigue of Risers In time-domain analysis, a time domain dynamic analysis is performed for all sea states in the wave scatter diagram, and for each direction with a non-zero probability of occurrence. In frequency-domain fatigue analysis of risers, the touch-down point is fixed. The time-domain analysis is applied when the soil-pipe interaction needs to be accounted for in order to remove the conservatism introduced in the frequency-domain analysis. Besides, the second order (drift) motions of the vessel may significantly affect the result of fatigue analysis. It is difficult to include the second-order motions using stress RAOs to transfer wave spectra into stress spectra. Based on the stress time histories from the time-domain dynamic analysis, the fatigue damage may be estimated as follows: The fatigue damage is estimated based on the moments of spectra (as those used in the frequency-domain analysis), and the stress-spectra are calculated using the Fast Fourier Transform algorithm. The fatigue damage is calculated directly from the stress time-history using a rainflow counting techniques. The dynamic simulation should be long enough because the dominant period of second order motions is of the order of 100 seconds. 20.3.4 Analysis Methodology for Time-Domain Fatigue of Nonlinear Ship Response Jha and Winterstein (1998) proposed a "Nonlinear Transfer Function (NTF)" method for efficient prediction of the stochastic accumulation of fatigue damage due to nonlinear ship loads in random seas. Nonlinear time-domain ship-load analysis may reveal asymmetry in sag and hog moment at mid-ship. The goal of the NTF method is derive accurate prediction using only a limited amount of nonlinear analysis based on regular waves. The analysis cost is reduced because expensive time-domain analysis over many cycles of ir-regular sea is replaced by a limited number of regular-wave analysis. The NTF is the generally nonlinear transformation from wave amplitude and period to the load amplitude measure of interest (e.g., total load range for rainflow-counting). Stochastic process theory is applied to Identify a minimal set of regular waves @e., wave heights and associated periods) to be applied based on a discretized version of the Foristall (1978) wave height distribution and Longuet-Higgins (1983) model for wave period selection. Assign an appropriate set of "side-waves" to be spatially distributed along the ship based on probability theory. Chapter 20 Spectral Fatigue Analysis and Design 379 Determine how these results should be weighted in predicting statistics of the loads produced in random seas. The prediction of the time-domain fatigue analysis was compared with frequency-domain stochastic fatigue analysis that assumes linear model of ship behavior. It was revealed that the nonlinear effect is significant. The NTF method may also be applied to any offshore structures. 20.4 Structural Analysis 20.4.1 Overall Structural Analysis Overall structural analyses are usually performed using space frame models and fine FEA models. The space frame analyses define the boundary loads for local structural models. To get the stress transfer functions for the fatigue damage assessment, these boundary loads are used to factor the results of fine, FEA unit load analysis results. This section presents aspects of modeling, load evaluation, and structural analysis applicable to the overall structural analysis. Space Frame Model The space frame model includes all the important characteristics of the stiffness, mass, damping, and loading properties of the structure and the foundation for the structural system. It consists primarily of beam elements. The accuracy of the calculated member end forces is influenced by the modeling techniques used. Figure 20.1 shows a space frame model for TLP hull primary structures and deck primary structures. Although not shown in this figure, tendons are included in the model as supporting structure to provide the proper vertical stiffness. Tubular beam elements are used to model the tendons. Applied load cases are, in general, self-balancing and should result in zero net load at the tops of the tendons. Thus, relatively flexible lateral springs are provided at the tops of all tendons in order to stabilize the analysis model against small net lateral loads. The hull's column and pontoon structures are modeled using beam-column elements. Joint and member definitions are interfaced from the global analysis model because interfaced loads from this analysis must be consistent with the model. Member properties are determined based on the member cross-sectional properties and material properties. Yield stresses of plate and stiffener components are input, along with the maximum bracket spacing for ring stiffener frames. Additional joints and members are included to ensure that the tendons and deck structure are structurally stable and as additional load collectors where appropriate. Deck members are modeled using the tubular or AISC (American Institute of Steel Constructions) elements. Deck equipment mass locations are determined for each major deck area and specifically included in the model so that proper inertial load magnitudes and centers of action are generated in the analysis. 