1、STP-PT-070DESIGN GUIDELINES FOR THE EFFECTS OF CREEP, FATIGUE as is the approach adopted in many other Codes. It is also recognized that if flaw size criteria are to be established using engineering mechanics then this can be accomplished independently of the overall approach to Design-By-Analysis.
2、In addition to the complexities highlighted by the outline of a Design-By-Analysis methodology, the implications for the overall structure and implementation of Design-By-Analysis methods within Section I are documented as a basis for future discussions to facilitate reaching consensus among the tec
3、hnical community on the best approach for specific aspects, and for the overall approach to adoption within the ASME Code. STP-PT-070: Design Guidelines for the Effects of Creep, Fatigue or may cover geometries, loadings, or modes of failure that are not easily dealt with in the context of “Design-B
4、y-Formula”. The two approaches can, and have, been shown to be able to be incorporated into common sets of rules. If the two approaches are consistent then they can be used in a complementary fashion. It is possible to use “Design-By-Formula” for basic sizing and for sizing of non-critical features
5、on a component. For critical features, where perhaps it is necessary to strike a compromise between creep and fatigue, or a creep-fatigue interaction when both mechanisms might be active, then “Design-By-Analysis” can be used STP-PT-070: Design Guidelines for the Effects of Creep, Fatigue this is th
6、e crux of the challenge for implementation of creep-fatigue rules within design ASME Codes. STP-PT-070: Design Guidelines for the Effects of Creep, Fatigue Section III, Subsection NH) use a Tresca criterion, whereas many other Codes (e.g. Section VIII, Div. 2) use a Mises criterion. Within EN13445-3
7、, Annex B the limit analysis can actually be performed with a Mises criterion (largely due to the fact that few finite element programs provide a Tresca yield criterion) but then the result must be “corrected” to the Tresca criterion. Hence, when comparing conservatism on loads and allowable stresse
8、s this “additional” conservatism (which is theoretically dependent on the stress state) should also be considered. This also raises the question as to “what is the most appropriate criterion?” The state-of-knowledge for yielding of steels indicates that for the time-independent regime the von-Mises
9、criterion is appropriate. However, for the time dependent regime then the multiaxial stress rupture criterion varies somewhat between materials with numerous models existing that combine von-Mises, principal and/or hydrostatic stress. Hence, for the time dependent regime, for protection against cree
10、p rupture, it would appear more reasonable to simplify to a Tresca criterion (per EN13445-3, Annex B). 3.1.2 Reference Temperature for Calculations Another topic that can complicate interpretation of conservatism and comparison of design margins, particularly for cyclic analysis, between Codes is th
11、at of temperature. In simple terms, the Section VIII, Div. 2 specifies that the temperature for calculations (material properties such as stress-strain response and S-N curve) shall be the average temperature of the cycle; implying an average between the maximum and minimum temperature of the cycle.
12、 The EN Codes (e.g. EN13445-3, Annex B) specifies that the temperature for calculations shall be 25% of the minimum temperature plus 75% of the maximum temperature, thereby producing a bias toward the higher temperature portion of the cycle and thus likely resulting in lower property (strength) valu
13、es than would be used for an equivalent calculation in the Section VIII, Div. 2. The Section III, Subsection NH generally requires that calculations are performed using properties for the maximum temperature of the cycle. To complicate comparisons further, within the EN Codes the distinction is made
14、 that the maximum and minimum temperatures shall be taken at the respective instants of the maximum and minimum stress in the cycle (and thereby might not have any correlation to the actual maximum and minimum temperatures for the cycle). This definition of maximum and minimum temperatures coincidin
15、g with maximum and minimum stresses does not appear to be inferred or implied by the rules in the Section VIII, Div. 2, although it is incorporated into some aspects of Section III, Subsection NH (e.g. selection of hot and cold yield strength). 3.1.3 Perfectly Plasticity vs. Strain Hardening For the
16、 basic load bearing capacity and shakedown / ratcheting design checks, the majority of Codes base the material behavior on a simplified elastic-perfectly plastic constitutive model. This simplification makes both monotonic and cyclic analysis easier to interpret (particularly in the case of cyclic a
17、nalysis where it removes the need to define a hardening criterion). While ASME (e.g. Section VIII, Div. 2) provides this option (at least for monotonic analysis), the option is also given to use the strain-hardening characteristic in monotonic (collapse) and cyclic (fatigue) calculations. The Sectio
