NUCLEAR REACTOR PRESSURE

VESSEL INTEGRITY INSURANCE

BY CRACK ARREST ABILITY
EVALUATION USING LOADS FROM
INSTRUMENTED CVN TESTS.
NUCLEAR REACTOR PRESSURE
VESSEL INTEGRITY INSURANCE
BY CRACK ARRESTABILITY
EVALUATION USING LOADS FROM
INSTRUMENTED CVN TESTS.

A. Fabry

SCK»CEN, Boeretang 200,

B-2400 Mol, Belgium

October 15,1997

BLG-750
 NUCLEAR REACTOR PRESSURE VESSEL INTEGRITY INSURANCE
BY CRACK ARRESTABILITY EVALUATION
USING LOADS FROM INSTRUMENTED CVN TESTS.

Albert Fabry

SYNOPSIS

This paper highlights some R&D accomplishments related to a series of contractual reports to
Tractebel Energy Engineering, prepared in support of giant Ufe management (PLIM) activities for
the Doel-I and Doel-II nuclear power facilities.

It is contended that the early concepts of fracture analysis diagram and of crack arrest temperature,
independently pioneered by Pellini and by Robertson in the fifties, can advantageously be applied
to insure appropriately conservative safety margins for ageing reactor pressure vessels- this while
keeping abreast with state-of-the-art fracture mechanics technology. This can be done by advanced
analysis of the crack arrestability features embedded in the response of the Charpy V-notch impact
test routinely used for commercial vessel surveillance.
 TABLE OF CONTENTS.

NUCLEAR REACTOR PRESSURE VESSEL INTEGRITY INSURANCE

BY CRACK ARRESTABILITY EVALUATION
USING LOADS FROM INSTRUMENTED CVN TESTS.

ABSTRACT

FOREWORD.

1. OVERVIEW OF BASIC CONCEPTS AND CHALLENGES.

2. FORMULATION OF "MASTER CURVE "FOR CVN ARREST LOAD CORRELATIONS.

2.1 Uncertainties of Drop Weight Test.

2.2 Uncertainties of CVN Arrest Load Determinations.
2.3 "Master Curve" Formulation.
2.4 "Master Curve" Validation.
2.5 "Master Curve" Relationship to Crack Arrest Fracture Toughness.

3. LINK OF CVN LOAD-TEMPERATURE DIAGRAM AND FRACTURE APPEARANCE.

3.1 Formulation.
3.2 Systematics and Limitations.

4. PERFORMANCE OF "MASTER CURVE'TOR CVN ARREST LOAD CORRELATIONS.

4.1 Application to a Commercial PWR Vessel.

4.2 Conservatism of Current Regulatory Approach.

5. MERGING OF CRACK ARREST AND CRACK INITIATION APPROACHES.

6. CONCLUSIONS.

7. ACKNOWLEDGMENTS.

8. REFERENCES.

FIGURES 1 to 18.
 Figure Captions.

Figure 1. Influence of Stress and Thickness on Crack Arrest

Figure 2. Uncertainties of "Pellini" Drop Weight Test

Figure 3. Statistical Distribution of CVN Arrest Load

3.a. ASTM Round Robin (Linde 80 Weld)

3.b. Three Parameter Weibull with Temperature-Dependent Threshold vs. Two-

Parameter Weibull

Figure 4. "Master Curve" Formulation for CVN Arrest Load

Figure 5. NDT Corresponds to Temperature at Median CVN Arrest Load of 3 kN

Figure 6. Validation of "Master Curve" for CVN Arrest Load Correlations

Figure 7. CVN Arrest Load Tracks Well Drop Weight NDT

Figure 8. Indexing Crack Arrest Fracture Toughness to Arrest Load of Instrumented CVN Impact Test

Figure 9. Link of CVN Load-Temperature Diagram and Shear Fracture Appearance

Figure 10. SFA Data Base for Unirradiated Plate HSST-02

Figure 11. Weibull Evaluation of CVN Fracture Appearance

Figure 12. Scatter Trends of CVN Fracture Appearance Change Drastically at NDT

Figure 13. In Service CVN-FATT Shift Is Upper Bound for NDT Shift

Figure 14. Evaluation of Margin Against Brittle Fracture for Circumferential Weld of Commercial PWR
Vessel Operated Well Beyond Plant Design Life

Figure 15. Irradiation/Anneal/Re-Irradiation (IAR) Path Depends on Indexing Concept for Fracture

Toughness

Figure 16. KIo - Shift for A302-B Reference Plate

Figure 17. CVN Arrest Loads for A302-B Reference Plate

Figure 18. Strain Rate Effect on RPV Steel Fracture Toughness

 NUCLEAR REACTOR PRESSURE VESSEL INTEGRITY INSURANCE BY CRACK
ARREST ABILITY EVALUATION USING LOADS FROM INSTRUMENTED CVN TESTS.

Albert Fabry
SCK-CEN, Boeretang 200, B-2400 Mol, Belgium

October 15, 1997

ABSTRACT: This work, undertaken in the context of nuclear reactor pressure vessel (RPV) surveillance, aims at re-
visiting the crack arrest approach to structural integrity insurance. This approach, mandatorily used under normal plant
operation conditions, can also offer an attractive alternative to the crack initiation philosophy promoted for accident
analysis. To this end, an appropriately conservative, cost effective and robust methodology is forwarded and demonstrated;
it makes use of the crack arrest information contained in the instrumented Charpy V-notch impact test and/or in the shear
fracture appearance of broken samples. Particular attention is paid at the appraisal of uncertainties and the related safety
margins. The resulting capability is placed in perspective with the state-of-the-art crack initiation methodology based on the
slow bend testing of precracked specimens, presently under standardization world-wide. The investigation leads to
highlight three conceptual weaknesses of current engineering and regulatory practices. Improved crack arrestability
evaluation emerges as an optimal approach to insure safe PWR operation up to design end-of-life and beyond.

KEYWORDS: Reactor pressure vessel steel, safety, regulation, surveillance, fracture toughness, crack arrest.

FOREWORD. them in capsules near the vessel inner wall. Due, in

particular, to space limitations within these
The strategical importance of permanently insuring capsules, the specimens cannot simulate properly
the integrity of nuclear reactor pressure vessels the stress-strain field conditions of the vessel.
(RPVs) has always been acknowledged. The Correlation methods are thus used to convert their
emphasis has increased as reactors age, because the response into vessel toughness. Intrinsically, such
interest has raised into the feasibility of safe and correlations entail significant uncertainties which
economically balanced plant operation beyond must be conservatively accounted for by
design life, even if at the cost of license renewal in appropriate margins. Fracture toughness can refer
some countries, of vessel anneal in others,... The to the material resistance to crack initiation -K]c,
central concern is prevention of the propagation of KId-, or to crack propagation/arrest -KIa-, to static
cleavage cracks through the vessel, possibly or to dynamic loading. Charpy V-notch impact
leading to its rupture; this could be triggered by a (CVN) surveillance specimens are sensitive to a
host of conceivable accidental pressure- mix of initiation and propagation effects,
temperature transients, if the steel fracture complicated by shear lip formation and by a three-
toughness was allowed to degrade exaggerately. dimensional evolution of the stress-strain field as
Another, less publicized concern entails the cycle deformation proceeds. For fracture mechanics RPV
operation cost, whose increase upon service evaluation, the static toughness KIc and the dynamic
depends also on vessel embrittlement margins toughness KIa are empirically linked to this
through start-up and shutdown pressure- intricate, dynamic CVN response, without any
temperature limitations. attempt to separate its components nor to address
the service-induced modification of strain rate
Pressure vessel surveillance aims at anticipating the effects. Yet, it has long been known that such
behavior of the relevant materials, by exposing improvements are possible if use is made of the
load-time signals recorded by strain gages on the long crack becomes impossible for a given applied
Charpy hammer. The implications to RPV safety stress; in a fracture mechanics perspective, this can
margins are considerable for a number of PWR also be seen as the temperature above which the
plants. crack arrest toughness exceeds a given magnitude,
100-200 MPaVm for stresses of relevant
1. OVERVIEW OF BASIC CONCEPTS AND engineering concern. Above the NDT, the stress
CHALLENGES. associated to the CAT, i.e. the CAT curve,
increases sharply with increasing temperature and
The concept of temperature of Nil-Ductility reaches the yield stress at the so-called Fracture-
Transition, abbreviated here as Nil-Ductility Transition Elastic (FTE) temperature.
Temperature (NDT). has been forwarded by
Pellini in 1953 during the development of the For mild steel structures of thickness about 50 cm,
explosion bulge test at the U.S. Naval Research FTE= NDT+33°C (At higher stress levels, the
Laboratories. By definition, NDT is the largest structure would fail by general plastic collapse).
temperature at which a crack, initiated by explosion The FTE can thus be seen as a safety criterion
loading of a large scale, structure-representative where the margins are provided by temperature
test plate containing a small flaw, will propagate alone, irrespective of the size of postulated flaws.
under purely elastic conditions (i.e., plate remains Furthermore, neither CAT nor FTE are affected by
flat at fracture). Investigation of service failures test specimen orientation nor by the specimen
showed that these could be prevented by requesting width along the crack path, but both depend on
to operate the structure at temperatures exceeding thickness, and to less an extent on the yield stress
NDT by a fixed, minimum amount. This approach (and the applied stress in the CAT case). This is
has been quantified by the well known "Fracture quantified by the full curves of Figure 1,
Analysis Diagram" (FAD) (Pellini, 1963). The constructed using a recent literature correlation
diagram also refers to the concept of Crack Arrest based on the compilation of an extensive
Temperature (CA T). a concept specifically experimental data base (Wiesner, 1996) [for
demonstrated by the development and application comparison, the dashed lines refer to the CAT and
of the Robertson large-scale crack arrest test to the small flaw configurations as evaluated in the
(Robertson, 1953). The CAT is the temperature original Pellini FAD diagram (Pellini, 1963)].
above which unstable propagation of an "infinitely"

