ML15160A220

NYS000520 - Kirk, M. Et Al., Assessment of Fracture Toughness Models for Ferritic Steels Used in Section XI of the ASME Code Relative to Current Data-Based Model, PVP2014-28540, Proceedings of PVP2014, 2014 ASME Pressure Vessels and Piping

ML15160A220
Person / Time
Site:	Indian Point
Issue date:	06/09/2015
From:	State of NY, Office of the Attorney General
To:	Atomic Safety and Licensing Board Panel
SECY RAS
References
RAS 27910, ASLBP 07-858-03-LR-BD01, 50-247-LR, 50-286-LR
Download: ML15160A220 (16)

Text

Proceedings of PVP2014 2014 ASME Pressure Vessels and Piping Division Conference July 20-24, 2014, Anaheim, CA, USA 1 PVP2014-28540

ASSESSMENT OF FRACTURE TOUGHNESS MODELS FOR FERRITIC STEELS USED IN SECTION XI OF THE ASME CODE RELATIVE TO CURRENT DATA-BASED MODELS Mark Kirk 1 Senior Materials Engineer, Nuclear Regulatory Commission, Rockville, MD, USA, mark.kirk@nrc.gov Marjorie Erickson President, Phoenix Engineering Associates, Inc. Unity, NH, USA, erickson@peaiconsulting.com William Server President, ATI Consulting, Black Mountain, NC, USA, wserver@ati-consulting.com Gary Stevens 1 Senior Materials Engineer, Nuclear Regulatory Commission, Rockville, MD, USA, gary.stevens@nrc.gov Russ ell Cipolla Principal Engineer, Intertek AIM , Sunnyvale, CA, USA, russell.cipolla@intertek.com ABSTRACT Section XI of the ASME Code provides models of the fracture toughness of ferritic steel. Recent efforts have been made to incorporate new information, such as the Code Cases that use the Master Curve, but the fracture toughness models in Section XI have, for the most part, remained unchanged since the K Ic and K Ia curves were first developed in Welding Research Council Bulletin 175 in 1972. Since 1972, considerable advancements to the state of knowledge, both theoretical and practical have occurred, particularly with regard to the amount of available data. For example, as part of thermal shock (PTS) re-evaluation efforts the NRC and the industry jointly developed an integrated model that predicts the mean trends and scatter of the fracture toughness of ferritic steels throughout the temperature range from the lower shelf to the upper shelf. This collection of models was used by the NRC to establish the index temperature screening limits adopted in the Alternate PTS Rule documented in Title 10 to the U.S. Code of Federal Regulations (CFR), Part 50.61a (10CFR50.61a). In this paper the predictions of the toughness models used by the ASME Code are compared with these newer models (that are based on considerably more data) to identify areas where the ASME Code could be improved. Such improvements include the following:

On the lower shelf, the low-temperature asymptote of the K Ic curve does not represent a lower bound to all available data. On the upper shelf, the de facto K Ic limit of applicability of 22 0 irradiation effects. The separation between the K Ic and K Ia curves depends on the amount of irradiation embrittlement, a functionality not captured

by the ASME Section XI equation

s. The temperature above which upper shelf behavior can be expected depends on the amount of irradiation embrittlement, a functionality not captured in the ASME Section XI equations.

1 BACKGROUND AND OBJECTIVE A key input to assessments of the integrity of operating structures in the presence of real or postulated defects is the fracture toughness of the material in question. Various parts of the ASME Code (that is, the Code itself, various Nonmandatory Appendices, and Code Cases) provide models of the fracture toughness properties of ferritic steels

principally the K Ic and K Ia curves developed in the early 1970s [1]. The Code has recently been expanded to include procedures to estimate

RT NDT using the Master Curve index temperature T o [2-3], and to estimate the temperature above which an EPFM-based analysis is needed [4]. Since the early 1970s, considerable advancements to the state of knowledge, both theoretical and practical have occurred, particularly with regard to the amount of data available and empirical models derived from these data. For example, as part of the U.S.

(PTS) re-evaluation efforts, reports from which were issued in early 2010, the NRC and the industry jointly developed an integrated model that predicts the mean trends and scatter of the fracture toughness of ferritic steels throughout the complete temperature range from the lower shelf to the upper shelf [5]. This collection of models was used by the NRC in the probabilistic fracture mechanics (PFM) Code, Fracture Analysis Vessels Oak Ridge (FAVOR) [6], to establish the index temperature screening limits adopted in the Alternate PTS Rule 1 The views expressed herein are those of these authors and do not represent an official position of the NRC.

This material is the work of the United States Government and is not subject to copyright protection in the United States. This paper is approved for public release with unlimited distribution.