18、n VIII, Div. 2 also allows the use of a strain hardening based on the average temperature for the cycle because of the restriction to the time-independent regime. For the time-dependent (creep) regime, the strain-hardening response can be quite different between extremes of the cycle (low temperatur
19、e and high temperature), so using an average is likely inappropriate, plus the strain hardening response will affect the stress in creep dwells (and hence the creep damage calculation). Properly accounting for these aspects in the time-dependent regime is complex and of the Codes and Standards revie
20、wed, only the extensive procedures (and associated material data requirement) of the R5 Assessment Procedure appear to provide a comprehensive and consistent technical treatment. STP-PT-070: Design Guidelines for the Effects of Creep, Fatigue hence the rules do not consider creep or creep-fatigue in
21、teraction. The procedures provide assessments that can utilize elastic analysis (using stress categorization methods) or inelastic analysis. The review here focuses on the inelastic approach since this overcomes many of the inherent difficulties of the stress classification approach (indeed, 5.2.1.2
22、 in Section VIII, Div. 2 recommends use of the inelastic approach for complex stress fields and loadings) and this approach can be compared and contrasted with other Codes following similar approaches. For global plastic collapse the Code offers a limit load analysis method which follows the normal
23、approach using an elastic perfectly-plastic constitutive model where the yield strength is equal to the appropriate material design strength. The von-Mises yield function is used for the calculation. The Code also offers an elastic-plastic analysis method where the actual strain hardening characteri
24、stic of the material may be considered, providing that the model provides perfectly plastic behavior beyond the true ultimate stress. This also uses a von-Mises yield function. As with a conventional limit analysis, the STP-PT-070: Design Guidelines for the Effects of Creep, Fatigue but it is noted
25、that the present Section VIII, Div. 2 is restricted to the time independent regime (although it is further noted that in Annex 3-D, which provides the cyclic stress-strain curves, parameters are given at temperatures that extend into the time-dependent regime for some materials). STP-PT-070: Design
26、Guidelines for the Effects of Creep, Fatigue that is, no enhancement of creep due to plasticity or vice versa is considered) and that the resulting independent damage values are then plotted on an interaction diagram that is material dependent. While a bi-linear interaction diagram is used suggestin
27、g stronger than linear interaction, the predictions could be non-conservative because any true creep-fatigue damage interaction is not accounted for. Appendix F provides the majority of material data required for deformation calculations. It should be noted that the Omega creep model does not includ
28、e primary creep which should be considered for accurate prediction of stress relaxation in many creep-fatigue cycles. Overall, this approach appears far from the state-of-the-art for creep-fatigue interaction calculations. Mode I and Mode II Creep-Fatigue assessment of Dissimilar Metal Weld (DMW) Jo
29、ints are also included, although this is restricted to 2.25Cr1Mo ferritic to austenitic steels with either a stainless-steel or nickel-based filler metal. 3.4 ASME BPVC Section III, Subsection NH This subsection of the ASME Code contains rules for nuclear facility components covering so-called Class
30、 1 Components in Elevated Temperature Service. The rules in this portion of the Code date back many years, with this originally being Code Case N47 (which is often the designation found in open literature referencing the ASME Nuclear Creep-Fatigue Rules). These rules are unique within the ASME Codes
31、 because they state the need to address time dependent behavior including creep-fatigue interaction. Within this subsection, Article NH-3000 Design includes NH-3200 Design-By-Analysis which provides background and definitions associated with the underlying approach which is based on stress categoriz
32、ation from an elastic analysis. However, inelastic analysis is also permitted (NH-3214.2) and it is noted that Appendix T (which will be discussed later) was established with the expectation that inelastic analysis would sometimes be required. For limits on primary stress, the Code provides time and
33、 temperature dependent strength values (rather than just a time independent, temperature dependent strength value as is the case for the current edition of Section I). The specific treatment of primary membrane, local and bending stresses, and the associated margins applied to the material strength