380 Part III Fatigue and Fracture \ Figure 20.1 Space Frame Model for a TLP Fine FEA Model A fine FEA model may be used to analyse the hull structure or a part of the hull structure in detail. All relevant structural components shall be included in the model. In the fine FEA model, major primary structural components are fully modeled using three- and four-node platelshell elements and solid elements. Some secondary structural components may be modeled as two-node beam elements. Design Loading Conditions To adequately cover the fatigue environment, fatigue design loading conditions consist of cyclic environmental load components at a sufficient number of wave frequencies. These loading conditions include: Other cyclic loading The loading components are either explicitly generated or interfaced from the global motion analysis. Load summaries are made for each design loading condition and checked for accuracy and load imbalance. Chapter 20 Spectral Fatigue Analysis and Design 381 The global motion analysis serves as a basis for dynamic load development. The actual interface from the global analysis to the structural analysis consists of several loading components for each analyzed wave period and direction: the real and imaginary applied unit amplitude, wave diffraction and radiation loads, the associated inertial loads and other cyclic loading such as tendon dynamic reactions. The successful interface of these load components is dependent on a consistent geometric and mass model between the motion and the structural analyses and is also dependent on a consistent generation of the loading components in the motion analysis. Consistent modeling is obtained by interfacing the model geometry directly from the motion analysis wherever possible. Consistent mass is obtained by interfacing with the same weight control database for both the motions and structural analyses, when available. Load combinations are formed for each wave period and direction. These combinations consist of the applied wave load, the generated inertial load, and the associated cyclic loadings such as tendon dynamic reactions for both real and imaginary loadings of the floating structures. These combinations form the total cyclic load condition for each wave period and direction to be used in the spectral fatigue analysis. Analysis and Validation Hull structural analyses are performed using linear finite element methods. The reaction forces include total force and moment reactions and the analysis results are verified. Symmetrical or asymmetrical load conditions are checked to confirm symmetrical or asymmetrical analysis results. 20.4.2 Local Structural Analysis Local structural details are included as a part of the analyses for the entire hull structure. The analysis of the structural details may be performed using the finite element program such as ABAQUS (HKS, 2002) and other software. The FEM model is three-dimensional and linear stress analysis is performed. The results from the FEA model are interfaced into the fatigue model for additional model validation and subsequent spectral fatigue analysis of the local structural details. The entire model is plotted and revised for accuracy both from the FEA model and after interface to the fatigue model. Loading conditions for finite element analysis of local structural details should be based on the hull's structural analysis since it includes all cyclic loadings of the structure. The unit loading conditions are frequently applied. The resulting stresses for each unit load condition are interfaced to the fatigue model for subsequent combination into fatigue design loads. 20.5 Fatigue Analysis and Design 20.5.1 Overall Design A spectral fatigue assessment should be carried out for each individual structural detail. It should be noted that every structural detail, every welded joint and attachment or any other form of stress concentration is potentially a source of fatigue cracking and should be considered individually. The UK DEn procedure or its modified versions are recommended in Europe for the fatigue analysis and design of floating structures since it is the most widely accepted code. Design 382 Part III Fatigue and Fracture standards such as AWS (1997) are used in the USA. However, it should be noted that different design standards provide different procedures in the fatigue stress determination and S-N classification, which result in large discrepancies in the predicted fatigue damages. Therefore, a consistent procedure based on one design standard shall be used. The safety factors for fatigue design of floating structures are given by the design standards listed in Section 20.2 based on: Criticality of the joint Inspectability and repairability The criticality of a join is determined based on its structural redundancy. A joint is critical if its failure will potentially lead to the failure of the structure. 