ML20059E307
| ML20059E307 | |
| Person / Time | |
|---|---|
| Site: | Yankee Rowe |
| Issue date: | 08/31/1990 |
| From: | NRC |
| To: | |
| Shared Package | |
| ML20059E305 | List: |
| References | |
| NUDOCS 9009100135 | |
| Download: ML20059E307 (28) | |
Text
.
4 2 J,.
i
- i ENCLOSURE s
j SAFETY ASSESSMENT OF YANKEE R0WE VESSEL I.
-INTRODUCTION In a letter dated l July 5,1990 from John D. Haseltine, the Yankee Atomic Electric Company (the licensee) submitted for staff review a report entitled,
" Reactor Pressure Vessel Evaluation Re p rt for Yankee Nuclear Power Station."
The report was_in response to NRC letters dated May 1, 7, and 15,- 1990.
The staff letters requested additional information( which was needed to assess the effect of vessel operating temperatures, be n.line material chemical composition;
-and material surveillance test results on thel integrity'of the Yankee Rowe reactor vessel. These concerns have potential impact on prior NRC reviews of, vessel integrity resulting from low irradiated Charpy Upper Shelf Energy (USE) and vessel integrity during postulated Pressurized Thermal Shock (PTS) and Low l
-TemperatureOverpressurization' Events (LTOP) events.
l The licensee's justification for operation of Yankee Rowe is that there is L
adequate assurance that risk of vessel brittle failure is very low.
This L
conclusion depends upon two factors:
(1)thefrequencyof.challengestothe
- 1. vessel, and (2) the probability of vessel failure given a challenge event (conditional vessel failure probability). Brittle failure challenge events 3
fall into 2 general categories:
(1) pressurized thermal shock (PTS) events, l
and(2)lowtemperatureoverpressurization(LTOP) events.
For both categories the 1.icensee has estimated a very low probability that a vestel failure will occur.
The frequency of challenge and probability of vessel failure for PTS
. and LTOP events are discussed in Section 11 and Section III respectively.
Additional information to support the licensee's conclusion was submitted in References 12 through 23.
b$$$00$
FDC
-.i
p.4, 2
,.s
'II. ' PRESSURIZED THERMAL SH0CK (PTS) EVALUATION II.1 Systems Evaluation of-PTS Limiting Events For PTS events the licensee has indicated that risk contributors can be divided into 3 groups:
(1) steam line breaks, (2) small break LOCAs, and (3) transients., For each PTS group the event resulting in the most limiting temperatureandpressureconditions(fromavesselfailureperspective)is considered to be representative for the group.
The frequency for a group is the sum of the frequencies for each event in the group.
For PTS the staff's review focused on the following considerations:
(1) completeness of the events considered; (2) the adequacy of the thermal hydraulic analyses; (3) adequacy of the event frequency estimates-including humanerrorcontributions;and(4)adequacyofthelimitingeventsselected.
II.1.1 Completeness of PTS Events Considered In its PRA submittal on PTS for Yankee Rowe, the licensee performed a systematic evaluation of initiating events (IEs) that could lead to primary system' overcooling coupled with primary system repressurization. These IEs were grouped into four major categories. Category. I is main coolant system (MCS)inducedevents. This category of events includes MCS - initiated cool-down events, depressurization events, and injection events, with both the MCS intact and faulted.
Category II is secondary system:. induced events.
This-category includes events initiated due to steam removal, feedwater flow, steam generator blowdown, and steam /feedwater flow control abnormalities.
Category
.III is general transients which do not directly result in' initial MCS cooldown
' and are not related to support systems but, if followed by other system failures could result in cooldown events. Category IV is events not necessarily resulting in initial MCS cooldown but involving support systems which have the potential to-impact other frontline systems which could cause MCS cooldown. The licensee also reviewed the PTS evaluations for H. B.
Robinson and Calvert Cliffs performed by Oak Ridge National Laboratories to
' p!