2 documented in Title 10 to the U.S. Code of Federal Regulations (CFR), Part 50.61a (10CFR50.61a) [7].

The objective of this paper is to compare the predictions of the toughness models within the ASME Code with the newer models that are based on considerably more data to identify areas where the ASME Code could be improved. ASME CODE TOUGHNESS MODELS The K Ic and K Ia curves that appear in Article A-4200 of Nonmandatory Appendix A to ASME Section XI [8] (the K Ic curve also appears in Appendix G [9]), are expressed as follows (all equations in this paper are expressed in SI units):

(1) (2) Eqs. (1-2) are intended to represent the lower bounds of K Ic and K Ia data. These estimates of K Ic and KIa depend on the index temperature RT N DT, which is determined per ASME NB-2331 [10]. Where appropriate, RT NDT is adjusted to account for the effects of neutron irradiation embrittlement. Neither Appendix A nor Appendix G, place an upper-limit on the K Ic value that may be estimates using eqs.

(1-2). Nevertheless, a valde facto limit on K Ic despite scarce mention or defense of this value in the literature. The basis for this limit and an assessment of its accuracy relative to data appears in [

11]. CURRENT DATA-BASED TOUGHNESS MODELS In 1984 Wallin and co-workers began publication of a series of papers that, collectively, describe what has come to b-14]. The Master Curve quantifies the temperature dependence and scatter of the fracture toughness (i.e., K Jc or J c values) of ferritic steels in the fracture mode transition temperature region.

Existing work in which large databases were examined demonstrates that the temperature dependence and scatter of the Master Curve are consistent for all ferritic steels [15-16]2. All that needs to be determined for a particular material is the Master Curve index temperature (T o), which positions the Master Curve on the temperature axis. Using ASTM E1921 protocols it is possible to estimate T o using as few as six fracture toughness specimens [19], providing the possibility that T o can be directly determined from the specimens already placed in the surveillance capsules of nuclear reactor pressure vessels. Over the past fifteen years papers have been published that expand upon and extend the Master Curve concept. These papers describe:

The temperature dependence and scatter in crack arrest fracture toughness (K Ia) [20]. The temperature separation of the K Jc and K Ia transition curves, and how this separation changes with the condition of the material [20-21] The temperature dependence and scatter in upper shelf fracture toughness data (JIc) [22-23], and 2 Some studies by Wallin suggest adjustments to the Master Curve lower bound value on the lower shelf [

17], and of the temperature dependence in the case of extremely high embrittlement [

18]. In absolute terms the magnitude of these adjustments are minor, having only small effects on the predicted values. These adjustments could be considered as a further improvement to the models suggested herein should the cognizant ASME Code groups decide to adopt these models. The relationship between transition fracture toughness and upper shelf fracture toughness values [24

]. References [

4, 5, 20-24] discuss both the empirical and physical bases for these relationships, which were developed from large databases (data numbering in the hundreds) covering a wide range of material conditions (e.g., different product forms, different irradiation exposures, different material chemistries). These papers provide information demonstrating that, like the Master Curve, these relationships can be expected to apply with comparable accuracy to all ferritic steels irrespective of composition, product form, heat treatment, degree of hardening, degree of irradiation damage, etc. Significantly, all of these models are linked via a single parameter:

T o. Once T o is determined the mean initiation and arrest toughness behavior, and the scatter about the mean, can be determined from lower shelf through upper shelf using combinations of the models presented in Table 1.

Figure 1 shows the curve shapes and relationships of the se models and defines their variables visual ly. In addition to the models shown in Figure 1, Table 1 includes the equation for RT To provided by Code Cases N-629 and N-631 (and now in the ASME Code Section XI, Appendices A and G).

RT To provides an alternative to RT NDT such that T o can be used to index the ASME K Ic and K Ia lower bound curves.

RT To thereby links the data-based models of Table 1 to the ASME models for K Ic and K Ia. This linkage enables comparison o f the ASME lower bound descriptions of transition initiation and arrest toughness to the data-based models of initiation and arrest toughness. COMPARISON OF ASME MODELS TO CURRENT DATA-BASED MODELS The data-based models of ferritic steel toughness summarized in Table 1 can be compared with the ASME models for K Ic and K Ia to identify situations where the ASME models adequately reflect the data versus situations where the ASME models could potentially be improved. In the next two subsections [A and B] the comparisons made can be interpreted in two ways: (1) either as an assessment of the accuracy of the ASME models when RT To (i.e., T o) is used as an index temperature, or (2) as an assessment of the accuracy of the ASM E models when the index temperature RT NDT is used and RTNDT exceeds T o -examines these analyses to assess the effect of the NDT/Charpy-based value of RTNDT being other than T o, which is often the case. Finally, an assessment is made in Subsection D of the influence margin terms have on the ability of the ASME models to represent, or conservatively bound, the fracture toughness data. A. Crack Initiation K Ic and Upper Shelf Figure 2 compares the predictions of the data-based models for K Ic / K Jc and J Ic with the ASME K Ic curve augmented by the de facto upper RT To values: -100, 0, and +100 °C (these being chosen to examine a range of hardening that could result from, as an example, neutron radiation embrittlement). These graphs support the following observations:

The ASME model over-estimates the lower shelf fracture RT To for all values of RT To. For un-irradiated materials such low temperatures cannot be achieved during normal operations. However, as radiation embrittlement causes the material transition temperature to approach regulatory limits (e.g., the PTS limits of 132 to 149 °C in 10CFR50.61 [25]) a temperature 60 °C below these values is clearly within the range achievable during a cool-down.

3 In the transition regime between lower shelf and upper shelf the ASME model maintains a consistent location below the data, thus providing a conservative estimate of K Jc. The well recognized difference between the temperature dependence of the data and that of the ASME model is also evident in these plots. While these differences could affect the outcome of probabilistic assessments, which are more sensitive to changes in the models near the lower bound, they are not expected to adversely affect deterministic assessments performed according to ASME SC-XI Appendix A or Appendix G.

Table 1. Summary of Data-Based Toughness Models for Ferritic Steels. refs Model Equations Eq. # [2-3] RT To Reference temperature for ASME K Ic and K Ia curves based on T o (3) [1 2-1 4] K Jc Temperature dependence of median fracture toughness of a 1T specimen (4) Scatter at a fixed temperature

, where (5) Size effect where B x is the thickness of the specimen of interest while B o is the reference thickness (1

-in., or 25.4 mm).

(6) [20] K Ia Temperature dependence of mean K Ia (7a) Scatter at a fixed temperature Lo g-normal with a variance () equal to 18% of the mean value.

(7b) [22-23] J Ic Temperature dependence (temperature is in Kelvin)

= 0.3 T ref = 288C (or 561K)

C 1 = 1033 MPa C 2 = 0.00698/K C 3 = 0.000415/K

= 0.0004/sec (8) Scatter at a fixed temperature where (9) [21 , 34] Linkage of K Jc and K Ia data The standard deviation of ln (T K Ia-T o ) is 0.383. (10) [24] Linkage of K Jc and J Ic data (11) The accuracy of the de facto ASME upper limit of on K Ic is strongly compromised by increasing RT To [11]. Above RT To exceeds the upper shelf fracture toughness of most RPV steels by a considerable amount, suggesting a practical limit on K Ic should be informed by the upper shelf fracture toughness.

The upper shelf of many ferritic materials falls below J0.1 is used as the characterizing parameter (see the Appendix for further discussion).

T US, defined in Table 1, 4 can be used to define the upper limit of applicability for the K Ic curve based on data [11].

3 B. Crack Arrest Figure 3 compares the predictions of the data-based models for K Ic / K Jc and K Ia with the ASME K Ic and K Ia curves for three RT To values: -100, 0, and +100 °C (these being chosen to examine a range of hardening that could result from, as an example, neutron radiation embrittlement). The data-based models show that as RT To increases the K Ic and K Ia curves converge. This convergence is not a feature of the ASME models, which maintain a constant temperature separation between them. The result of using a constant temperature separation to represent the actual material behavior is that the ASME model is overly pessimistic for high values of RTTo indicative of highly irradiated material and over-estimates K Ia at low values of RT To. C. Sensitivity Study Based on the T o - RTNDT Relationship As described in

[2-3], RT To was defined as T o + 19.4 °C so that the K Ic curve indexed to RT To toughn[26] as a curve bounding approximately 95% of the data, a finding also validated by Wallin [27]. The consistent placement of the reference

K Ic curve enabled by the use of a true fracture toughness measure like T o ensures that this degree of bounding will occur for all materials. Thus, the analysis of the previous section [B] applies to situations where RT To is used as an index temperature, or to materials for which RT NDT exceeds T o RT NDT does not always equal T o +19.4 °C. Figure 4 draws together data from two sources (one being the source used to establish the epistemic uncertainty in

RT NDT for the PFM computer code FAVOR [6]) to illustrate that the degree by which RT NDT exceeds (or in one case does not)

T o varies by a considerable degree. The following values of (RT NDT -T o) were taken from these data and used to re-plot the ASME models in the format of Figure 2 and Figure 3 to assess the degree to which the ASME models represent the data for different values of RT NDT: RT NDT = T o + 0 °C: This is an extreme case, but nonetheless possible based on these data and moreover is representative of a worst-case assessment of the conservatism of the ASME models. RT NDT = T o + 47 °C: This represents the median of the distribution of (RT NDT -T o) values shown on Figure 4. Figure 5 illustrates that the deficiencies of the ASME K Ic model i n representing the lower shelf and upper limit of applicability noted in the discussion of Figure 2 are not influenced by the difference between RT NDT and T o. For materials where RT NDT exceeds T o by less than 19.4

°C, Figure 5 illustrates that non-conservative predictions of K Ic in transition can also be expected.