Applied Stress /Yield Stress

1.5
Small Flaw ,--CAT For Thickness^
All Thicknesses,' ,' Small /Large

NDT

-20 20 40 60 100
T-NDT (°C)
Figure 1. Influence of Stress and Thickness on Crack Arrest.
It is seen that the difference FTE-NDT increases by Code indexes the fracture toughness curves (KIa
only 20°C for an increase of the yield stress from and KId, but also static Klc) to T-RTNDT; here, the
400 to 1000 MPa - a range essentially Reference Temperature for Nil Ductility
encompassing the spectrum of possibilities for Transition RTNDT is defined as the lowest of: 1)
PWR vessels- and furthermore, this 20° increase NDT, and of: 2) TCV-33°C, where T c v is the
does not depend on thickness. For a 200 mm thick temperature at which each of three CVN impact
PWR vessel, on average: FTE= NDT+85°C, with a specimens exhibits at least 68J absorbed energy
maximum variation of ± 10°C for the considered and 0.89mm lateral expansion. This modification
yield strength range. This decreases to NDT+ 80°C was done with the intent to expediently provide
(± 10°C) for a thinner vessel of 150 mm, and the some insurance against possible tearing instability
thickness effect is thus small as well as easily for materials displaying low CVN upper shelf
accountable for. This effect is not linear at smaller energies. The mismatch is double:
thickness: for a hypothetized 10mm thick
specimen, one would have: FTE= NDT+29°C, 1) One ignores recent, albeit well
calling for a penalty of +56°C if transposing to the established technical and regulatory
200 mm thick vessel. progress (see for instance USNRC
Regulatory Guide 1.161, June 1995):
The Drop Weight ("Pellini") Test was developed rather than adding some margin to brittle
(Puzak, 1952) and eventually standardized (ASTM initiation and arrest curves, any potential
E208, 1963) in order to allow NDT determinations tearing instability concern can and should
with relatively small-scale specimens. For this be separately evaluated on basis of the
three-point bending impact test, the specimen is a ductile initiation fracture toughness and of
rectangular coupon (of minimum thickness 16 mm J-resistance curves;
x minimum width 51mm x 127 mm); a brittle weld
bead, containing a notch, and deposited at the 2) Static toughness is indexed to dynamic
center of one face, serves as crack starter; the quantities; this neglects the known fact
amount of deflection is controlled by a stop so as that the strain rate effect on toughness
not to exceed a few mm.; NDT is the maximum depends on the yield stress.
temperature at which the specimen breaks- an event
defined to occur whenever the brittle crack extends What actually seems to transpire here is some
to just reach the edges of one (or both) of the amount of confusion between the crack initiation
tensile surfaces on either side of the weld bead. and crack propagation approaches to safety.
Basically, the drop weight test measures the Indeed, for normal operation of a PWR reactor
resistance against continued crack propagation, pressure vessel, the ASME Code Section III
which is equivalent to crack arrestability. Thus, Appendix G procedure defines pressure-
NDT has initially been used to index the arrest temperature limitations with reference to the K,R
fracture toughness K]a of ferritic steels: more curve, which is essentially equivalent to the lower
specifically, the lower bound Kfa curve has been bound KIa curve; the philosophy for brittle fracture
considered to be an unique function of the prevention is crack arrest and the onset of upper
"reduced" temperature T-NDT. This is an shelf temperature is defined in relation to the KIa
approximation: fracture mechanics interpretations curve. If the CVN upper shelf energy falls below
of the drop weight test suggest that NDT better the 68J. regulatory limit, a tearing instability
correlates to K[a normalized by the dynamic yield analysis is nowadays required and therefore, the
stress (e.g. Irwin, 1967, Shoemaker, 1971, rationale having led to replace NDT by RTNDT
Sumpter, 1986). becomes obsolete. On another hand, under accident
conditions, the current nuclear philosophy for
Subsequently, significant conceptual mismatch brittle fracture prevention as implemented for
has been introduced by the fact that the ASME instance by the July 1991 U.S. Code of Federal
 Regulation, Title 10, Part 50, Article 61 - the Not until the early sixties however could an
pressurized thermal shock (PTS) rule - is primarily experimental separation of these energy
crack initiation; however, indexing the K[c curve to contributions be attempted: this was then made
RTNDT , as done in this context, is technically possible by the successful development of strain
incorrect. It is generally believed that the crack gage instrumented Charpy tups, which allow the
arrest approach is more conservative than the on-line recording of load-time (deflection) signals
initiation one. By contrast, it has been found during the entire impact event. In particular, the
(Fabry, 1997.a) that, for irradiated reactor pressure load at arrest of a brittle crack can be detected in
vessels, the KIc curve tends towards the KIa curve as function of the test temperature, and the post-arrest
a result of service-induced steel strengthening: the energy (associated to shear lip formation) can be
two philosophies thus tend to become equivalent. derived as well. This is used to define a
This important finding, substantiated further in § 5, characteristic temperature at which a reference
is consistent with earlier studies entailing KId versus level of the considered parameter is reached and to
Klc patterns for unirradiated steels spanning a large look for correlations of this temperature with either
range of strengths (Barsom, 1970). The existence CAT (Fearnehough, 1973, Hagedorn, 1983) or
of this effect- not acknowledged by current nuclear NDT (Berger, 1979, Ahlf, 1986, Schmitt, 1990,
regulation- actually means that the arrest fracture Pachur, 1994). Good to excellent correlations have
prevention approach is ultimately not penalizing to been reported; most importantly, while arrest-
the nuclear industry, yet provides the most secure related parameters were always found, as expected,
safety margins, given that crack initiation can never to increase linearly with CAT or NDT, this was not
be guaranteed in an absolute sense (e.g., local always the case for the corresponding 41J.
brittle zones, undetected weld flaws, residual temperatures: larger increases are observed in a
stresses,...). number of cases; the reasons for such "outlier"
behavior have been quantified in detail (Fabry,
1996.a). More recently, excellent correlations
Aside of the above NDT/RTNDT mismatch in
between the CVN arrest load and the crack arrest
assessing beginning-of-life (baseline) vessel
fracture toughness KIa have also been documented
condition, another, most significant source of
(Wallin, 1995, Nanstad, 1996, Iskander, 1997). The
uncertainty and of unwarranted conservatism stems
evidence supporting the feasibility to use CVN
from the current practice to assume that service-
arrest loads in the frame of the crack arrest fracture
induced shifts of fracture toughness curves to
prevention philosophy is altogether compelling, but
higher temperatures are uniquely proportional to
the many disparate details/ resp. the sometimes
the upward shift of the temperature at which the
apparent divergences of published correlations
CVN impact energy reaches an arbitrary level,
need to be unified/ resp. clarified. This can be and
chosen for instance as 41 J. in the USNRC
is done herein by: 1) Addressing the relevant
regulation. The eventual biases stem from the fact
physical and statistical aspects of the CVN test; 2)
that the CVN impact energies, not only are of a
Considering the relationship between CVN load
dynamic nature unsuitable to static toughness
signal and shear fracture appearance (SFA), along a
indexing (see above), but still more importantly,
previously outlined formulation (Fabry, 1996.b).
the biases are inherent to the fact that these
This second line of inquiry was encouraged by
energies are a mix of crack front formation and
extensive literature evidence regarding SFA
initiation energy, crack propagation energy, plastic
correlations to crack arrest [see for example the
deformation energy and shear lip formation energy.
references mentionned at the beginning of § 3],
The relevance of CVN energy partitioning has
been qualitatively grasped ever since unexpected
structural engineering failures have been clarified In summary, the outlined CAT/FTE approach is
by the discovery of the ductile-brittle transition attractive in the nuclear reactor pressure vessel
temperature (DBTT) phenomenon in the fourties.
safety context for the following, inter-related complete redistribution of the applied stress among
reasons: the unbroken, plastically reinforced ligaments left
in the wake of the fast running crack (Cottrell,
1) The approach allows to establish the 1995). In any case, the CAT/FTE/NDT correlation
most secure, yet not necessarily penalizing approach itself would of course remain applicable;
safety margins for vessel operation under FTE, primarily a temperature-based, rather than a
the life management (or life extension) stress-based criterion, is probably also the most
conditions prevailing for all plants once in restrictive, safety-wise (Smith, 1997).
full-fledged service regime; furthermore,
the approach is often less unduly
conservative than the crack initiation 2. FORMULATION OF "MASTER CURVE"
approach as applied to-day for accident FOR CVN ARREST LOAD CORRELATIONS.
analysis
2.1 Uncertainties of Drop Weight Test.
2) The correlation CAT/FTE to NDT is
sufficiently well established (Wiesner, In order to judge the performance of the semi-
1996) and is sufficiently insensitive to empirical formulation proposed below, it is
postulated flaw size, to stress, to necessary to first develop some estimate of the
constraint, to service strengthening accuracy with which the quantity to be predicted is
effects,... so as to make one able to itself usually defined. In this section, this quantity
guarantee confidence in predicting crack is chozen as the drop weight nil-ductility
arrest behavior of actual reactor pressure temperature, but it must be kept in mind that the
vessels from the knowledge of NDT only ultimate objective entails the previously outlined
CAT/FTE/NDT approach. Thus, at this stage, the
3) Furthermore, as elaborated hereafter, issue is to decide whether or not unbiased NDT
NDT can be reliably correlated to the estimates can be obtained from CVN arrest loads
shear fracture appearance and to the crack within confidence intervals comparable to the ones
arrest load of the instrumented CVN of the "Pellini" test. Only subsequently will one
impact test, as routinely used in address the final stage of accuracy for structural
surveillance programs; this confers to the integrity application (§ 4.1).
approach the needed practicability for
straightforward use by the engineering The drop weight testing conditions may affect NDT
community- while not requiring any extra determinations for nuclear pressure vessel steels,
testing leading in some cases to unconservative
evaluations (Holt, 1986). Of primary importance
4) Finally, as illustrated in § 4.1, the safety are the crack starter bead welding and specimen
margins predicted by this approach are cooling rate conditions; in particular, two-pass bead
consistent with the ones resulting from the application (ASTM Method E208-81) tends to
application of state-of-the-art fracture cause high toughness of the heat-affected zone
mechanics. (HAZ), as well as some quenching and tempering
of the specimen; excessive HAZ hardness acts as a
An important comment is called for by the fourth barrier against crack propagation, i.e. lowers the
item. That fracture mechanics applies to crack observed NDT with respect to its unperturbed
arrest at the NDT is clear, but that it may apply also value. Additional uncertainties are associated to the
to crack arrest at the FTE can another hand be material variability as well as to the testing
questionned on fundamental grounds; in particular, temperature steps (usually 5 or 10 °C). It is difficult
this should not be so if the FTE actually represents to assess the exact magnitude of all these
the temperature at which arrest stems from uncertainties when establishing CVN to NDT
correlations, because these depend on the
Fallura Probability
considered experiment and steel. Many of the input
data for the present study have been collected from
literature. An across-the-board estimate of the
overall NDT uncertainty has been attempted by
examining a large set of data (71 tests) compiled by
EPRI for unirradiated plate HSST-02 (Server,
1978). This is shown by Figure 2. The median
NDT value is -28 °C, with a 2a confidence interval
of ± 15 °C. Two separate determinations, one at
ORNL (Childress, 1971) and one by the Belgian
program, give respectively -18 and -11 °C. The
value stemming from the correlation presented
below is equal to -11 °C. The correlation is
actually consistent with a 2a NDT uncertainty of± 0.2