20.5.2 Stress Range Analysis A stress range analysis is performed using the fatigue software as a precursor to the fatigue damage calculation. The FEA unit load, model geometry and element stress results are interfaced into the fatigue calculation model. Loading combinations will then be defined for each fatigue wave load based on the applied boundary loads. Geometry and element properties from the space frame model are plotted and revised for accuracy. Any detected errors are corrected in the FEA input file and the FE analysis repeated. The finite element model of the specific hotspot region shall be developed based on the procedures, finite element size requirement defined by the design standards. In the FEA model, unit load results will be interfaced into the space fiame model database. These unit loads are then appropriately combined based on the applied boundary loads. 20.5.3 Spectral Fatigue Parameters Wave Environment The wave environment consists of wave scatter diagram data and wave directional probabilities. The scatter diagram data consists of annual probabilities of occurrence as fimctions of significant wave heights and peak periods in the structure installation site. For spectral fatigue analysis, a wave spectrum (e.g. Pierson-Moskowitz) is associated with each cell of the scatter diagram. Directional probabilities for fatigue waves are also included in the fatigue assessment. It is usually unconservative to ignore any non-uniform distribution in directional probabilities. However, in lieu of such information, the wind directional probability may be used to account for the non-uniformity in the wave approaching direction and to provide conservatism in the fatigue damage calculation. Stress Concentration Factors The determination of the appropriate SCF in the fatigue analysis is a complex task. It is also dependent on the S-N classification and stress analysis methods. The general rule of thumb is that the stress used in the fatigue analysis should resemble the fatigue stress obtained from the specimen tested when deriving the S-N curves. The fatigue stress does not mean the most accurate stress determined by the high-resolution fine mesh FEA. It is the pertinent stress, in Chapter 20 Spectral Fatigue Analysis and Design 383 Full penetration welds - T curve Partial penetration welds - W curve accordance with the chosen S-N curves. A discussion of the SCF and S-N classification is given in later Sections. The SCF can be determined based on parametric equations and finite element analysis. S-N Curves In the United States, the AWS (1997) S-N curves are used to analyze structural details of floating structures. Where variations of stress are applied to conventional weld details identified in Figure 9.1 of AWS (1 997), the associated S-N curves in Figures 9.2 or 9.3, should be used, depending on the degree of criticality. Where such variations of stress are applied to situations identified in AWS (1997) Table 10.3. The associated S-N curves are provided in AWS D1.l, Figure 10.6. For referenced S-N curves in AWS (1997), Figures 9.2,9.3 and 10.6, are Class Curves. For such curves the nominal stress range in the vicinity of the detail should be used. In Europe, UK DEn (1 990) S-N curves are used for structural details in floating structures. The S-N classification is determined based on the structural configurations, applied loading and welding quality. As discussed earlier, the UK DEn procedure is recommended in this chapter. Therefore, the S- N classification based on UK DEn curves will be discussed in detail, see Table 20.1. X curve is sufficiently devalued to account for thickness/size effect rable 20.1 Co! Subject S-N Curves S-N Classifications Fatigue Damage Assessment Cathodic Protection Welding Improvement parison between European Standards and US Standards Europe Standards (refers to e.g. UK DEn, 1990 ) US Standards (refers to e.g. AWS D1.l, 1997) Mean-minus-two-standard deviation Lower bound curves. One of 8 classes: B, C, D, E, F, F2, G and W, depending on geometry, stress direction, and method of fabrication and inmection. l Smooth weld metal merging with parent metal - X curve, otherwise, X’ curve Simplified Fatigue - The long-term wave height distribution can be represented by the Weibull distribution Or Spectral Fatigue Analysis Simplified Fatigue - The long-term wave height distribution may be represented by the sum of two Weibull distributions one for noma1 and the other for hurricane conditions Or Spectral Fatigue Analysis Cathodically protected joints in Seawater equivalent to joints in air. Unprotected