Figure 6 illustrates that RT NDT would need to exceed T o by more than 47 °C (a value of approximately 65 °C would be needed) to ensure that the ASME K Ia model provides a conservative bound to the K Ia data for the range of RT To values considered. The data in Figure 4 demonstrate that such materials make up only 25% of the population of RPV steels. Thus, the use of RT To to index the K Ia curve will provide a non-conservative representation for approximately 75% of RPV materials.

3 It should be noted that with the exception of Code Case N-749 [4] the ASME Code does not now explicitly consider the area from the upper transition region to the upper shelf. Instead the Code independently assesses cleavage fracture in the transition region using linear elastic fracture mechanics (LEFM), and on the upper shelf using elastic-plastic fracture mechanics (EPFM), in each case with appropriate margin terms. Another paper presented at this conference [34] corrects this situation by using eq. (10) to adjust RT To so that it can be used to index the K Ia curve. D. Effect of Margins Taken as a whole, the information in subsections [A], [B], and [C] leads to the conclusion that the ASME models are out of date and, in some cases, non-conservative. Especially on the lower shelf, the upper transition limit, and for all of the crack arrest transition, the ASME models fail to capture trends clearly evident in data now available.

Even the application of large margins implicit to some RT NDT values does not fully ameliorate the non-conservatisms identified here (see Figure 4). Table 2. ASME Code margin terms.

Code Ref. Margin Values Equation Appendix G 2 for current practice 1 for risk informed IWB-3612/ Appendix A anticipated loading faulted loading In view of these observations, the question arises as to how the ASME Code models survived for over 40 years without these inaccuracies becoming evident. The primary explanation is the margins placed on loading, flaw size, and the effects of radiation embrittlement. The deterministic method described in the ASME Code, Appendix G, uses these margins with the expectation that they account for these discrepancies between the data and the ASME models..Some of the specific margins applied are listed below:

Margin on irradiated RT NDT: The ASME Code states that RT NDT should be adjusted to account for the effects of radiation damage. Regulatory Guide 1.99 Revision 2 suggests a margin term that accounts for uncertainty in the unirradiated value of RTNDT and in the estimated shift of this value due to irradiation damage [29]. The value for each of these terms depends on several factors (e.g.,

product form, generic or measured values, credibility of surveillance data, etc.). To provide a sense of the magnitude of the margin term the range of values adopted for the various

database is provided in Figure 7 [30]. Margin on K I: Table 2 summarizes some of the different margins applied to the value of K I in different parts of the ASME Code. Margin due to flaw size

This margin is generally coupled with the aforementioned margin on K I. For ASME Code Appendix G calculations for pressure temperature operating curves, an assumed large flaw size of 1/4-thickness of the vessel wall is used. In IWB-3611/Appendix A where a detected flaw is being evaluated, the margin on flaw size is a factor of 10 relative to the critical flaw sizeappears in Table 2. By way of example Figure 8 illustrates the effect of these margin terms on the situations from Figure 5 and Figure 6 where the ASME models are most in need of margin to be conservative.

The upper graph in Figure 8 shows that a RT NDT margin of +20 °C (approximately the median from Figure 7) plus a K I margin of K Ic model bounding on lower shelf and in transition for most steels. However, a larger K I margin (=2) would be needed to ensure bounding on the upper 5 shelf for high RT To materials. It should be noted that the effectiveness of the Appendix G K I margins, which are just applied to the K arising due to pressure, are not assessed here. Since these margins apply only to a portion of K, greater values (i.e., greater than 2) would be needed to ensure bounding on the upper shelf. The lower graph in Figure 8 shows that a RT NDT margin of +40 °C (higher than any of the values in Figure 7) plus a K I margin of K Ia model bounding on lower shelf and in transition for most steels.