3 assure that the Yankee Rowe: evaluation took into account sequences found to be
-significant contributors to thermal shock at these plants.- The licensee examined the operating experience at Yankee Rowe (including all the trip logs);
and concluded that there has never been an overcooling event at Rowe. The plant design and the Yankee Rowe Probabilistic Safety Stud / were likewise reviewed to identify any plant unique cooldown sequences.
11.1.2 Thermal-Hydraulic Analyses for Transients Affecting PTS Based on system and thermal-hydraulic considerations, each of the initiating events were evaluated and the initiators relevant to PTS concerns were
-identified.
Event tree sequences were then developed for each event associated with the relevant initiators concerning PTS. Support systems were treated in a separate auxiliary tree. Quantification of event sequences and endstate: was
. performed based on the system models, dependencies, and human actions.
~Endstates with frequencies higher than 10-8/ reactor year were selected for potential further thermal-hydraulic and fracture mechanics analysis.
Based on groupin'g sequences with similar plant thermal-hydraulic behavior, this process resulted in-the final set of. initiating events being grouped into three
- categories with four corresponding event trees
- ' steam line breaks upstream or-downstream of non-return valves, small break LOCAs, and transients.
For each of the above identified four event trees, thermal-hydraulic analyses were performed to model the spectrum of. overcooling events. The transient downcomer temperature and MCS pressure were calculated and bounding cases affecting PTS concerns.were identified.
TheLlicensee used the CEPAC computer code to perform. scoping calculations for.
=the' events of concern to predict limiting cooldown transients at Yankee Rowe.
Based upon pressure and temperature response a small break LOCA of 15/16 inches at the reactor coolant pump suction and three cases of main steamline break were-found as the limiting transients relative to PTS concerns. These limiting transients were analyzed in greater detail using the RETRAM computer code, the combination of RETRAN and EPRI models, or the combination of RETRAN
q 4
4 and REMIX codes. The RETRAN computer < code is designed to analyze the response Jof plant systems during both normal and transient conditions. The licensee's capability of using RETRAN for main steam line break analyses was reviewed and approved by the staff in 1983.
The licensee asserted that.the CEPAC code is.
similar but simpler than the RETRAN code. The CEPAC code has not been reviewed by the staff. However, the limiting transients results were not based upon CEPAC calculations.
The EPRI model has been_used for the-non-stagnant flow conditions in the Calvert Cliffs PTS analysis.
The REMIX code was used for the SBLOCA case without offsite-power available, where flow stagnation occurred.
The staff has evaluated the adequacy of the licensee's use of REMIX for the Yankee Rowe plant SBLOCA case. We feel that sufficient conservatism exists in this analysis. There are other conservative assumptions considered in the SBLOCA analysis such as early stagnation in the downcomer area, low decay heat, coincident loss of offsite power and an assumption that all three trains of
~
p safety injection are injecting water to the MCS. The first three of these assumptions result in minimal mixing of the cold SI water with the hot primary p
system water. The fourth assumption maximizes the amount of cold water added to the primary system. The result is a conservative (colder) downcomer water temperature.
In the_ main steam line break cases, there are conservative assumptions spplied such as zero power at event initiation, low decay = heat, dry steam to-the break, coincident loss of offsite power, non-return valve failure, etc. As in-the SBLOCA cases these assumptions minimize mixing in the primary system and maximize primary system cooldown.
The following' design features were found to be significant in the analyses:
The ' charging pumps trip on a safety injection signal.
This feature helps assure that the maximum repressurization achievable during a.LOCA or transient that may initiate safety injection is limited to the shutioff head of the safety injection system (1550 psig).
The safety injection pumps have relatively low capacity and a shutoff head of 1550 psig when HPSI and LPS are aligned in series. When not aligned in series the shutoff head.s limited to 800 psig.
i
I.;
R H
5
]
There is only one pressurizer PORV.