15 °C, and this estimate is accepted. It is

nevertheless recognized that state-of-the-art drop
weight tests are probably more accurate, with a 2a
-so -40 -30 -20 -10
confidence interval somewhere in the range ±8-15 T«st Temperature i°C)

Figure 2. Uncertainties of Drop Weight Test.

2.2 Uncertainties of CVN Arrest Load dependent threshold fa (T) and shape parameter
Determinations. m=2. The scale parameter, or median, is denoted
Fa(T). Considering a given temperature T, i.e.
The statistical distribution of the CVN arrest load dropping the T dependency, one can write the
Fa has been found to be representable by a three-
cumulative distribution function as
parameter Weibull function with temperarure-

V (F.) = 1 - exp(-[(Fa -fa)l{Fa- fa (1)

The mean and standard deviations are respectively equal to

(2)

cr = 0.463F, (3)

Figure 3.a compares this distribution to the best 1997); all specimens have been broken at ORNL
two-parameter Weibull fit (m= 4). The 61 tests on a same machine; the exercise entailed the
used for this evaluation are relative to the low comparison of ASTM and DIN tups, and statistical
upper shelf Linde-80 weld selected for the ASTM evaluation showed that no tup influence could be
round robin on CVN reconstitution (Onizawa, detected for the arrest load. The existence of a
 Ln Ln (V Wl) Teat Temperature: NOT • 41 °C Probabinty Denilty Function
0.C

/ \ Tmn-PirwMttr Wtlbuli
0.4 -

t
j
0.3
t i ^00^*-i Three-Parameter Welbull

m-4 0.2

0.1
IDT A HDT.41°C N^

15 20
» 12 14 M
ArrHt Low) F, (kN)
CVN Arrest Load (kN)

Figure 3. Statistical Distribution of CVN Arrest Load

a) ASTM Round Robin (Linde 80 Weld) b) Three-Parameter Weibull with Temperature-
Dependent Threshold vs. Two-Parameter Weibull

temperature-dependent arrest load threshold is to correlation to NDT. Many experiments that would
be expected on straigthforward physical grounds: appear inconsistent with one another for the [m=4,
indeed, there is always a temperature above which f=0] distribution are recounciled with the [m=2,
brittle crack propagation cannot take place, f(T)] statistics.
irrespective of the applied stress (upper bound of
upper shelf onset temperature) while on another Two other important considerations entail
hand, at decreasing temperature, the arrest load precision and validity limits. This can be briefly
tends towards zero. Actually, and as independently discussed at the light of Figure 4, where arrest
confirmed in a subsequent section, it is reasonable loads for unirradiated plate HSST-03 have been
to define NDT as the temperature below which the analyzed using the "master curve" formulation
arrest load threshold is zero and above which it further described below.
departs from zero. This is not without reminding
one procedure adopted for direct correlation of
CVN arrest loads to the thickness-corrected CAT
Arraal Load (kN)
(Hagedorn, 1983); yet the referred correlation - 20
0 MEA (L-T) MEDIAN
CAT= temperature where Fa drops to zero- seems • MOL (L-T>
ONSET OF

to conflict with a previous study suggesting NDT © MOLIT-O

to be equal to the temperature at which the average

CVN arrest load is 5 kN (Berger, 1979). It could
be concluded, and has sometimes been considered
(Ahlf, 1986), that the Fa level corresponding to
NDT depends on the steel. These apparent
disparities are resolved by the present results.
Indeed, a reliable definition of the statistical
distribution of Fa in function of temperature is 60

essential before any generalized correlation be

seriously attempted. As Figure 3.b illustrates, the
present Weibull distribution is significantly wider Figure 4. "Master Curve" Formulation for CVN
at NDT than for instance the m=4 two-parameter
Crack Arrest Load
Weibull distribution; this in turn affects the

definition of the best median ./^(T^) relationship Incidentally, it can be seen that no orientation
effect (L-T versus T-L) is detectable, and that tests
and therefore of the median Fn level for by two independent laboratories are in good
agreement; these two observations are general and the median temperature above which Fu< Fm, where
have been confirmed by examination of a large Fm is the maximum load: significant ductile stable
data bank relative to many pressure vessel steels crack growth then precedes the brittle initiation.
tested at a significant number of laboratories. However, extensive evaluation has shown that a TN
Figure 4 indicates that the optimal testing rejection criterion would be unnecessarily severe.
temperature range for sufficiently precise Along the same lines, it has been found that, in
application of the approach extends approximately case of multiple brittle initiation and arrest, the first
from NDT-15°C to NDT+60°C. At lower arrest point must be used. All in all, the presently
temperatures, the master curve is too flat and, as accumulated experience indicates that, in order to
already suggested by Figure 3, its confidence achieve an unbiased, confident application of the
interval is too large. But at higher temperatures, proposed correlation, a minimum of four tests
validity questions arise, requiring some amount of within the temperature range NDT-15°C to
data rejection. More exactly, rejection is called for NDT+60°C are needed. Such requirement is
whenever the test temperature exceeds the median generally fulfilled by routine surveillance testing,
onset temperature T o of the CVN upper shelf: this and if not the case, CVN reconstitution (van Walle,
characteristic temperature is defined as the median 1996) can often provide the necessary remedy.
temperature at which the brittle crack initiation
load Fu equals the arrest load Fa and the median 2.3 "Master Curve" Formulation.
shear fracture appearance thus reaches 100%. The
procedure to determine T o is illustrated in a The CVN arrest load "master curve" formulation
subsequent section. From a "physical" viewpoint, illustrated by Figure 4 is specified as an increasing
the need for such rejection criterion stems from the exponential function of the "reduced" temperature
fact that at or above T o , plastic deformation of the T-NDT, with a modest slope increase at the NDT.
specimen becomes excessive: the near- plane strain The associated statistical distribution is the three-
conditions prevailing at NDT, and which constitute parameter Weibull function with temperature-
one crucial facet of the rationale under the CVN- dependent threshold.This formulation is expressed
drop weight correlation, are violated. Strictly by the following series of simple equations:
speaking, such "violation" may already occur at TN,

Fa(T) = 3.00exp(ar*) T* = T- NDT (4)

a = 0.026 for T*>0

a = 0.020 for T* < 0

(Load in kN, Temperature in °C), with the 2cr confidence interval given by

Fa ± A = [(Fa - fa M l ± 0.762)] + fa (5)

where the temperature-dependent threshold is

fa (T*) = 2.05[exp(0.026?*) - l] for T*>0 (6.a)

= 0 forT*<0 (6.b)
For this Fa master curve formulation, the
median nil-ductility temperature NDT temperatures up to at least the ductility temperature
corresponds to a median CVN arrest load TN (§ 2.2). A change of the slope a of the
indexing level of 3.00 kN. exponential behavior was intuitively expected to
occur at TN, a temperature at which a fully formed
A qualitative physical argument has been ductile crack front begins to propagate; the change
previously given for the existence of the was found instead to take place already at NDT,
temperature-dependent threshold. That this which may suggest concurrence with the median
threshold would depart from zero exactly at the temperature for onset of local ductile initiation.
NDT and increase according to equation (6) was an
initial working assumption, but it has been 2.4 "Master Curve " Validation.
subsequently confirmed by the evaluation of a
substantial arrest load data bank, and by the The formulation of § 2.3, applied with due
statistical features of fracture appearance data, as attention to the precision requirements and validity
reviewed in a forthcoming section. The selection of limits outlined in § 2.2, does adequately meet its
the exponential increase of the threshold above purpose in providing a one-to-one correspondance
NDT, which implicates a corresponding median between the drop weight NDT and the temperature
exponential behavior (but not necessarily of the at a median CVN crack arrest load of 3 kN. This is
same slope), was inspired by the ASME demonstrated by Figure 5 and Table 1. Basically,
representation of the lower bound Kla curve. the "CVN- predicted" NDT values are consistent
Whenever linear elastic fracture mechanics is with the drop weight values to within the 2a
adequate to describe crack arrest, the assumption of confidence interval of ± 15°C estimated in § 2.1. It
proportionality of K, to stress, thus of K,a to arrest is reasonable to conservatively assign a 2a
load, is indeed reasonable insofar as plane strain confidence interval qf± 22°C to the median NDT
conditions are nearly respected: this should be the values projected through the present "master
case for CVN specimens broken at curve" correlation.
Drop Weight NDT PC)
160 W Unirradiated Weld and Base
• Irradiated HSSI Welds 72W & 73W
H Irradiated 22NiMoCr37 Forgings