These examples illustrate that in many, but certainly not all, cases combination of the ASME models and these margin terms results in a conservative characterization of K Ic and K Ia relative to data now available. Nevertheless, the degree of conservatism is by no means consistent across the range of conditions found in the operating fleet. CONCLUSIONS A. For RTNDT-based Characterizations For RT NDT-based assessments the comparisons of the ASME model predictions to the data demonstrate the necessity of maintaining the current margins on RT NDT and on K I to compensate for the inaccuracies of the ASME models. Maintenance of these margins is expected to enable the ASME models to produce conservative characterizations in most cases. However, the information presented herein demonstrates that conservatism cannot be guaranteed in all cases. Additionally, the actual margin achieved by the combination of explicit and implicit margins currently adopted by the Code varies considerably across the fleet, and instances can be found where the actual margin achieved is less than one would wish. In our view, the best remedy to ensure consistent and quantifiable margins is the adoption of a T o-based assessment strategy. Such a goal has been the focus of Code Cases and various activities within the Code over the past fifteen years. The following section summarizes briefly these past efforts, and goes on to suggest a more integrated approach that could be adopted moving forward. B. For T o-based Characterizations Existing Code activities address some, but not all, of the needed components of a model that is fully consistent with the data and with current-day understandings of the fracture toughness behavior of ferritic steels. These activities, which are in some cases in accord with the data and in other cases are not, are listed in Table 3. To augment these activities and bring them into closer accord with the underlying data , this paper summarizes a model of the fracture toughness behavior across the entire temperature range from lower to upper shelf. This model, which can be based on the information summarized herein, depends only on a value of T o and on a selected bounding probability value, p, where (for example), p could equal 0.05 as is typical for many engineering applications.

Table 4 summarizes the equations, inputs and constants, and limits that fully define this model for calculation of central tendency (median, mean) values while Table 5 summarizes the same information for calculation of a bounding value of fracture toughness.

These tables do not present new equations, but rather recast those previously presented in Table 1 in a compact format. The equations of Table 4 and Table 5 capture the effect of radiation embrittlement (or, equivalently, of T o) on the interrelationships of K Jc , K Ia , J Ic, and J0.1. This feature absent from existing ASME models, and its absence is responsible for many of the inaccuracies of ASME models noted herein. What remains absent from Table 4 and Table 5 are equations to predict the effects of radiation on T o if this effect is not measured. Various embrittlement trend curves are available for this purpose [29, 31-33] for ASME use in replacing current toughness models the equations given in Table 5 could be expressed as a low probability (p) value (e.g., 5%), with the specific p-value perhaps tied to the loading condition as is now the practice in IWB-3612. The appropriate margins to use with this data-consistent approach, beyond the margin associated with this selection of a p-value, merits further discussion. Reconsidering the types of margins used currently: Table 3. ASME Code use of models from Table 1. ID Summary Accuracy N-629 [2] N-631 [3] App. A App. G Use RT To to index the K Ic and K Ia curves K Ic: OK in transition. Need margin on lower shelf K Ia: inconsistent with the data, but being fixed [34]

N-749 [4] Defines a temperature (T c) above which EPFM analysis is needed The temperature T c is not a function of T o, which is not consistent with the data. Proposed N-830 [35] Re-defines the K Ic curve as a fixed percentile of the Master Curve in transition and a constant value on lower shelf. The constant lower shelf value follows current ASME practice but is not consistent with the Master Curve bound and fracture toughness data.

Margin on T o: With the elimination of the uncertainty on the unirradiated value of RT NDT (which, as illustrated here, consumed a large part of the existing margin), less margin seems necessary. It is suggested that two factors be considered: the measurement uncertainty on T o (see [19]), and the uncertainty in the embrittlement shift prediction (see [29, 31-33]). As is current practice a square root sum of squares combination of these two factors can be used. Additionally, should a direct measurement of T o in the irradiated condition be made then some reduction of these margin terms would be in order. Margin on K: The values of K Ic and K Ia on the lower shelf that are exhibited by the data, and are therefore reflected by equations listed in Table 4 and Table 5, are lower than the values predicted by current ASME models. It is recognized that this presents a particular challenge within the ASME Section XI Appendix G context of setting P-T limits for normal operation because a positive head pressure on the pumps needs to be maintained. Figure 8 margin term on K I produces K Ic and K Ia values roughly equivalent to a 2.5 th percentile curve of the equations listed in Table 5 In view of the added accuracy associated with a To-based approach the use of a bounding p-value and no additional margin on K I seems appropriate. Should this not produce a sufficiently wide P-T corridor for routine heat-up and cool-down it is suggested that the conservatism inherent to the use of a 1/4t flaw in establishing the P-T limits be revisited by the Code.