This reduces the probability of a stuck open PORV (relative to two PORYs) initiating a cooldown event.
l There is only one turbine bypass valve, and it has low capacity. This limits-therateofpotentialcooldown(ifthevalvefailsopen).
The emergency atmospheric steam dump valves have low capacity. This.
i limits the rate of cooldown should the valves fail open.
The condensate pumps trip following a steam break in the vapor containment.
Emergency feedwater pumps must be manually started.
The Emergency Operating Procedures direct the operator, in response to iminent~ PTS conditions, to stop safety injection pumps and low pressure-safety injection pumps if there is sufficient subcooling and pressurizer, level.
Although the plant has primary system loop isolation valves, emergency.
J l
operating procedures only require their. operation during a steam generator tube rupture in order to isolate the faulted generator.
For other LOCAs y
inside the vapor containment, the operators are instructed to not isolate L
the break location.
Isolation of a break could result in significant repressurization.
The'feedwater pumps trip on reactor scram or low suction pressure. Above 15 percent power, operators are instructed to isolate feedwater ficw by closure of'the feedwater regulating valves and the feedwater motor-
. operated isolation valves. These measures limit the chance and severity.
of-an overcooling event caused by overfeed of the steam generators.
i
/
f Yankee Rowe is also unique in the large number of ways in which water can be supplied.to the steam generators. Among these multiple paths, all flow sources
-however, are dwarfed in volume by the boiler feedwater pumps. The feedwater
6 w,
control system has independent controls for each steam generator such that a single failure in the control system would not result in overfeeding more than one steam generator.
If another system should begin to supply additional water j
to the steam generators (e.g., the charging system),'the feedwater control system would cut back on the flow from the boiler feedwater pumps to maintain steam generator level.
In view of these plant specific h atures and the l
modelling assumptions used by the licensee, the staff considers that the thermal-hydraulic analyses are conservative and reasonable. We note that the results are also consistent with other similar analyses such a the Robinson and Calvert Cliffs PTS studies.
L 11.1.3 Frequency of Cooldown Events Threatening the Vessel Yankee Atomic has estimated that the frequency of sequences that would i
~
significa..tly challenge the integrity of the reactor vessel due to pressurized thermal shock to be about 5 E-4 per reactor year. Smali break LOCAs result in
.the most limiting thermal hydr 6ulic conditions of any of the sequences analyzed.
Yankee Atomic estimated this frequency by partitioning the WASH-1400 small break LOCA frequency (for break sizes between 0.5 and 2 inches) based on the number of pipe segments inside the vapor containment-that were between 1 L
.and2inchesininteriordiameter(I.D.).
The limiting sequence (combination
~
~f frequency and thermal hydraulic conditions) was estimated by Yankee to be a o
l LOCA about 1 5/16 inches I.D. where the estimated minimum downcomer temperature L
was 151'F and the maximum RCS pressure after cooldown was 670 psi.
This
~
analysis did not, however, take into account the possibility-of the operator violating his Emergency Operating Procedures and attempting to isolate the break.
Such action could lead to an RCS maximum pressure equal-to the shutoff head of the safety injection pumps. At the staff's request Yankee Atomic performed an analysis of such a sequence. The licensee concluded that it was not a significant event because of the small amovnt of small bore piping which is isolable, the frequency of a small break in any location, ard the operator-training and procedures which direct operators not to isolate breaks inside the vapor. containment, i
7
- 3 The staff'has reviewed the licensee's event frequency estimates in considera-tion of the plant specific features of Yankee Rowe. The limiting event frequencies are reasonably consistent with values used in other studies.
The treatment of human error in the Yankee Rowe PTS PRA is judged to be conserva-tive or non-conservative depending on the timing of the error. The PTS thermal hydraulic analyses indicate that small break LOCAs give the worst combination of low primary system temperature, high primary system pressure, and high cooldown rate. 'The staff believes that the_ licensee's estimate of 5x10-4 per reactor year as the frequency of a small break LOCA is consistent with the frequency of 1x10-3 per reactor year typically used in PRAs.