joints in Seawater require S-N curve to be reduced by a factor of 2 on life. Included S-N curves (X’ and X) presume effective cathodic protection. Fatigue provisions of AWS D1.1 apply to members and joints in atmospheric service. Does not recommend further reduction of S-N curve for free corrosion. Not covered Use X curve rather than X’ curve 384 Part III Fatigue and Fracture Joint Classification Guidelines on joint classification may be found from the UK DEn (1990). Note that the S-N curves in the UK DEn (1 990) was modified by HSE( 1995). The UK DEn (1990) guidelines apply only to welded joints that are free from serious defects or discontinuities. Factors such as undercut at the toe, internal or surface breaking defects or cracks, and geometric irregularities may cause a reduction in fatigue strength and should be evaluated separately. The UK DEn (1990) guidelines allocate various types of welded joints into one of nine joint classes. To determine the correct classification for a particular weld detail, it is necessary to identify the weld type, the direction of the applied loading, and to consider all potential cracking locations. For most types of joint, the weld toes, weld ends, and weld roots are considered the most important locations. The joints with the highest classifications are those that are stressed in a direction parallel to the weld. Fillet or butt weld joints fall into Class C or B in the UK DEn (1990) guidelines depending on whether the manufacturing process is manual or automatic. Such joints seldom govern the fatigue strength of a welded details since other joints are likely to fall into lower joint classes. The classification of transverse butt welds is more complex. They can fall into Class D or E, depending upon the details of the manufacturing process, position, and location, all of which may influence the weld profile. Class C may be justified if the weld overfill is removed by grinding or the weld is shown to be free from significant defects by using non-destructive testing. However, if access is limited and the weld must be made from one side only, a lower fatigue strength should be assumed. The UK DEn (1 990) guidelines downgrade butt welds, made onto a permanent backing strip, to Class F. The guidelines also warn against the use of tack welds within small distances of the plates edge, in which case, the classification is lowered to Class G. Tack welds are a controversial topic. A number of studies have been conducted for different methods of attaching the backing to the plates prior to making the butt weld. Tacking the backing strip to the root preparation, and incorporating this into the final weld, gives small improvement in fatigue strength over joints in which the backing strip is fillet welded to one of the plates. However, the increase is not sufficient to warrant a higher joint classification. In both cases, failure may initiate at the root of the butt weld. Currently butt welds made onto temporary backing such as glass or ceramic backing strips are not classified and require further research. The availability of electrodes designed specifically for root runs has resulted in an improvement in the quality of single-sided welds made without backing. In recognition of this welding quality improvement, such joints can be considered as Class F2 if full penetration is achieved. This classification should be used with caution, because fatigue strength in some areas may be much lower due to lack of penetration at the root. The fatigue strength is seldom governed by butt welded joints, because these joints in general posses a superior strength over fillet welded joints. Fillet welds fall into Class F, F2, or G depending on their size, orientation, and location in relation to a free plate edge. However, recent studies have shown that fillet welds posses a fatigue strength lower than that predicted by Class G, if the weld is continued over the comer of the plate. Chapter 20 Spectral Fatigue Analysis and Design 385 In addition to the weld toe, which is the most usual site for fatigue cracking to occur, all load carrying fillet welds and partial penetration butt welds must be evaluated to assess possible weld throat failure. To avoid this type of failure, it is necessary to ensure that these joints are adequately dimensioned. This may be achieved using the Class W design S-N curve. One should note that the maximum shear stress range is associated with the class W design S-N curve. Structural Details The UK DEn fatigue design and assessment guidelines provide sketches, which provide assistance in the S-N classification of structural details. According to UK DEn (1990) guidelines, joints are subdivided into the following types: Metal free from welding Transverse butt welds Details in welded girders The UK DEn Curves were developed based on small test specimens. In the S-N classification of structural details, the