100

60

-60

-100 -60 0 60 100 160

Median 3 kN CVN Arrest Load Temperature
Figure 5. NDT Corresponds to Temperature at Median CVN Arrest Load of 3 kN.
 10

TABLE 1
Comparison between NDT Determinations by Drop Weight Test
and by CVN Arrest Load Correlation

Steel and Condition NDT: Nil Ductility Temperature (°C)

Drop Weight CVN Correlation

Unirradiated (Baseline")

A302-B Reference Plate -12 to-23 -14

A533-B Plate HSST-02 -11 to-28 -11
A533-B Plate HSST-03 -12 +4
KS-01 22 NiMoCr 37 Forging +5 +5
KS-02 22 NiMoCr 37 Forging 0 -6
A508 C1.2 Forging (PTSE-1) +66 +83

Doel-I,-II Surveillance Forgings -24 -24

Doel-IV A508 C1.3 Surveillance Forging -35 -19
22 NiMoCr 37 Surveillance Forging -20 -3

HSSI Weld 72W -27 -27

HSSI Weld 73W -34 -34
ASTM Round Robin Linde-80 Weld -23 -39

Quad Cities-2 Surveillance Weld -46 -50

Doel-I,-II Surveillance Welds -30 -30 to -45
22 NiMoCr 37 Surveillance Weld -35 -45
Midland Vessel Weld -55 -55

Irradiated

KS-01 22 NiMoCr 37 Forging

2E19cm" 2 +45 +45
7 E19 cm"2 +100 +90
KS-02 22 NiMoCr 37 Forging
2 E19 cm"2 +40 +40
8 E19 cm"2 +105 +105
HSSI Weld 72W 1.5-2 E19 cm"2 +34 +34
HSSI Weld 73 W 1.5-2 E19 cm"2 +38 +38

Figures 6, a to d, offer four selected illustrations reevaluation of the considered experiments on

relative to unirradiated pressure vessel steels: a) Figures 7, a and b: the published index is 3.3 kN
ASTM reference A302-B plate; b) German 22 for the first case (KS-01 forging) and 2.3 kN for
NiMoCr 37 weld and forging; c) two U.S. Linde- the second case (KS-02 forging); all these data and
80 welds; d) Belgian A508 Cl. 3 forging. their scatter are also accounted for by the present
The suggestion (Ahlf, 1986) that the Fa indexing "master curve" formulation, which, despite its
level may depend on the steel is examined by unique 3.0 kN index, allows to predict NDT to
 11

a) Unirradiated A302-B Reference Plate b) 22 NiMoCr 37 Weld and Forging

Arrest Load (kN) Arrest Load (kN)
DROP WEIGHT TESTS
Submerged Arc Weld
BAPL - 23.3 °C Drop
p Weight
eg - 36 %
NRL - 12.2 °C CVN Correlation - 46 °C
16 EPftl - 18.0 °C / / /''
© Forging
CVN CORRELATION Drop Weight - 20 °C
- 13.8 °C CVN Correlation -
10

-60 60 100 -100 -60 60 100

T-NDT (°C) T-NDT ( t )

c) Unirradiated Linde-80 Welds d) Unirradiated Doel IV Forging (A508 Cl. 3)

NDT Nominal: -35 °C
Arrest Load (kN) Arrest Load (kN) NDT Thia Work: - 19 °C
A
Quad Cities-2 Weld (W-E) @ Standard CVN
Drop Weight: - 46.6 °C o Miniature CVN
CVN Correlation: - 49.6 °C (Normalized by
16 ' men ASTM Round Robin Weld Ligament)

Drop Weight- - 23.3 °C

CVN Correlation: - 39.3 °C

-100 60 -60 0 60 100

T-NDT (°C) T-NDT {°C)
Figure 6. Validation of "Master Curve" for CVN Arrest Load Correlations

a) Unirradiated and Irradiated 22NIMoCr37 Forging KS-01 (T-L) b) Unirradiated and Irradiated 22NIMoCr37 Forging KS-02 (T-U
Arrest Load (kN) Arrest Load (kN)
so CVN
20
FLUENCE >t MeV DROP WEIGHT FLI1EMCE >IMeV DROP WEIGHT QJflJ
O NONE 6.0 °C 6.0 0C O NONE 0.0 °C -6.4 °C
0 2 E19 cm" 2 46.0 °C 46.0 °C @ 2 E19 cm'* 40.0 °C 39.8 °C
16 • 7 E19 cm-2 100 °C 90.0 °C I 1 1' • 8 E19 cm-2 106 °C 106 °C

jj
10

0
-200 -160 -100 -60 0 60 100 -160 -100 -60 0 60 100
T-NDT (°C) T-NDT (°C)

Figure 7. CVN Arrest Load Tracks Well Drop Weight NDT

 12

within better than ±5°C, except in one case (- Nanstad, 1996, Iskander, 1997). No major
10°C). Clearly, the differences of interpretation discrepancy should be expected (nor is observed)
among various authors primarily stem from the beween the referred publications, insofar as the
formulation (or lack of it) adopted to represent the input data bases tend to significantly overlap. The
temperature dependency of the arrest load, and in major interest however in re-examining these
particular to assess precision, validity limits and experiments is that, in addition to drop weight tests,
confidence intervals. To this respect, it is finally they also encompass fracture mechanics K[a tests,
relevant to mention here that the parameters from which one can, in particular, derive the
adopted to describe the shape of the median and median temperature at a reference level of 100
threshold curves in equations (4) and (6) account MPaVm, noted TA herein. Figure 8 gathers the
also for a number of experiments in which no results. On the left part, the excellent performance
NDT, CAT,....measurements were performed, but of the present formulation can once more be
which entailed rather extensive CVN arrest load appreciated; note that for weld 73W, the ORNL Fa
and fracture appearance determinations at data have been augmented by tests done at Mol
numerous temperatures within the transition. using specimens reconstituted from the ORNL
broken compact tension remnants; also, the outlier
2.5 "Master Curve" Relationship to Crack point (ORNL) does not meet the T o validity
Arrest Fracture Toughness. criterion. The right part of Figure 8 has been
constructed by simply shifting, by 20°C along the
abscissa, the Fa median, threshold and confidence
The formulation has been applied to some
interval master curves from the left Figure. The
experiments performed at Oak Ridge National
point to emphasize is that this shift is unique for the
Laboratories (ORNL) and analyzed elsewhere with
five experiments.
purposes similar to the present ones (Wallin, 1995,

Arrest Load (kN) Arrest Load (kN)

20 20
HSSI Weld 72W HSSI Weld 72W
HI Baseline £3 Baseline
• Irradiated • Irradiated
HSSI Weld 73W HSSI Weld 73W
© Baseline © Baseline
15 # Irradiated 16 • Irradiated
MIDLAND Weld MIDLAND Weld
0 Baseline 0 Baseline

10 10

100 100

Figure 8. Indexing Crack Arrest Fracture Toughness to Arrest Load of Instrumented CVN Test:
TA (Temperature at Median K,aof lOOmpaVm) = "Pellini"NDT+20°C.
 13

For the present Fa master curve formulation, the issue is re-visited now by further consideration of
temperature TA at which the median fracture the physically-grounded link between SFA and the
toughness Kla is equal to 100 MPaVm instrumented CVN load diagram (Fabry, 1996.b):
corresponds to a CVN arrest load indexing level
of 5 kN (exactly: 5.05 kN); furthermore:
SFA (%) = [l-(Fu-Fa)/{Fm+ k(Fm-Fy)}] xlOO (8)

TA= NDT+20°C (7)

where Fy, Fm, Fu and Fa are respectively the general
yield load, the maximum load, the brittle initiation
with a 2CT uncertainty of ± 12°C. load and the arrest load, while the parameter k has
been introduced to quantify the occurence of local
This compares favourably with a recent correlation ductile initiation
(Wallin, 1995): TFa4= TA-10°C (2a = ± 27°C)
where TFa4 is the temperature at which the median k=0 Initiation at maximum load
CVN arrest load is 4 kN; indeed, this transforms k=l Initiation at general yield load
into TA= NDT+21°C upon consideration of k=0.5 Initiation at load equal to
equation (4) for the master curve. Good (F y +FJ/2
concurrence is also found with the work of the two
other previously mentionned authors (Nanstad, Figure 9 illustrates application of this model to the
1996, Iskander, 1997). unirradiated A533-B reference plate HSST-02 in
3 t r . l t (MP«)
3. LINK OF CVN LOAD-TEMPERATURE
T. • - 78 C as°c
DIAGRAM AND FRACTURE APPEARANCE.

3.1 Formulation.

As already mentionned in § 1, brittle crack

arrestability has, since the early sixties, been shown
by many studies to approximately correlate with
CVN shear fracture appearance (SFA): namely,
CAT is linearly linked to the temperature at which 0 100 200
Test Temperature (°C)
SFA reaches a given level, such as 25% (e.g.,
Nichols, 1965, Cowan, 1967, Fearnehough, 1973,
Shur <*)
Kussmaul, 1977) or 50% (e.g., Hagedorn, 1971,
T--S4OC
Akiyama, 1985, Smedley, 1989, Wiesner, 1993).
These correlations differ somewhat from one
another and their scatter is usually large.
Nevertheless, it has been proposed, in the context
of nuclear power operation, that the 50% Fracture
Appearance Transition Temperature (FATT)
allows to better quantify service embrittlement than N O T - - I t °C
the temperature at which the CVN absorbed energy SfA-U*

reaches "conventional" energy levels such as 41 J., -200 -WO 0 WO 200 300

e t c . : namely, FATT shifts are less biased, less Test Temperature (°C)
scattered and often less conservative than energy Figure 9. Link of CVN Load-Temperature Diagram
shifts (Fabry, 1996.a and b, McElroy, 1996). This and Shear Fracture Appearance.
 14