SUMMARY

In 1972 when ASME adopted the K Ic and K Ia curves and much of the current margin approach, the margins were established based mostly on engineering judgment. Now, ample data exists to build more accurate toughness models and to quantify the amount of margin needed to bound the data; this paper addresses both topics. The models summarized in this paper are based on T o; they produce safety margins that are both consistent and quantifiable across the fleet, a benefit that cannot be obtained within the current correlative 6 framework based on RT NDT. Additionally, while the information presented here demonstrates the conservatism of the current RT NDTbased approach in most cases, available data shows it may be non-conservative in some situations. It is suggested that the cognizant ASME Code committees consider these models so that the Code can achieve the benefits made possible by use of current technology and data. REFERENCES [1] Recommendations on Toughness Requirements for Ferritic August 1972. [2] American Society Of Mechanical Engineers, Use of Fracture Toughness Test Data to Establish Reference Temperature for Pressure Retaining Materials,Section XI, Division 1, ASME Boiler and Pressure Vessel Code Case N-629, ASME, New York (1999). [3] American Society Of Mechanical Engineers, Use of Fracture Toughness Test Data to Establish Reference Temperature for Pressure Retaining Materials Other than Bolting for Class 1 Vessels,Section III, Division 1, ASME Boiler and Pressure Vessel Code: An American National Standard, Code Case N-631, ASME, New York (1999). [4] American Society Of Mechanical Engineers, Alternative Acceptance Criteria for Flaws in Ferritic Steel Components Operating in the Upper Shelf Temperature Range,Section XI, Division 1, ASME Boiler and Pressure Vessel Code Case N-749, ASME, New York (2011

). [5] Unified Model for the Fracture Toughness of Ferritic Steels in the Transition and on the Upper Shelf in Fitness-for-Service Assessment and in the Design of Fracture Toughness Conference, PVP2006-ICPVT11-93652. [6] Williams, P., DicksonVessels Oak Ridge FAVOR, v12.1, Computer Code: Theory and Implementation of Algorithms, Methods, and ORNL/TM-2012/567, United States Nuclear Regulatory Commission ADAMS Accession Number ML13008A014, (2012). [7]

http://www.nrc.gov/reading-rm/doc-collections/cfr/part050/part050-0061a.html. [8] ASME Boiler and Pressure Vessel Code, Rules for Inservice Inspection of Nuclear Power Plants,Section XI, Appendix

A [9] ASME Boiler and Pressure Vessel Code, Rules for Inservice Inspection of Nuclear Power Plants,Section XI, Appendix Failure [10] ASME NB-2331, 1998 ASME Boiler and Pressure Vessel Code, Rules for Construction of Nuclear Power Plants, Division 1, Subsection NB, Class 1 Components [11]

Proposal for the Maximum K Ic for use in ASME Code Flaw Vessel and Piping Conference, PVP2011-57173. [12] WallK Ic Fracture Mechanics, 19(6), pp. 1085-1093, 1984. [13] K Ic Fracture Mechanics, 22, pp. 149-163, 1985.Table 4. Equations to estimate central tendency (mean, median) fracture toughness values for ferritic steels, and their variation with temperature.

Toughness Equations Eq. # Inputs and Constants Limits Cleavage crack initiation, lower shelf and transition (4) T o measured as per ASTM E1921 US , T US per eq. (11)

Cleavage crack arrest (10) T o, as above

--- (7a) Ductile crack initiation on the upper shelf (11) T o, as above n, from fit to J-R curve of the form = 0.3 C 1 = 1033 MPa C 2 = 0.00698/K C 3 = 0.000415/K

= 0.0004/sec T > T US , T US per eq. (11)

(8) (A1) Notes: By performing these calculations beginning at the top and moving to the bottom of this table, all variables will be defined in the order that they are needed in later equations.

7 Table 5. Equations to estimate bounding fracture toughness values for ferritic steels, and their variation with temperature.

Toughness Equations Eq. # Inputs and Constants Limits Cleavage crack initiation, lower shelf and transition

, where (5) T o measured as per ASTM E1921 p is a selected bounding value (e.g., 0.01, 0.025, 0.05) US , T US per eq. (11)

Cleavage crack arrest (10) T o as above --- (7a) (7b) p M p 0.01 2.33 0.025 1.96 0.05 1.64 Ductile crack initiation on the upper shelf (11) T o, as above n, from fit to J

-R curve of the form = 0.3 C 1 = 1033 MPa C 2 = 0.00698/K C 3 = 0.000415/K

= 0.0004/sec T > T US , T US per eq. (11)

(8) where (9) (8-9) M p as above (A1) --- Notes: By performing these calculations beginning at the top and moving to the bottom of this table, all variables will be defined in the order that they are needed in later equations.