users first carefully relate the fatigue stress in tests with the stress of structural details under consideration. For example, the fatigue stress in the test for the weld shown in Figure 20.2a, would be the tensile stress, S, on the cross-section, but for the weld shown in Figure 20.2b, it would be SCF S , where SCF is the stress concentration factor caused by the hole. This is due to the fact that at point x, the stress near the weld is SCF S . However, for a small cutout in Figure 20.4c, the stress concentration due to the small hole shall not be included since micro-structural effects have been included in the S-N curves. Continuous welds essentially parallel to the direction of applied stress Weld attachments on the surface of a stressed member Load-carrying fillet and T butt welds ts C Figure 20.2 Explanation of Fatigue Stress When Weld is Situated in Region of Stress Concentration Resulting from Structure’s Gross Shape Theoretically, structural details should be classified and considered for each loading step throughout the fatigue analysis since different loading steps result in different applied loading 386 Pari III Fatigue and Fracture directions. This approach is generally prohibitively complex. Therefore, simplified S-N classification is used based on the rule of thumb in engineering applications. When classifylng the weld's structural details in large, complex structural systems from a series of design drawings, it is important to: Consider each weld individually Figures 20.3 and 20.4 show two typical examples of details found in a floating structure. In the section shown in Figure 20.3, the classifications range from C to F2 and W, depending upon the direction of the applied stress. In these examples, stresses in the three principal directions S, , S, and S, , are not equal. Thus the design stress range for each class will differ. However, for simple design purposes, the maximum principal stress and F2 classification are assigned for the overall structural details. It is particularly difficult to classify the details that have a hole and to identify potential crack locations. Holes in a continuous longitudinal weld are covered in the UK DEn fatigue design guidelines as Class F, without requirement for an additional stress concentration factor. However, a web should be incorporated to this detail. The end of a web butt weld at the hole is a more severe detail that should be ground. For the ground detail, Class E or D is recommended. Due to the presence of the hole, a stress concentration factor of 2.2 or 2.4 should be included. If the end of the butt weld is not ground, a Class F or F2 curve, together with the geometrical stress concentration factor (2.2-2.4), is recommended. Consider each direction of applied stress Evaluate all possible cracking locations, because each may yield a different classification Consider any possible stress concentration effects Figure 20.3 S-N Classification of Structural Details Subjected to Triaxial Loading Chapter 20 Spectral Fatigue Analysis and Design 387 y,Lx 0 :F Figure 20.4 S-N Classification of Structural Details If concerns remain about the use of a cope hole, it is possible to improve its fatigue strength by cutting back and grinding the weld end as shown in Figure 20.3. In such cases, the weld between the flange and web should be full penetration over the regions on either side of the cope hole in order to avoid failure through the weld throat (W class). Figure 20.4 illustrates the third example of S-N classification of structural details. It’s the small bracket between the pontoon and the base node in a TLP structure. Based on the UK DEn (1990) Guidelines and published fatigue test data, the hotspot areas can be classified as F or F2. S-N classification of the structural details in floating structures is a challenging task. During the design process, there are many structural details, which cannot be classified based on the UK DEn (1990) guidelines. In this case, other design standards such as AWS (1997) or published fatigue test data may be used to justify the classification. 20.5.4 Fatigue Damage Assessment The fatigue life of structural details is calculated based on the S-N curve approach assuming linear cumulative damage (Palmgren-Miner rule). A spectral fatigue analysis is used where the long term stress range distribution is defined through a short term Rayleigh distribution within each short-term period for different wave directions. A one-slope or bi-linear S-N curve may be assumed. Fatigue lives are determined by the service life and safety factors. Additional margin is desirable due to the uncertainties associated with fatigue assessment procedures. Initial Hotspot Screening The objective of the initial screening is to identify the fatigue critical areas based on the experience and the in-service data. Fatigue damage is calculated for each element in the group assuming a conservative S-N curve and upper-bound SCF for each element. The calculated damages are reviewed and all elements with fatigue lives less than the minimum required, are analyzed in further detail in the specific hotspot analysis. [...]