the L-T orientation. Also indicated are all the pressure vessel steels. It is now possible to
characteristic temperatures for the CVN test: TN, scrutinize further the fundamental features outlined
T o , NDT, FATT, as previously defined, and TI; the in § 2.2. The Weibull evaluation of all data on
"brittleness" temperature, in general equal to the Figure 10 is partially illustrated by Figure 11 and
median temperature at which SFA equals zero. the results are summarized by Figure 12. It is found
Note that k=0.5 for T>TN and k=0 for T<TN. It has that the SFA scatter trends change drastically at the
been easy to establish the SFA confidence intervals NDT, left part of the Figure: the slope parameter m
from equation (8). Indeed, the scatter associated to suddenly increases if the two parameter Weibull
Fy and Fm plays a negligible role (the two- description is adopted. However, the alternate use
parameter Weibull distribution has been found of the three parameter Weibull function, with
applicable, with a slope parameter m always in constant slope (m=2), leads to the departure from
excess of 35): this is because these loads reflect the zero at NDT of the temperature-dependent
material flow properties and are not affected by the threshold, right part of Figure 12. This second
inhomogeneity of the distribution of fracture interpretation is recommended: indeed, it hinges
triggering particles. As the top part of Figure 9 upon the qualitative, yet physically- defendable
shows, the role of Fu is minor for the considered grounds discussed in § 2.2,; this argumentation is
steel. In general terms, it has been found that the clearly vindicated by the present statistical analysis.
scatter of the arrest load governs the SFA
scatter. Therefore, the SFA scatter can be The recommended SFA representation is therefore
modelled from equation (8) if one describes the Fa given by equation (8), complemented, from the
scatter by equations (5) and (6). The bottom part of statistical point of view, by the three-parameter
Figure 9 has been derived in this way. The validity Weibull distribution with constant slope m equal
of this statistical model is confirmed by the
to 2 and with temperature-dependent threshold:
extensive SFA data bank available for this plate
(Server, 1978, Stallmann, 1987), see Figure 10
(this excludes the data of Figure 9, which are the
only ones for which load-time traces could be sfa (T*) [%] = 12 [exp (0.026 T*)-l] (9)
located). Statistical tests do not reveal any 0 < T* < 85°C
orientation effects here, in line with a previous
investigation of plate HSST-03 (Vandermeulen,
1993); this observation is in general true for reactor where, as before, T*= T-NDT (°C).

ShHr <*>

L-T Orientation T - L Orientation

100 I0O
o Y~W " 7 " ^
a LA 7 /ofo /
80

80
iff 80

80
<7o/ /
Jo /
oil
oh 1 I
oral o/
/

40 Li I 40
i l l
<hob I*6
crcajao
20 JsxP/u a// SO
/oj/jp
0 i a-^s
-100 -SO 0 SO -100 -CO 0 SO

Test Temperature (°C) Test Temperature (°C>

Figure 10. SFA Data Base for Unirradiated Plate HSST-02

 15

TEST TEMPERATURE: 4.4 °C TEST TEMPERATURE: 16 °C

Ln < Ln 1/1-PI) Ln ( Ln 1/1-PI)

© L-T
O T-L

-4

m-15 ,'"m-2
_i i i i i i i i
-6 -6
1 10 wo wo
Shear Fracture Appearance (%) Shear Fracture Appearance {%)

TEST TEMPERATURE: 29 °C TEST TEMPERATURE: 68 °C

Ln ( Ln Vi-Pf ) Ln ( Ln 1/1-Pf)

0 L-T @ L-T
O T-L o T-L
<

••'1
-'•' m-2 /
Threshold: 79 * ^ J
'2'' m-4,
m- 2 , m- 16/
-6
1 10 10
Shear Fracture Appearance (%) Shear Fracture Appearance (%)

Figure 11. Weibull Evaluation of CVN Fracture Appearance

Welbull Slope m SFA (*> Shear (%)

|Q m| 0 MEDIAN
• THRESHOLD

10 100 CONFIDENCE INTERVAL

© 80 | Welbull SFA Analysis

Based on Loads
60

1
?>^^ Confidence Interval
40

O o
20

01
-60 0
T-NDT (°C)
-~ Threshold
for m- S (Constant)

60 at 100
0
-60
Xll 0
T-NDT (°C)
60
[ «p(0.026 T) -1
• T-NOT PC)>0

wo

Figure 12. Scatter Trends of CVN Fracture Appearance Change Drastically at NDT
 16

3.2 Systematics and Limitations.

HSST-02 can be summarized by the following
The results obtained above for unirradiated plate Table 2.

TABLE 2
Systematics of CVN Crack Arrest Indicators

Index Typical Temperature of CVN Crack Arrest Indicator (°C)

Median Confidence Intervals(2a)
95% 5%
NDT -11 -42 +19
TA +9 -4 +29
FATT +29 +16 +49
To +54 +40 +75

It is seen that the confidence interval for NDT is FATT

160
almost the double of the ones obtained for the three
+ Plate HSST-01
other crack arrest indicators. This of course is true 0 HSST-02
for all steels and conditions, because simply • Doel-l,-ll Weld
reflecting in all generality the statistical behavior of O Doel-l,-ll Forging
CVN arrest loads. For this material, FATT 100 -
(median)= NDT(median) + 40°C, while T o (upper
bound)= NDT (median) + 86°C, and FTE=NDT
(median)+ 85°C.

The correct description of the confidence intervals

of the arrest correlation indicators (e.g. Table 2) is
crucial for the least squares definition of their best
median values. The application margins however
are not equal to these confidence intervals; this is
briefly discussed in § 4.1.

It may also be wondered how far these indicators

are equivalent in terms of structural application. It -50 0 50 100
will subsequently be shown in § 4.1 that NDT and
Nil Ductility Temperature NDT (°C)
TA are indeed equivalent. By contrast, using FATT
or T o for nuclear applications is not necessarily Figure 13. In Service CVN-FATT Shift is Upper
equivalent to using NDT or TA. It is interesting to Bound for NDT-Shift
briefly discuss the FATT case. A compilation of in-
service observations for three different types of For plates HSST-01 and HSST-02 in all considered
reactor pressure vessel steels is presented on Figure surveillance capsules whatever their exposure, an
13. unique, "best" relationship FATT= NDT+42°C
holds and, within its 2a confidence interval of ±
 17

5°C, is in agreement with the unirradiated HSST- appearances (e.g. Figure 9) allows to minimize
02 results of Figure 9; NDT does approximately uncertainties and to detect eventual anomalies.
correspond to a constant SFA level of 16%. For the
unirradiated condition of the other two steels 4. PERFORMANCE OF "MASTER CURVE"
however, NDTs FATT, and this 50% equivalence FOR CVN ARREST LOAD CORRELATIONS.
level tends to decrease towards 16% with
increasing embrittlement: the in-service FATT- 4.1 Application to a Commercial PWR Vessel.
shift is an upper bound for the NDT shift. Quite
generally speaking, it has been found that It is instructive to compare three "recipes" to define
crack arrestability in a reactor pressure vessel. As
an example, this is done on Figure 14 for the
FATT< NDT+50°C (10) critical, circumferential weld of a commercial, 180
mm thick vessel in the beginning-of-life (BOL)
condition as well as in an extended end-of-life (E-
The use of T o entails even more conservatism, EOL) condition, well beyond the plant design life:
although never as much as when using the 41J. this corresponds to a maximum 290°C neutronic
indexing. exposure of 5.0 E19 cm"2 (>lMeV). Here, the
unirradiated R.T. yield strength of 428 MPa was
One can qualitatively grasp the patterns of Figure found to have increased in service up to an upper
13 at the light of equation (8). For a given test bound value of 570 MPa, while the unirradiated
temperature, any service-induced plasticity NDT of -30°C (identical for drop weight and Fa
decrease tends to decrease the difference Fm-Fu master curve method) was found to have increased
between the maximum load and the brittle to a conservatively estimated value of +73 °C, with
initiation load Fu: namely, the amount of ductile a 2o upper bound of+95°C (TA= +115°C). Note
stable crack growth decreases, thus Fu increases that the median NDT increase of 103°C is at least
and may even reach Fm at the limit of extremely 40% less in this case than the 41 J. CVN shift.
large embrittlement (there is then essentially no
ductile stable crack growth anymore, but shear lip The three "recipes" are as follows:
formation without brittle initiation at T o ). So, the
less plasticity, the larger Fu for a given arrest load 1) Use the ASME Code K,R curve, where
Fa and the lower SFA: FATT increases relative to RTNDT is replaced by NDT as defined by
NDT. Conversely, another simple way to put it is the CVN arrest load
that, the more ductile the steel, the larger the
fraction of shear to be expected at the NDT. 2) Use Wallin's 2c lower bound KIa
"master curve" where the indexing
These considerations explain why literature temperature is taken equal to the TA value
correlations of CVN-SFA to crack arrest defined by the CVN arrest load
temperatures can only be approximative: the SFA
"indexing" level cannot be unique- yet, 3) Determine FTE for the considered
paraphrasing, it is "more unique than others", -such vessel thickness, for the relevant R.T.
as the 41 J. energy level. In absence of instrumented uniaxial yield strength and for the relevant
information, the 50% FATT index can be used to NDT defined by the CVN arrest load.
conservatively measure embrittlement shifts, while
avoiding the possible biases of energy/lateral The three approaches are consistent with one
expansion indexing. More generally, the another and indicate a substantial safety margin of
combination of CVN loads and fracture
 18

Fracture Toughness K|a (MPaVm)

ASME CODE INDEXED TO NOT
260 WM.LIN 'MASTER CURVE" LOWER BOUND VESSEL CHARACTERISTICS
INDEXED TO T . • NDT* 20 °C
m FTE

200 •
Thickness 180 mm
NDT- - 3 0 °C
(BOD

160 -
Diameter 3330 mm

Operating Pressure 16MPa

100 •

Membrane Stress 155 MPa

LIMIT
SO
(Normal Operation)

-100 0 100 200

Temperature (°C)

Figure 14. Evaluation of Margin Against Brittle Fracture for Circumferential Weld of Commercial PWR
Vessel Operated Well Beyond Plant Design Life.

