XA9744631

On the Proper Fracture Toughness Properties to be Used forPressurized Thermal Shock Evaluations

by

William L. ServerATI Consulting

3860 Blackhawk Road, Suite 160Danville, California 94506 USA

Embrittlement of nuclear reactor pressure vessel steels has been monitored throughextensive surveillance programs conducted routinely for U.S. nuclear power plants. Thelevel of embrittlement is factored into normal operating procedures for heatup andcooldown of the vessel. Additionally, since the early 1980s, pressurized thermal shock(PTS) has been a non-design condition that all U.S. plants have had to prove adequatetoughness relative to a severe overcooling and pressure-increasing transient event. Theregulations in the U.S. were revised in 1985 to include a simple screening criterion whichwas called the PTS Rule (Title 10, Code of Federal Regulations, Part 50.61). Thisscreening criterion is for the value of RTPTS which cannot exceed 270°F (132°C) for basematerials and axial welds and SOOT (149°C) for circumferential welds. RTPTS is theprojected value of the reference temperature, RTNDT, that corresponds to the end-of-license fluence for the vessel. Although these values are termed screening criteria, inpractice they are really limits that require serious attention if they are going to be reached.

In the early 1990s, a plant in the U.S. was projected to exceed these screening limits evenbefore the end-of-license based upon very conservative re-evaluation of the degree oftoughness degradation. This plant was Yankee Rowe, one of the earliest nuclear plantsbuilt in the U.S. There were several other integrity issues for the Yankee Rowe vessel inaddition to the level of radiation embrittlement. The ability to perform a reliableinspection of the beltline region was questionable due to the type of spot-welded claddingthat was used. Other issues had to do with defining the PTS transients and the severity ofthe transients based upon operating actions and reliability. Later studies indicated that theassumptions used by NRC in evaluating the Yankee Rowe embrittlement were veryconservative.

The traditional approach in the U.S. for evaluating PTS has relied upon probabilisticstudies in which the toughness has been based upon the data used to generate the lowerbound ASME Code Kic and KIR curves. A mean curve through this data with a Gaussianstatistical distribution assumed, except for a lower bound cutoff of somewhere between 2and 3 standard deviations, has been used. The RTKDT normalizing concept has beenmaintained which then requires the measured shift in Charpy V-notch toughness at the 41J (30 ft-lb) energy level be used to adjust the position of the Code curves. The Master

- 137 -

Curve method provides a unique alternative in providing a much better measure of realfracture toughness, plus the opportunity to use a more refined statistical distribution usingWeibull statistics. There are active moves in the U.S. to Standardize and Codify theMaster Curve (also termed To method). Benefits to both deterministic and probabilisticanalyses will be realized since more realistic measures of toughness can be used.

- 138 -

/v 11 consulting

I

I

Background

Fracture toughness of irradiated RPV steels is a key element of a PTSevaluation

Fracture toughness is needed over a temperature range extending intothe transition regime for RPV welds and base metal irradiated todifferent levels

Surveillance program materials should be utilized to the maximumextent possible

PTS evaluations have been performed both using deterministic andprobabilistic methods- Deterministic approaches require some sort of lower bound estimates of

toughness as a function of temperature or a mean curve with a safetyfactor added later (or both)

- Probabilistic methods require a statistic mean curve and an appropriatedefinition of the distribution characteristics

ATI Consulting

U.S. PTS Evaluations

• Combined deterministic and probabilistic calculations were used todevelop the NRC PTS Rule (10 CFR Part 50.61) in 1985

• Projected value of reference temperature, RTNDT, at end-of-license,, termed RTPTS

S - Circumferential welds: RTPTS < 149° C (300° F)1 - Axial welds and base metal: RTPTS < 132° C (270° F)

• Emphasis was placed upon probabilistic justification which includedRegulatory Guide 1.154 for assessing future plant-specificprobabilistic evaluations