34、data vary with the level classification applied to the service loadings. To apply the rules, the operating times for various service loadings must be classified into a Load Histogram (NH-3114) and a linear time fraction damage summation is used to control their cumulative effect. The effects of seco
35、ndary stresses are managed through Limits on Deformation Controlled Quantities (NH-3250) for which acceptability criteria and material properties are contained in Appendix T (although it is noted that alternative criteria may be applied subject to approval by the owner). The non-mandatory Appendix T
36、 is entitled “Rules for Strain, Deformation, and Fatigue Limits at Elevated Temperatures.” This appendix includes limits for inelastic strains which can be demonstrated to be satisfied by either elastic analysis, simplified inelastic analysis, or detailed inelastic analysis. The methods for detailed
37、 inelastic analysis are not specified but the description of inelastic analysis provided in NH-3214.2 notes “The constitutive equations, which describe the inelastic behavior, should reflect the following features STP-PT-070: Design Guidelines for the Effects of Creep, Fatigue primary creep and the
38、effects of creep strain hardening as well as softening (due to reverse loadings); and the effects of prior creep on subsequent plasticity, and vice versa.” Practically, such constitutive equations are not available and hence elastic or simplified inelastic analyses are the only feasible options. If
39、elastic analysis is used then some simple tests are provided to demonstrate that the combination of primary and secondary stress remains within the elastic range and that creep damage and strains are negligible. Such restrictions are invariably prohibitive thereby requiring simplified inelastic anal
40、ysis. In addition to some basic tests on deformation (NH-T-1330), a general procedure for creep-fatigue evaluation is provided (NH-T-1400) which is the widely referenced “ASME Creep-Fatigue approach”. A number of calculations utilize a core stress, which is defined as the stress controlling the ratc
41、heting creep strain for the cycle (which is approximately the stress within the core of the structure). In other methodologies, such as R5 which will be discussed later, this would be termed a shakedown reference stress. Estimation of this core stress is based on the ODonnell/Porowski modification o
42、f the Bree diagram (Figure NH-T-1332-1) which was derived for the wall of a pressurized tube with constant membrane stress (due to pressure) combined with a cyclic bending stress (due to through-wall temperature gradient). For this specific case, the diagram provides the core stress for any combinat
43、ion of primary and secondary stress. An alternative diagram (Figure NH-T-1332-2) is provided for determination of core stress in general structures but this is highly conservative in many cases. Strictly the core stress is a function of both the geometry and loading so the simplifications introduced
44、 are quite dramatic. Much has been written about the overall approach to creep-fatigue calculation within these rules so only an outline is provided here of some of the key features. (a) The total strain range is estimated (NH-T-1432) using Neubers method in conjunction with the (modified) time inde
45、pendent isochronous stress-strain curve. This strain range includes adjustments for multiaxial plasticity and Poissons ratio. Also, an enhancement is applied for the creep strain increment during the cycle due to creep at the core stress. The fatigue damage is calculated based on this total strain r
46、ange. (b) Creep damage is calculated (NH-T-1433) by considering the stress relaxation during the cycle. The stress history is estimated by assuming that stress relaxation occurs at constant strain (no elastic follow-up) using isochronous stress strain curves. However, the stress cannot relax to a va
47、lue less than 1.25 times the core stress. This provides a stress history from which creep damage is evaluated using a life fraction rule. (c) The calculated creep and fatigue damage values are plotted on an interaction diagram to demonstrate that the combined effect is acceptable. The interaction di
48、agram is material dependent, but in all cases a bi-linear interaction is specified with the result that the sum of the creep and fatigue damage is less than unity (in some cases greatly so). This methodology has been in place for many years without significant modification but the principal concerns
49、 with the approach stem from the lack of generality in some of the simplifying assumptions. Specifically, calculation of the core stress is overly simplified so, for complex geometry and/or complex loading the core stress estimates which are key to both creep and fatigue damage predictions in the time dependent range are invariably inaccurate. The other principal contention with the approach is that stress relaxation is assumed to occur without elastic follow-up from the peak stress in the cycle. Few real components would satisfy that requirement which results in a more rapid stress r