[14] Toughness Transition Curve Shape for Reactor Vessel Int. J. Pres. Ves. & Piping, 55, pp. 61-79, 1993 [15] Establishing a Physically Based, Predictive Model for Fracture Toughness Transition Behavior of Ferritic Steels (MRP-53): Materials Reliability Program (MRP), EPRI, Palo Alto, CA:2001. 1003077. [16] Validation Of The Master Curve For Irradiated And of the 1998 ASME/JSME Pressure Vessel and Piping Symposium, July 26-30, 1998, San Diego, California, USA. [17] IAEA TRS-curve approach to reactor pressure vessel integrity in nuclear Austria, 2005. [18] Final Workshop, 15-16 January 2014, Dresden Germany, http://projects.tecnatom.es/webaccess/LONGLIFE/. [19] ASTM E1921- Reference Temperature, T o, for Ferritic Steels in the Transition Range. [20] Master Curve Based Correlation between Static Initiation Toughness K Ic and Crack Arrest Toughness K Ias of the 24 th MPA-Seminar, Stuttgart, October 8 and 9, 1998. [21] Kirk, M. T., Natishan, Physics-Based Model for the Crack Arrest Toughness of Fatigue and Fracture Mechanics, 33 rd Volume, ASTM STP-1417, W. G. Reuter, and R. S. Piascik, 8 Eds., American Society for Testing and Materials, West Conshohocken, PA, 2002. [22] Upper-Shelf Fracture Toughness Master Curve for Ferritic International Journal of Pressure Vessels and Piping , 83 (2006) 571583 [23] Materials Reliability Program: Implementation Strategy for Master Curve Reference Temperature, T o (MRP-101), EPRI, Palo Alto, CA, and U.S. Department of Energy, Washington, DC: 2004. 1009543

. [24] Relationship between the Transition and Upper Shelf Fatigue Fract Engng Mater Struct 29, 672684 (2006). [25]

http://www.nrc.gov/reading-rm/doc-collections/cfr/part050/part050-0061.html. [26] Application of Master Curve Fracture Toughness Methodology for Ferritic Steels (PWRMRP-01): PWR Materials Reliability Project (PWRMRP); EPRI, Palo Alto, CA: 1999. TR-108930. [27] Seminar, University of Stuttgart, Germany, October 1997. [28] B. Houssin, R. Langer, D. Lidbury, T. Planman and K. Wallin, "Unified reference fracture toughness design curves for RPV steels - Final Report", CEC-DG XI Contract B7-5200/97/000809/MAR/C2, EE/S.01.0163.Rev. B 200

1. [29] U.S. Nuclear Regulatory Commission Regulatory Guide 1.99 Radiation Embrittlement of Reactor Vessel Materials , May1988, http://pbadupws.nrc.gov/docs/ML0037/ML003740284.pdf [30] Nuclear Regulatory Commission Reactor Vessel Integrity Database, Version 2.1.1, July 6, 2000. [31] Charpy Embrittlement CorrelationsStatus of Combined Mechanistic and Statistical Bases for U.S. RPV Steels (MRP-45): PWR Materials Reliability Program (PWRMRP),

EPRI, Palo Alto, CA: 2001. 1000705. [32] an Embrittlement Trend Curve Using the Reliability Program (MRP-289), EPRI, Palo Alto, CA: 2011. 1020703. [33] -Range Embrittlement Trend Curve for Radiation on Nuclear Materials on June 15, 2011 in Anaheim, CA; STP 1547, Takuya Yamamoto, Guest Editor.,

pp. 132, doi:10.1520/STP103999, ASTM International, West Conshohocken, PA 2012. [34] Kirk, M., Hein, H., Erickson, M., Server, W., and Stevens, Fracture-Toughness based Transition Index Temperatures for use in the ASME Code with the Crack Arrest (K Ia) Curvessure Vessel and Piping Meeting 2014, PVP2014-28311. [35] Proposed Code Case N-830 (BC 09-182), Direct Use of Master Fracture Toughness Curve for Pressure Retaining Materials for Vessels of a Section XI, Division 1, Class.

Figure 1. Illustration of the variables used by the models in Table 1 to describe the fracture toughness of ferritic steels.

9 Figure 2. Comparison of ASME K Ic curve (truncated at the de facto lto data-based models of K Ic / K Jc (pink) and J Ic (blue). The shaded regions depict the 97.5% / 2.5% confidence bounds for the data-based models. Within the overlap of the shaded regions there is competition between cleavage and ductile fracture.

10 Figure 3. Comparison of ASME K Ic and K Ia curves (truncated at the de facto to data-based models of K Ic / K Jc (pink) and K Ia (green). The shaded regions depict the 97.5% / 2.5% confidence bounds for the data-based models.