55°C for the E-EOL condition. Bv contrast- reflect its median properties (Cottrell, 1995). But
application of the current regulatory concepts to the median property here is derived through a
this vessel leaves almost no margin. correlation (Fig. 5). The 2CJ confidence interval
associated to the prediction has been estimated as ±
It is not too surprising that the first two "recipes" 22°C (§ 2.4); this has been found consistent with
above agree so well: the shapes of the KIR curve, of the NDT embrittlement trends and uncertainties for
Wallin's lower bound KIa curve and of the present the considered weld, as revealed by the data from
Fa "master curve" are all close to exponential up to six surveillance capsules and ten material
functions with quite similar slopes; the agreement, conditions (i.e. including baseline, ageing and
or lack thereof, depends only on the accuracy of annealing at two temperatures).
the difference TA-NDT (only 20°C, with an
uncertainty less than 15°C). Concurrence with the The view discussed in § 1 that the ASME K!R curve
third "recipe" is gratifying, given that the should always be indexed to NDT rather than to
FTE/NDT relationship is based on correlations RTNDT can now be further justified. Asking for
independently developed for non-nuclear steels, additional protection against low upper shelf
and from experiments seldomly involving toughness, especially in the case of an accidental
thicknesses as large as the ones of ASME-III pressure-temperature transient, amounts to ask
designed reactor pressure vessels. whether the arrested crack could reinitiate in a
ductile mode (in a regulatory perspective, brittle re-
It must be noted that the bounding rather than the initiation is excluded because FTE at E-EOL is
best estimate value has been used for the single, significantly less than the value corresponding to
experimentally defined input parameter (NDT or the PTS screening criterion). In the example of
TA) needed to apply the "recipes". This 2a upper Figure 14, the FTE occurs at NDT+80°C and
bound is distinct from the one considered in § 3.2- corresponds to a toughness K Ia of 140 MPaVm for a
which characterizes the material scatter at the stress of almost four times the estimated membrane
relatively local scale of a CVN specimen. Any stress in normal operation. In theory, and in the
large, non-arresting crack in a vessel would most unlikely event that the applied stresses reach
necessarily sample a large volume of material and yield magnitude, ductile re-initiation would thus be
 19

possible if the Kjc level corresponding to Jlc is less of this vessel in 1984 necessary, nor was the 1987
than 140 MPaVm; the tearing modulus would plant shutdown (a decision largely due to the cost
furthermore have to be quite low for such crack to of answering new requests to perform additional,
propagate to any significant degree. Let now ask extensive probabilistic PTS studies). Actually, the
what happens if FTE has been underestimated. It is fracture toughness of the material has been
obvious from Figure 14 that the critical KJc level adequate all along. The same is true for the vertical
for ductile re-initiation after arrest will change little Linde-80 weld. Investigation of a representative,
(at E-EOL, it is only 14 MPa Vm larger than at commercial surrogate weld irradiated in a test
BOL) and the same is true for the corresponding reactor at the plant operation temperature shows
steel J-R curve. Thus, correcting (i.e. increasing) that service-induced transition temperature shifts
FTE would not affect the margin against ductile re- are exceedingly less than according to regulatory
initiation, and this in turn illustrates well that expectations. The relevant comparison for BR3
replacing NDT by RTNDT does not meet its intent of surrogate weld and plate materials is illustrated by
providing protection against tearing instability for Figure 15 (note that the plate orientation is T-L).
steels with low CVN upper shelf. The only sound
way to address ductile initiation or post-arrest re-
initiation is through a fracture mechanics tearing Transition Temperature Increase ( t )
evaluation, as already stated (§ 1). 200 -

WELD
OHIO.*
Finally, it is useful to emphasize that the present 160
Mh O »

FTE/CAT/NDT approach applies to any service

condition. Insofar as its associated margins are
100
conservatively assessed, it is an absolute
formulation rather than a formulation based on
differences between E-EOL and BOL (baseline) 60 260 °C Irradiation ANNEAL
349%
TT41
• WELD
conditions; this is in contrast to current regulation Q PLATE (T-L)
and may prove helpful for oldier vintage vessels
1 2 3 4 6
whose unirradiated mechanical properties are ill-
Neutron Fluence >1MeV (cm-2 X1E19)
defined, with no archive material left for an
improved characterization. Regulatory Indexing

4.2 Conservatism of Current Regulatory Indexing to CVN Arrest Load

Approach.
Transition Temperature Increase (°C)

The practical implications of the present concepts

can be considerable when assessing safety margins
for PWR pressure vessels. The case of the vessel 160 •
discussed in § 4.1 is by far not unique. That the
current regulatory approach may be unduly
100 •
conservative has been known for some time (Fabry
96.a,b) and significant work is in progress with a
view to promote future regulatory acceptance of an
enhanced RPV surveillance methodology. In
particular, it has been shown by sampling the
1 2 3 4 6
Belgian BR3 vessel that the margins estimated in
Neutron Fluence >1MeV (cm-2 X1E19)
the past for the base metal of this vessel are much
larger than indicated by the prevailing engineering Figure 15. Irradiation/Anneal/Re-Irradiation Path
methods (Fabry 97.b): neither was the wet anneal Depends on Fracture Toughness Indexing Concept
 20

At 260°C irradiation temperature, the NDT shifts a constant term to which is added an exponential
as determined according to the present approach function of the reduced temperature T- T^,.0 where
are significantly less than the corresponding 41J. TKJ,.0 is the temperature at a reference level of 100
CVN shifts and are even below the predictions of MPaVm. By convention, T,^ 0 is usually quoted so
the USNRC Regulatory Guide 1.99 Rev.2 for the as to correspond to a specimen thickness of 25.4
quite less detrimental irradiation temperature of mm (one inch). The scatter is described by a
288°C (according to the Guide, correction to 260°C Weibull three-parameter distribution of slope m=4
calls for an additional penalty of +28°C). It is and of constant threshold of 20 MPaVm. The only
encouraging that a crack arrest philosophy can parameter to be experimentally determined is 7^°
offer such benefit, not only in terms of pressure- and in principle, this can be done to within a 2cr
temperature limitations during normal plant confidence interval of ± 15-20°C by testing a
operation, but also for plant management decisions minimum of six Charpy-size (or larger) samples at
with respect to the avoidance of accidental threats a properly selected temperature. A particularly
beyond vessel design life. useful feature of this approach in the present
context is that reconstitution technology allows to
5. MERGING OF CRACK ARREST AND fabricate precracked Charpy-V (PCCV) specimens
CRACK INITIATION APPROACHES. from the existing inventory of broken surveillance
remnants (van Walle, 1996).
Another direction of recent experimental and
theoretical efforts to modernize RPV surveillance Wallin's KJc evaluation procedure has been
centers on the crack initiation approach. It is systematically applied to surveillance and test
common knowledge that prohibitively large reactor irradiation results for which extensive
specimens, incompatible with insertion in a experimental documentation was available,
commercial surveillance capsule, are needed for the together with the CVN load-time traces needed for
direct measurement of valid static initiation crack arrest analysis by the present formulation. As
fracture toughness KIc in the transition temperature typical illustration, Figure 16 shows the application
range - i.e., vadid in the sense of linear elastic of Wallin's procedure to the USNRC-sponsored
fracture mechanics (LEFM). However, elastic- investigation of dose rate effects for the A302-B
plastic fracture mechanics (EPFM) indicates that reference plate (Hawthorne, 1990) while Figure 17
the space compatibility requirements can be met by condenses the companion Fa master curve
J-integral testing. The temperature range for evaluation. The R.T. static yield strengths have also
LEFM-valid KIc measurements or for EPFM-valid been carefully assessed, among other mechanical
KJc determinations depends on the yield strength of properties not considered herein. The major
the steel and on the specimen size, but the size outcome of this study in the present context is
restrictions are relaxed if some amount of plasticity presented on Figure 18.
can be accomodated, as feasible with the J integral
concept. For RPV steels in general, constraint Figure 18 clearly confirms that RPV steel
corrections remain small or negligible for median strengthening tends to remove the strain rate effect
Kjc levels up to = 100-120 MPaVm, even for the on fracture toughness: TA= 7^° for yield strength
deeply-precracked Charpy-V geometry tested > 600 MPa. This figure is very comparable to
under three-point slow bend (Anderson, 1993). similar displays of T ^ / - T ^ 0 (dynamic versus
Thickness corrections are easily applied if one static initiation) in function of the R.T. static yield
accounts for the fact that, in the transition range, strength (Barsom, 1970, Fabry, 1996.b). This
cleavage statistics is governed by the weakest link should be so insofar as crack arrest obeys a stress-
model. A "master curve" procedure has been related cleavage criterion, i.e. at NDT and at TA:
developed to incorporate these advances (Wallin, the strain rate corresponding to KIa is larger than for
1991) and is presently being standardized by the usual K,d tests, but this obviously does not
ASTM. The median Kjc (T) curve is represented by
 21

IC-1 0.53 E19 cm-2 IC-2 1.02 E19 cm-2

K j c Static Initiation Toughness (MPaVff) K JC Static Initiation Toughness (MPaVffi?
300 300
/ / 16%
260 260

200
Baseline
200 B
"""n9// / / /
1
160

100

60

O
• Pure Cleavage Failure
© Failure on R- Curve
160

100

60
• Mr • Pure Cleavage Failure
@ Failure on R- Curve

-60 0 60 100 160 200 260 300 -100 -60 0 60 100 160 200 260 300
Test Temperature (°C) Test Temperature CC)

IC-3 1.95 E19 cm-2 CE-1 0.88 E19 cm-2

K J c Static Initiation Toughness (MPaVm) K j c Static Initiation Toughnaaa (MPaVm)

96% 1 16%
/
260 260 Baseline /

200
Baseline I
200
_ 1
/©/a /
160 160

100 100

60 ^ ^ 60
• Pure Cleavage Failure • Pure Cleavage Failure
@ Failure on R- Curve 0 Failure on R- Curve
n
-mo -60 0 60 100 160 200 260 300 -WO -60 0 60 100 160 200 260 300
Test Temperature (°C) Test Temperature ( t )