... of the material The TWI CTOD Design Curve was also adopted by the American Petroleum Institute in its API 110 4 (1983) as a basis for its fitness-for-purpose criteria 5 4 4 3 2 Wide Plate T s s et (shaded area) 1 0 0 1 2 3 4 5 €laY Figure 21.1 The British Welding Institute CTOD Design Curve Part 111 Fatigue and Fracture 394 21.3 Level 2: The CEGB R6 Diagram This Level 2 Assessment provides a simplified... reference during fabrication 20.6.3 Incorporation of Comments from Classification Society Comments on the design brief and the task report should be incorporated into the applicable revised document The revised document is issued for record and final approval, if required 20.7 References 1 ABS (2002), “Rule for Building and Classing Steel Vessels”, American Bureau of Shipping 2 API (1997), “Recommended... analysis results and design brief revisions 20.6.2 Submittal and Approval of Task Report A technical task report is issued after the analysis is completed to document the analysis and design results This report should follow the analysis methodology documented in the design brief and discuss any variations from the design brief The task report includes supporting information, hand calculations and computer... Assessment Using Failure Assessment Diagram The failure assessment diagram (FAD) may be used to predict residual strength of a cracked at member for a given set of fracture toughness and defect size, see P r 111 Section 21.1.2 If the peak stress exceeds the residual strength derived through FAD, failure may occur For the accurate prediction of residual strength, it is important to properly assess the maximum... Analysis”, in an ASCE book f entitled “Tension Leg Platforms - A State o the Art Review” Edited by Demirbilek, Z 10 HKS (2002), “ABAQUS/Standard User‘s Manual, Version 5.6”, Hibbitt, Karlsson & Sorensen, Inc 11 HSE (1995), “Offshore Installation, Guidance on Design, Construction and Certification”, UK Health and Safety Executives, 4th Edition, Section 21 12 Jha, A.K and Winterstein, S.R (1998), “Stochastic... Computational Services 16 UK DEn (1 990), “Offshore Installations: Guidance on Design, Construction, and Certification”, 3rd Edition, UK Department of Energy (Now UK Health and Safety Executives) Part I11 Fatigue and Fracture Chapter 21 Application of Fracture Mechanics 21.1 Introduction 21.1.1 General Applications of the fracture mechanics in marine structural design include: Assessment of final fracture,... conditions - each applied loading condition is checked for accuracy Analysis and validation - Analysis results are checked step-by-step; discrepancies between expected and obtained analysis results should be documented and explained Loading combinations checked for accuracy - each applied loading combination should be summarized and Environmental conditions - wave scatter diagram and directional probability... constant amplitude, the above equation may be re-written as: (2 1.10) If F does not dependent on a, the above equation may lead to (Almar-Naess, 1985): I-n/2 N, = acR ( c SAF 1-m/2 -a0 Y for m # 2 (21 .11) (1-rn12) The Paris parameters C and m may be found from Gurney (1979), IIW( 1996), BS 7910 (1999) and API 579 (2001) The values of C and m depend on the material, service environment and stress ratio... exists: CD =- E 0.0259 - 0.25 = - E, a a mx and therefore, the maximum half-width is am =O.O74m 21.9 References 1 Almar-Naess, A (1989, “FatigueHandbook, Oflshore Steel Structures”,Tapir, Norway 1 API 110 4 (1994), “Alternate Standards for Acceptability for Girth Welds Appendix A, Standards for Welding Pipelines and Related Facilities”, 18” edition, American Petroleum Institute 2 MI 579 (2001), “Recommended... Engineers 9 Dawes, M.G (1974), “Fracture Control in High Strength Weldments”, Welding Journal 53~369-S-379-S Edition, Cambridge University 10 Gurney, T.R (1979), “Fatigue of Welded Structures”, 2nd Press 11 Harris, D.O (1995), “Fatigue and Fracture Control in the Aerospace and Power Generation Industries from the book entitled “Prevention of Fracture in Ship Structures” by the Committee on Marine Structures, . is completed to document the analysis and design results. This report should follow the analysis methodology documented in the design brief and discuss any variations from the design brief. The. the Class W design S-N curve. One should note that the maximum shear stress range is associated with the class W design S-N curve. Structural Details The UK DEn fatigue design and. and potential redesign andor modification of welding procedures, and reanalysis. Specific Hotspot Design Structural details that do not pass the specific hotspot analysis are redesigned to improve