• Yankee Rowe was the first vessel to attempt to use Regulatory Guide1.154 and a detailed plant-specific PTS evaluation to substantiatecontinue plant operation

ATI Consulting

Yankee Rowe (continued)

•P -

I

Other important vessel integrity issues were related to flaw distributionwhich hinged on a reliable inspection and transient severity as dictatedby plant operation

NRC had a difficult time assessing the Yankee Rowe probabilisticanalysis since many diverse fields are brought together

Low temperature (wet) thermal annealing was being considered as amitigative measure (similar to BR3)

Economics and uncertainty in the regulatory process finally led toearly retirement of Yankee Rowe in 1992

Later studies and analyses (some related to the sister vessel BR3 inBelgium) have shown that the degree of embrittlement in the YankeeRowe materials was not nearly as severe as projected by NRC

ATI Consulting

Yankee Rowe was being evaluated as a lead plant for license renewal(extension) in the late 1980s and early 1990s, but key issuesconcerning vessel integrity were revealed which led to early shutdownLack of knowledge concerning vessel material embrittlement was akey concern- RPV materials chemistry (and its variability) were not clearly known

- Surveillance program results were limited and inadequate

- Projections of embrittlement were made by NRC consultants based uponconservative approaches that indicated an immediate problem

- A supplemental surveillance program was initiated using surrogatematerials, but the question of surrogate quality was raised

A complete probabilistic evaluation using Regulatory Guide 1.154 wasconducted, but other issues and generic questions were raised

ATI Consulting

Standard Approach for Fracture Toughness ofIrradiated RPV Steels

Assume that we have a general fracture toughness curve shape that isdictated by a fracture parameter that we can easily measure fromsurveillance capsule materials

Ideally, the fracture parameter would be fracture toughness itself, butlimited irradiation space and the infancy of fracture mechanics goingback to the 1960-1970 time frame has generally forced us to useCharpy V-notch properties

U.S. approach uses the RTNDT parameter which is initially a marriageof Charpy V-notch at 68 J (50 ft-lb) and drop-weight NDT tests, andlater is adjusted based upon Charpy V-notch shifts at 41 J ( 30 ft-lb)

Other countries use something similar (such as Tk) which is also basedupon Charpy V-notch tests

ATI Consulting

Current U.S. Reference Toughness Approach

Initial RTNDT is based upon measured nil-ductility temperature, NDT,and attainment of at least 68 J (50 ft-lb) at a temperature 33°C (60°F)above the NDT

Adjusted RTNDT is based upon measurement and prediction of the shiftin Charpy properties at the 41 J (30 ft-lb) level:- Adjusted RTNDT = initial RTNDT + A RTNDT + margin

- Margin is a comfort factor based upon quality of measured surveillanceresults and general uncertainty in predictive equations (e.g.., RegulatoryGuide 1.99, Rev. 2)

Reference toughness curves in ASME Code provide a fixed curveshape for lower bound fracture toughness for use in performingdeterministic vessel integrity evaluations (and determining operatingpressure-temperature curves)

Fracture toughness K!c, K,R, KIa, (ksivin.)

N ) 4o o o

o o o r o j 0 5o o o o oro rooo

r o * » . 0 5 c » o 10

Fracture toughness Kjc, K,RJ Kla, (MPavhi)

33

oo

Xfl

©

o

Hoc

QTQ

n

ro

3crc

- 145 -

ATI Consulting

Initial RTNDT Concept

o a.t- xO Ul

a. i—

33en•z.