CE-2 1.64 E19 cm-2 CE-3 4.01 E19 cm-2

K j c Static Initiation Toughness (MPaVm) K j c Static Initiation Toughness (MPaVm)
300 300
16%
/ " / / 96% I / 16%

•IL 1
260 260
Baseline/ / / Baseline/
200 / 200

160 160 i
•— ' • _ _
/ BS'tt / / / / 9193 /. 0
100 100

60 60
• Pure Cleavage Failure ^ & • Pure Cleavage Failure
0 Failure on R- Curve 0 Failure on R- Curve
0 0
-100 0 60 100 160 200 260 300 -wo -60 0 60 100 160 200 260 300
Test Temperature ("C) Test Temperature {°C)

Figure 16. Kjc- Shift for A302-B Reference Plate (Half Thickness, L-T)
 22

IN CORE CAPSULES CORE EDGE CAPSULES

ArraM Load (kN) Arrest Load (kN)
20
IC-1 0.66 E19 cm-2
NDT- 16.B °C
IC-2 t 0 4 E1S cm-2
WOr- 28.3 °C
16 IC-3 2.23 E1B cm-2
NOT- 32.8 °C

10

60 100
T-NDT (°C) T-NDT (°C)
Figure 17. CVN Arrest Loads for A302-B Reference Plate.

affect the nature of the physical effect responsible Note that "blue brittleness" can complicate this
for the trends. A simple physical rationalization has simple picture in some cases. Much more
been previously outlined to explain the importantly, it must be stressed that the ideal,
disappearance of the strain rate effect on crack physically-grounded correlation parameter for
initiation at increasing strength, and applies here as strain rate effects is not the yield stress, but the
well. Basically, strengthening entails an increase of ratio of the yield stress to the microcleavage
the athermal contribution to the flow stress and this fracture stress (Fabry, 1997.a). The simplified
shifts upwards the temperature range relevant to representation adopted here does not fundamentally
initiation (and arrest), bringing it increasingly affect the engineering and regulatory implications
closier to the cut-off temperature above which the of the present findings. The application details
thermally activated component of the flow stress however should not be based on the crude,
vanishes; once this cut-off is exceeded, all static indicative linear fit of Figure 18.
and dynamic properties become equivalent.
Temperature at Median K!a of 100 MPaVm
Temperature at Median K Jc ol 100 MPaVTn
T
120 A
-
A302-B Ref. Plate
a 1/4-T
100 0 1/2-T
D 18MND5 Forging
S3 Soudotenax Forging
80 • Soudotenax Weld
© Welds 72W, 73W
® 22NiMoCr37 Weld
60

40
ASME CODE „
20 . | OFFSET - 35 °C
(Constant I)

-20
300 400 600 600 700 800 900
R.T. Yield Strength (MPa)
Figure 18. Strain Rate Effect on RPV Steel Fracture Toughness
 23

By contrast to the above results, the ASME Code arrest formulation is entirely equivalent to the T,^ 0
and its nuclear application are based on the initiation approach based on the slow bend testing
assumption of a constant offset: T^-T^ 0 s35 c C. of precracked specimens; considering uncertainties,
Strictly speaking, this regulatory fix holds for the "break-even" may even hold at lower service-
lower bound toughness curves while Figure 18 induced strengthening. From a cycle-to-cycle
pertains to median "indexing" temperatures, but economy viewpoint, pressure-temperature
this distinction is primarily of academic nature; it is limitations are the paramount consideration and
clear that the Code does not properly account for these are dictated by K,a . On another hand,
the experimental observations. Why does the Code margins against accident govern any concern or
entertain such bias, especially if one considers that decision relative to end-of-life, extended end-of-
such type of observations and some grasp at their life, license renewal, ...; these margins are usually
physical underpinnings have existed since the calculated in terms of Klc/Kjc . Thus, the two
seventies ? At least two main factors have approaches to safety seem complementary, but this
presumably contributed. First, sustained interest in work indicates that they also are mutually
the crack initiation approach is relatively recent: reinforcing to a degree not always sufficiently
this seems traceable to the realization that design comprehended. When moving from the BOL to the
basis accidents did not comprehensively address EOL condition in most PWRs however, it is
the whole spectrum of hypothetical pressurized believed that the "last word" remains with crack
thermal shock transients. As long as crack arrest arrest, whether in normal operation or under
remained the linchpin of regulation, the strain rate hypothetized accident regimes.
effect did not matter much in practice: dynamic
properties are the relevant ones and it certainly 6. CONCLUSIONS.
seems a-priori convenient to "index" KIa to the
dynamic CVN test. Second, observations that the The results of this work lead to the view that
Klc/Kjc shift can for some melts exceed the 41J. current regulation and engineering practices in the
CVN shift are also somewhat recent (Hiser, 1985). field of nuclear reactor pressure vessel integrity
Furthermore, this effect seems to mostly affect insurance are affected by some conceptual flaws,
plates and forgings, by contrast to welds- which, and appear to deserve improvement in three main,
more often than not, govern safety margins. The interrelated respects:
reason for such patterns (Fabry, 1996.b) is that the
strain rate effect on fracture toughness tends to be
1) The influence of service exposure on the strain
compensated and obscured by biases stemming
rate sensitivity of fracture toughness is, and
from the CVN energy/lateral expansion indexing
should not be, neglected
procedure. The balance between the two sources of
bias depends on the steel. If the 41J. level is always
2) The 'physics' of the CVN impact test- the test
"triggered" by the initiation-linked energy fraction,
central to the presently accepted toughness
as often happens for base metals in the L-T
indexing procedure- is, and should not be, ignored
orientation, there is no CVN indexing bias for
(separation should be done between the roles of
service-induced shifts, and the KIc/KJc shift exceeds
crack initiation and propagation, shear lip
the CVN shift. At the opposite end of possibilities,
formation,..; the same integral features should not
if control of the 41J. level upon service moves from
be used irrespective of whether one wants to
the initiation-linked to the post-arrest energy
monitor service effects on initiation or on arrest
fraction, the 41 J.-shift will be excessive and may
fracture toughness)
even overcompensate the strain rate effect: this
happens for instance with Linde-80 flux welds.
3) Unwarranted and unnecessary mismatch is
introduced between brittle cleavage and ductile
It is clear from Figure 18 that, at yield strengths of tearing fracture modes when replacing NDT by
= 600 MPa and above, the present, Fa-based crack RTNDT for the (ill-served) purpose of trying to
 24

incorporate some protection against low CVN 8. REFERENCES.

upper shelf through the indexing of the unirradiated
cleavage fracture toughness. Ahlf.J., Bellmann,D., Fohl, J., Hebenbrock, H.D.,
Schmitt, F.J., and Spalthoff.W., "Irradiation Behavior of
These simplifications can cause significant biases- Reactor Pressure Vessel Steels from the Research
unacceptably overconservative ones sometimes- Program on the Integrity of Components", Radiation
Embrittlement of Nuclear Reactor Pressure Vessel
and seem to have obscured the cost/benefit
Steels: An International Review (Second Volume),
appraisal of the relationship between the crack ASTM STP 909. L.E. Steele, Ed., American Society for
initiation and crack arrest philosophies to assess Testing and Materials, Philadelphia, pp. 34-51 (1986).
RPV safety margins against brittle fracture.
Akiyama, T., Hirase, Y. and Urabe, N., "Development of
It is contended that the crack arrest approach, simple test method to evaluate crack arrest property",
mandatory under normal and upset plant operation Trans. ISIJ, Vol. 25, pp. 1267-1275 (1985).
conditions, is also, if properly implemented, the
most attractive and efficient approach for faulted Anderson, T.L., and Dodds, R.H. Jr., "Simple Constraint
Corrections for Subsize Fracture Toughness Specimens",
and accident regimes at ageing plants. A simple,
Small Specimen Test Techniques Applied to Nuclear
reliable implementation strategy using the load Reactor Vessel Thermal Annealing and Plant Life
signals of the instrumented CVN test has been Extension, ASTM STP 1204. W. R. Corwin, F.M.
formulated. A number of reasons behind the stated Haggag, and W.L. Server, Eds., American Society for
preference have been advanced and justified, the Testing and Materials, Philadelphia, pp. 93-105 (1993).
two main ones being:
Barsom, J.M., and Rolfe, S.T.,"Correlations between Klc
1) Less conservatism in defining heat-up and Charpy-V notch test results in the transition-
and cooldown operation limits temperature range", Impact Testing of Metals. ASTM
STP 466. American Society for Testing and Materials,
Philadelphia, USA, pp. 281-302 (1970).
2) Greatest safety guarantee against
accidents at no obvious economic penalty Berger, C, Ewald, J., Wiemann, W. and Wojaczyk,
when approaching end-of-life or extended H.G., "Ermittlung von Sprodbruch und
end-of-life. Bruchmechanischen Kennwerten mittels
Kerbschlagproben". Proc. Conf. Sitzung DVM AK
Bruchvorggnge. pp. 182-195, DVM (1979).