11 Figure 4. Combination of data from [6] and [28] where measurements of RTNDT and T o are available for the same materials to illustrate the considerable range by which RTNDT can exceed T o.

12 Figure 5. Comparison of ASME K Ic curve (truncated at the de facto data-based models of K Ic / K Jc (pink) and J Ic (blue). The shaded regions depict the 97.5% / 2.5% confidence bounds for the data-based models. The ASME K Ic curve is shown for various temperature differentials between T o and RTNDT.

13 Figure 6. Comparison of ASME K Ia curve (truncated at the de facto -based models of K Ic / K Jc (pink) and K Ia (green). The shaded regions depict the 97.5% / 2.5% confidence bounds for the data-based models. The ASME K Ia curve is shown for various temperature differentials between T o and RTNDT.

14 Figure 7. Values of RT NDT margin reported in RVID2 [30].

Figure 8. Comparison of ASME models (curves) with various margin terms applied to data-based models of K Ic / K Jc (pink), J Ic (blue), and K Ia (green). The ASME models are all truncated at the de facto 97.5% / 2.5% confidence bounds for the data-based models.

15 APPENDIX: RELATIONSHIP BETWEEN J 0.1 AND J Ic Background Nonmandatory Appendix K of the ASME Code, Reactor Vessels with Low Upper Shelf Charpy Impact Energy [A1] adopts an elastic-plastic fracture mechanics assessment method based on the J-R curve. Appendix K does not use J Ic as a parameter characterizing ductile crack initiation, but rather adopts the parameter J0.1, with the subscript denoting that this is the value of J at 0.1 inches (2.5 mm) of stable ductile crack growth.

J Ic is determined at the onset of ductile crack growth [A2]; since the J-R curve is rapidly increasing for most reactor materials at loading levels around

J Ic values of J Ic can exhibit considerable scatter. The ASME Code therefore adopted J0.1 as an engineering measure of ductile crack initiation; it generally exhibits less scatter than does J Ic. J0.1 Model In NUREG/CR-5729 Eason et al. assembled from the literature a considerable collection (over 500 specimens) of J-R curve data, including both unirradiated and unirradiated RPV materials, welds as well as base metal, and also nuclear grade piping materials [A3].

Figure A1 shows all of these data, plotted as a function of temperature, where upper shelf is characterized using both J Ic and J0.1. As previously reported by Kirk et al. [A4], the J Ic characterization demonstrates de facto upper limit on K Ic of 220 MPa-conservative. Conversely, the J0.1 characterization of upper shelf shows that for de facto upper limit on K Ic Figure A1. Data from [ref] plotted in terms of both J Ic (top) and J0.1 (bottom). The de facto ASME limit on K Ic to J units, is shown on each graph. The J-R curve data in NUREG/CR-5729 exhibit a clear relationship between the ratio of J0.1/J Ic and the J-R curve exponent n (see Figure A2), as follows:

(A1) This relationship can be used to convert the J Ic-based K Ic limit proposed by Kirk et al. [A4

]: (A 2) to one based on J0.1: (A 3) Figure A3 compares the limit of eqn. (A2) to that o f eqn. (A3), and to the de facto J-R curve exponent

n. For high values of n and low values the RT To reference temperature the ASME limit is appropriate or conservative.

However, in view of the tendency for irradiation damage to increase RT To and also reduce n, it seems that the continued use of a 220 K I be re-examined. Figure A4 provides an illustrative example of the elevation of J0.1 above J Ic for two different values of n. Figure A2. Data from [ref] showing a clear relationship between the ratio J 0.1/J Ic (top) and the J-R curve exponent

n.

16 Figure A3. Illustration of the effect of J-R curve exponent n on the K Ic upper shelf limit supported by upper shelf toughness as characterized by J 0.1. The K Ic limit based on J Ic was proposed in

[A4]. Figure A4. Illustration of the effect of J-R curve exponent n (n=0.8 top, n=0.2 bottom) on the magnitude of upper shelf toughness characterized by J 0.1 for a material having RT To = 102 C.

References

[A1] ASME Boiler and Pressure Vessel Code, Rules for Inservice Inspection of Nuclear Power Plants,Section XI, Assessment of Reactor Vessels with low Upper Shelf Charpy Impact Energy [A2] ASTM E1820, Standard Test Method for Measurement of Fracture Toughness, ASTM International, West Conshohocken, Pennsylvania, USA. [A3] Eason, E.D., WrightModeling of Pressure Vessel and Piping J-NUREG/CR-5729, United States Nuclear Regulatory Commission, 1991. [A4] Proposal for the Maximum K Ic for use in ASME Code Flaw Pressure Vessel and Piping Conference, PVP2011-57173.