RTNDT = NDTT

RTNDT = TCV " 6 0 ° F

TEMPERATURE (°F)

(a )

I

1/1

K I R = 26.777 + 1.223 EXP{0.014493 [T - (RTNDT - 160]}

T - RTNDT

TEMPERATURE RELATIVE TO R T N ( ) f ( ° F )

( b )

ATI Consulting

Charpy Curve Shift with Irradiation

I

-240200

5 1 5 0

CD

•5 10°"5

Cha

i

50

-129

Temperature (°C)

•18 93

IUnirradiated

Irradiated

204

Upper shelf energy (USE)(Ductile failure)

30jt-lbLower shelf-i

(brittle failure) 1

-400 -200 0 200

Temperature (°F)

•AUSE[

5ptt1lb

316271

203

CD

136 oo

68

400L I o

600

JoSI

o

ATI Consulting

00

I

Key Issues for Current Reference ToughnessApproach

Heat-to-heat variability was to be removed using RTNDT method;results looking at more than 50 heats of RPV steels showed thatabsolute temperature was statistically better than T - RTNDT

Curve shape is assumed constant even for some low upper shelftoughness steels; limited results do not show a curve shape change forlower bound toughness

For probabilistic analyses, researchers have taken the original set ofdata used to derive ASME reference curves and fit a mean curve

Statistics applied to this mean curve are generally assumed to beGaussian, but a lower bound value of 2-3 standard deviations isassumed to account for potential non-Gaussian behavior on the lowtoughness end

ATI Consulting

ASME Code KIR Curve Dynamic Data andStatistical Fit

I

•C-

I

(TEMPERATURE-RTHDT), DEC C

50 100 150 200 250 300

A533B (HS5T 02)A508-2K, . DATA FROM ORIGINAL K1D CURVE

Id IK

T B - IIS.7H

B- I37.HS

B- IIB.BI

C- 119.BB

320

280

240

200

160

120

00

40

(HMI'tRATURF.-RTNnT), DEC. I

ATI Consulting

Key Issues for Current Reference ToughnessApproach (continued)

• Code curves were based upon a limited amount of data generatedbeginning in the late 1960s; additional data have not shown Codecurves to be non-conservative as compared to original data setsnormalized using RTNDT

g • Some materials can show a greater shift in static fracture toughness1 than the Charpy shift would suggest

• Dynamic and crack anest toughness shifts more closely match Charpyshifts (less data available, however)

• Precracked Charpy impact tests have been shown to be a betternormalizing parameter than RTNDT

• Direct measurement of "valid" toughness is better than using Charpyinferred toughness

Original KIR Curve fromWRC 175

i

Ln

I

O SHABBITSIWCAP-7623)

A RIPLING AND CROSLEY HSST,5th ANNUAL INFORMATIONMEETING, 1971 PAPER NO. 9

• UNPUBLISHED W DATA

MRL ARREST DATA 1972 HSSTINFO MTG

I I I II I I I-80 -40 0 40 . 80

TEMPERATURE RELATIVE TO NOT (°F)

ASME Code KIC Curve Basis200

Legend:HSST-01

• HSST-02+ HSST-03

A609 Class 2o HSST-01 Subarc Welda A633B Class 1 Subarc Weld• A633B Class 1 WeldA A533BWeldHAZ

K|,: " 33.2 + 2.806 exp (.02 (T - R T N D T + 100vF>|

J I-150 -100 100 160

T-RTNDT<°FI

ATI Consulting

Master Curve Toughness Approach is ActivelyBeing Codified in U.S.

Uses Weibull statistics and cleavage weak link concepts to defineposition of a master curve (better definition for deterministic andprobabilistic analyses)

Based upon elastic-plastic toughness results using small specimens(including precracked Charpy specimens tested in slow three-pointbending as fracture mechanics tests)

Number of specimens needed and test temperature requirements arebeing established by ASTM and ASME Code bodies (PVRC TaskGroup and International Programs)- ASME Section III - initial RTNDT, RTTo

- ASME Section XI -- adjusted RTNDT

- ASTM E08 ~ Standard for Master Curve and To determination- ASTM E10 ~ application to surveillance programs

ATI Consulting

Application of Master Curve in U.S.