7. ACKNOWLEDGMENTS. Childress, C.E., "Manual for ASTM A-533 Grade B

Class 1 Steel (HSST Plate 03) Provided to the
International Atomic Energy Agency", Report ORNL-
This study has been jointly sponsored by SCK- TM-3193. March d 97 n .
CEN and by Tractebel. Special thanks are
expressed to Robert Gerard, Tractebel project Cottrell, A., "Keynote address: Introductory Comments
manager, for his competent support. The author is on Crack Arrest", in Proceedings of the TWI-HSE
also indebted to his colleagues, in particular Eric Seminar on "Crack Arrest Concepts for Failure
van Walle, Rachid Chaouadi, Jose Van de Velde Prevention and Life Extension". TWI, Abington,
and Alfons Verstrepen, for their help and interest, Cambridge, UK, 27 September (1995).
to Kunio Onizawa (JAERI), who provided the data
from the ASTM round robin on CVN Cowan, A., Fearnehough, G.D., Gilles, J., and Lees,
G.M., "Irradiation damage in mild steel: a comparison of
reconstitution, to Fred Stallmann (ORNL) for
crack arrest tests and Charpy criteria and influence of
making available the CVN load-time traces for the neutron spectrum". ASTM STP 426. pp. 107-134,
A302-B reference plate and to Mike Sokolov and American Society for Testing and Materials,
Shafik Iskander (ORNL) for kindly forwarding Philadelphia (1967).
advance copies of work to be published.
 25

Fabry, A., Van de Velde, J., Puzzolante, J.L., Van Changes in Reactor Pressure Vessel Steels". Report
Ransbeeck, T., Verstrepen, A., Biemiller, E.C., Carter NUREG/CR-5493. MEA-2376 (1990).
R.G., and Petrova, T., "Research to Understand the
Embrittlement Behavior of Yankee/BR3 Surveillance Hiser, A.L., "Correlation of Cv and KIc/KJc Transition
Plate and Other Outlier RPV Steels", Effects of Radiation Temperature Increases Due to Irradiation". Report
on Materials: 17th International Symposium. ASTM NUREG/CR-4395. MEA-2088 (1985).
STP 1270. D.S. Gelles, R.K. Nanstad, A.S. Kumar, and
E.A. Little, Eds., American Society for Testing and Holt, J.M., and Puzak, P.P., "Drop-Weight Test for
Materials, Philadelphia, pp.138-187 (1996.a). Determination of Nil-Ductility Transition Temperature:
User's Experience with ASTM Method E 208". ASTM
Fabry,A., van Walle, E., Van de Velde, J., Chaouadi, R., STP 919. American Society for Testing and Materials,
Puzzolante, J.L., van Ransbeeck, Th., and Verstrepen, Philadelphia, (1986).
A., "On the Use of the Instrumented Charpy 'V Impact
Signal for Assessment of RPVS Embrittlement", Irwin, G.R., Kraftt, J.M., Paris, P.C., and Wells, A.A.,
Evaluating Material Properties by Dynamic Testing, "Basic Aspects of Crack Growth and Fracture", NRL
ESIS 20 (Edited by E. van Walle), Mechanical Report 6598. Naval Research Laboratory. Washington,
Engineering Publications, London, pp. 59-78 (1996.b). D.C. (1967).

Fabry, A, "Characterization by Notched and Precracked Iskander, S.K., Nanstad, R.K., Sokolov, M.A., McCabe,
Charpy Tests of the In-service Degradation of Reactor D.E., Hutton, J.T., and Thomas, D.L., "Use of Forces
Pressure Vessel Steel Fracture Toughness", Small from Instrumented Charpy V-Notch Testing to
Specimen Test Techniques. ASTM STP 1329. W.R. Determine Crack-Arrest Toughness", Effects of
Corwin, S.T. Rosinski and E. van Walle, Eds., American Radiation on Materials: 18th International Symposium,
Society for Testing and Materials (1997.a). ASTM STP 1325. R.K. Nanstad, M.L. Hamilton, F.A.
Garner, and A.S. Kumar, Eds., American Soiety for
Fabry, A., Chaouadi, R., Puzzolante, J.L., Van de Velde, Testing and Materials, (1997).
J., Biemiller, E.C., Rosinski, ST., and Carter, R.G.,
"Embrittlement and Annealing of Reactor Pressure Kussmaul, K., Krageloh, E., Kochendorfer, A.,
Vessel Steels: Comparison of BR3 Surveillance and Hagedorn, E., "The Relationship between Notched Bar
Vessel Plates to the Surrogate Plates Representative of Impact Bending Test, Robertson Test and Drop-Weight
the Yankee Rowe Vessel", Effects of Radiation on Test". Nuclear Engineering and Design. 43, pp. 203-217
Materials: 18th International Symposium. ASTM STP (1977).
1325. Randy K. Nanstad, Ed., American Society for
Testing and Materials, Philadelphia (1997.b). McElroy, R.J., and Lowe, Jr., A.L., "Irradiation
Embrittlement Modeling of Linde 80 Weld Metals",
Fearnehough, G.D., "The small-scale test and its Effects of Radiation on Materials: 17th International
application to fracture propagation problems", Proc. Symposium. ASTM STP 1270. D.S. Gelles, R.K.
Conf. 'Dynamic Crack Propagation', pp. 77-102, Nanstad, A.S. Kumar, and E.A. Little, Eds., American
Noordhoff Publishing (1973). Society for Testing and Materials, Philadelphia, pp.68-85
(1996).
Hagedorn, K.E., Kochendorfer, A., "Measurement of the
crack arrest temperature in the test according to T.S. Nanstad R.K., Sokolov, M.A., and McCabe,
Robertson and comparison with the transition D.E.,"Fracture Toughness Curve Shift Method", Heavy-
temperatures as measured in the notch bend test", Arch. Section Steel Irradiation Program, Progress Report,
Eisenhuttenwes.. Vol. 42, pp. 39-47 (1971). NUREG/CR-5591. ORNL/TM-11568, Vol.7, N l,pp.
10-1 to 10-4 (1996).
Hagedorn, K.E., "Comparison of transition temperatures
from instrumented impact tests with the Robertson crack- Nichols,R.W., "Fast and brittle fracture studies related to
arrest temperature", Arch. Eisenhuttenwes. 54, N 6, pp. steel pressure vessels". Proc. Rov. Soc. A. Vol. 285,
239-244 (1983). pp.lO4-119(1965).

Hawthorne, J.R. and Hiser, A.L., "Influence of Fluence Onizawa, K., van Walle, E., Nanstad, R., Sokolov, M.,
Rate on Radiation-Induced Mechanical Property Pavinich, and W., "Critical Analysis of the ASTM
 26

Round-Robin on Reconstitution", Small Specimen Test of Commercial Power Reactors". Report NUREG/CR-
Techniques, ASTM STP 1329. W.R. Corwin, S.T. 4947, ORNL/TM-10459 (1987).
Rosinski and E. van Walle, Eds., American Society for
Testing and Materials (1997). Sumpter D.G., "Fracture Mechanics Interpretation of the
Drop Weight NDT Temperature", Drop- Weight Test for
Pachur, D., "Comparison of Drop-Weight and Determination of Nil-Ductility Transition Temperature:
Instrumented Charpy Impact Test Results for Irradiated User's Experience with ASTM Method E208, ASTM STP
RPV Steels", Effects of Radiation on Materials: 16th 919. J.M. Holt and P.P. Puzak, Eds., American Society
International Symposium. ASTM STP 1175. A.S. for Testing and Materials, Philadelphia, pp. 142-160
Kumar, D.S. Gelles, R.K. Nanstad, and E.A. Little, Eds., (1986).
American Society for Testing and Materials,
Philadelphia, pp. 195-210 (1994). Vandermeulen, W., Fabry, A., Puzzolante, J.L., Van de
Velde, J., Van Ransbeeck, Th., and van Walle, E.,
Pellini, W.S. and Puzak, P.P., "Fracture analysis diagram "Charpy Specimen Reconstitution as a Means of
procedures for the fracture-safe design of steel Providing Data for PLEX Licensing Purposes", Int. J.
structures". Welding Res. Counc. Bull., n 88 (May Pres. Ves. & Piping. Vol. 54, pp. 89-98 (1993).
1963).
van Walle, E., "Reconstitution: Where Do We Stand ?"
Puzak, P.P., Eschbacher E.W. and Pellini, W.S., Effects of Radiation on Materials: 17th International
"Initiation and propagation of brittle fracture in structural Symposium. ASTM STP 1270. D.S. Gelles, R.K.
steels". Weld. J. Vol. 31, pp. 561-s to 581-s (Dec. 1952). Nanstad, A.S. Kumar, and E.A. Little, Eds., American
Society for Testing and Materials, Philadelphia, pp. 138-
Robertson, T.S., "Propagation of brittle fracture in steel", 187(1996).
Journ. of the Iron and Steel Institute. Vol. 175, pp. 361-
374(1953). Wallin, K., "Statistical modelling of fracture in the
Schmitt, F.J., "Correlation of Cv and Drop-Weight ductile-to-brittle transition region", Defect Assessment in
Transition Temperature Increase Caused by Irradiation", Components- Fundamentals and Applications,
Effects of Radiation on Materials: 14th International ESIS/EGF9. Edited by J.G. Blauel and K.-H. Schwalbe,
Symposium (Volume II), ASTM STP 1046. N.H. Packan, Mechanical Enginering Publications, London, pp. 415-
R.E. Stoller, and A.S. Kumar, Eds., American Society 445(1991).
for Testing and Materials, Philadelphia, pp. 373-384
(1990). Wallin, K., Planman T., and Rintamaa, R., " Comparison
of some indirect measures of crack arrest".Proceedings
Server, W.L., and Oldfield, W., "Nuclear Pressure Vessel of the TWI-TSE Seminar on "Crack arrest concepts for
Steel Data Base". Report EPRINP 933 (1978). failure prevention and life extension". Abington,
Cambridge,UK, 27 September (1995).
Shoemaker, A.K. and Rolfe, ST., "The Static and Wiesner, C.S., Hayes, B. and Willoughby, A.A., "Crack
Dynamic Low-Temperature Crack Toughness of Seven Arrest in Modern Steels and their Weldments-
Structural Steels", Engineering Fracture Mechanics. Comparison between Small and Large Scale
VoI.2,N 4, pp. 319-340(1971). Experiments". Int. J. Pres. Ves. & Piping. Vol. 56, pp.
369-385 (1993).
Smedley, G.P., "Prediction and specification of crack
arrest properties of steel plate", Int. J. Pres. Vess. & Wiesner, C.S., "Predicting structural crack arrest
Piping. Vol. 40, pp. 279-302 (1989). behaviour using small-scale material characterization
tests". Int. J. Pres. Ves. & Piping. 69, pp. 185-196 (1996)
Smith, E., "The criteria for the arrest of a crack in a
pressure vessel". Int. J. Pres. Ves. & Piping. 71, pp. 55-
60 (1997).

Stallmann, F.W., "Analysis of the A302B and A533B

Standard Reference Materials in Surveillance Capsules