Redefinition of initial RTNDT for one Linde 80 weld metal (WF-70)

Other weld metals (particularly Linde 1092) and at least one platematerial show much better toughness response than RTNDT approachwould suggest

Technical concerns:

- Constraint and specimen size effects- Uncertainty in To determination- Curve shape: KJc(med) = 30 + 70 exp [0.019 (T - To)]

- Dynamic and static toughness data- Definition of lower bound for deterministic use (lower

confidence/tolerance bound levels)

ATI Consulting

Application for Linde 80 Initial RTNDT

FRACTURE TOUGHNESS - WF-70DYNAMIC TOUGHNESS DATA, T = 0 F

300

250

200

I 150o2*

100

50

0

TEST DATA•

Kjc(med)

Kjc(95%)

KIR (-27)

-200 -10010

TEMPERATURE, F100 200

A l l Consulting

Recent Results for a Linde 1092 Weld Metal(1/2-T CT)

300

250+-

200

«M

<

< 1505

100

50 4 ^

. . . . ,

| To = -90.6C |

j7 /

/• 1/i j

i /i /

/ /

/ // /

/ /

ii1 — —

— — K J C Median

O KJC Raw Data (Adjusted to 1-1}

KJC (0.05 Tolerance Bound)

Margin Adjusted KJC

- * _ _ J I _ H— - . • 1 • 1 I. . ,1 . i . , . • i i

-1S0 -100 -50 50 100

Temperature (Deg C)

150

ATI Consulting

I

Ln

I

Recent Results for a Linde 1092 Weld Metal(Reconstituted Precracked Charpy)

300

250

200

^

£

5

150

100

50

-150

— — K J C Median •*•

O KJC Raw Data (Adjusted to 1T)

KJC (0.05 Tolerance Bound)

Margin Adjusted KJC

-100 50 100 150

Temperature (Deg C)

ATI Consulting

oo

I

Recent Results for a Linde 1092 Weld Metal(Precracked Charpy)

300

250

200.

<E< 150.

O

5100

50

a >

* ^ \ ^

—i—•___•—i—i

To = -100.3C

/I / —

//

/ ./ /

' / /s s-/ /

f // /

*>/

... . • u_.-»

//

/ J/ /

/ // /

/ ' // / /

/ // /

/ // • / •

/ /

/ /

/ /

/ /

LJ/ /

t /t /

/

—.—.—^__.—i

-riii I

//

i

-

KJC Median

D KJC Raw Data (Adjusted to 1T)

KJC (0.05 Tolerance Bound)

Margin Adjusted KJC

*• » ' I

!

-150 -100 -50 50 100 150

Temperature (Deg C)

ATI Consulting

vO

I

Master Curve Application to ProbabilisticEvaluations

Effect of a reduced RTNDT

- Approximately an order of magnitude reduction in conditional failureprobability for a reduction of 15°C (27°F)

- Modified surveillance programs can provide this key information

Effect of different curve shape and statistics- Initial calculations of conditional failure probability suggest similar results- Further calculations are needed

Need to carefully assess technical issues for direct comparison

ATI Consulting

oI

Original NRC Evaluations for PTS Rule

10"

LONGITUDINAL CRACK EXTENSION NO ARREST

NRC STAFF PRA RESULTS.-2

<Lul> -

oo<CXL

CxiLJCL

>-

ozuIDoUJDC

10"

10'-t.

10"

10"

LEGEND• PRA TOTAL9 STEAM L[NE BREAKSXOiOtMlIVSMALL BREAK LOCA

71

-EC

"

175 200 225 250 275 500 325 350

ILTI Consulting

Conclusions

Key element for PTS evaluations (either deterministic or probabilistic)is the RPV materials fracture toughnessYankee Rowe attempted probabilistic analyses with uncertaintoughness propertiesPast studies have relied upon conventional approaches using CharpyV-notch data for plant-specific materialsMaster Curve methodology provides distinct advantages for futureanalyses if modified surveillance results are utilizedU.S. Standards and Codes are implementing the To and Master Curveapproach