11.1.4 Adecuacy of PTS Limiting Events The licensee performed a systematic review of the Yankee Rowe features in order to identify potential overcooling sequences. The licensee then grouped the possible events on the basis of similarity in thermal hydraulic (TH) response.
For each group a limiting event was determined based upon consideration of
(
event frequency and 'the severity of pressure temperature conditions (relative to vessel failure) resulting from the event. The staff concluded the. events considered are reasonably comprehensive, the thermal hydraulic analyses, methods, assumptions and results are reasonable.
With regard to the frequency estimates, the most important considerations are the insensitivity to human error and the relative frequency values. The systems failure estimates used are considered to be reasonable because they are consistent with state-of-the-art PRA applications. The event frequencies were also foun' to be d
relatively insensitive to human error since the limiting events would not change significantly even if the human error probability (at times greater than 1l hour) changed by a factor of 100. Therefore, based on these systems, thermal hydraulic, and event frequency studies, the staff concludes that there'is reasonable assurance that the limiting events have been properly identified, t'
g
. j.
8
+
II.2 PTS Materials Evaluation 11.2.1 Background-The Pressurized -Thermal Shock (PTS) rule,10 CFR 50.61, adopted on July 23, 1985, establishes a screening criterion that is a limiting level of enbrittle-ment beyond which operation cannot continue without further plant-specific evaluation. The screening criterion is given in terms of RTHDT, calculated as' j
a function of the beltline material chemical composition (copper and' nickel contents) and the neutron fluence according to the procedure given in the PTS rule, and called RT to distinguish it from other procedures for calculating PTS RT The greater the amounts of copper, nickel and neutron fluence the HDT.
higher the RT for the material and the lower its fracture resistance. The NDT screening criterion is 270'F for plates and axial welds and 300'F for the
.circumferential weld. The rule does not consider the effect of vessel operating temperature and material surveillance test results on the calculated RT The rule is currently being amended to calculate the RT using the l
PTS.
PTS trend curves in Regulatory Guide (RG) 1.99, Rev. 2.
The licensee, in response.to our concerns about embrittlement; previded the following significant information:
1.
The reported copper and nickel contents of the weld metal are now assumed to be' higher, because the actual values are unknown, and.the licensee elected to report measurements made for a " sister" vessel, the Belgian BR-3 reactor, instead of previously-reported measurements for a weld in the upper head of the Yankee Rowe vessel.
2..
The nomina 1' operating temperature is 500 F, whereas the data base for R.G. 1.99, Rev. 2 and the PTS rule is from reactors that operate at a nominal temperature of 550*F.
(Lowerirradiationtemperatureincreases RTNOT*)
3..
The surveillance data from the Yankee Rowe vessel, all of which date from the late 1960's, show high sensitivity to neutron embrittlement, even l
3-
.4=
7,..
9
,o consider;ng the effect of the lower irradiation temperature. These data were known to the AEC but were discounted because the operating tempera-ture inLthe first few fuel cycles was known to be low (500'F), and there were. coast down periods involving low operating temperature of several j
months duration at the end of the fuel cycles.
11 2.2 -Evaluation of Material Properties
- l The beltline in the Yankee Rowe reactor vessel consists of an upper plate, a lower plate, two axially oriented welds and one circumferential1y oriented 1
weld. The only surveillance data from these materials is from the YNPS beltline upper plate. The chemical composition and heat numbers for the upper cnd lower plates are known.
The chemical composition and heat numbers for the axial and circumferential welds'are unknown.
Eighty-five percent of' the accumulated irradiation occurred at a cold leg temperature between 500*F and 520'F. The remaining fifteen percent of the accumulated irradiation occurred.
at cold leg temperatures less than 500'F.
The. staff's estimato and licensee's estimate of the mean value-reference temperature in 1990 for each Yankee Rowe beltline material at its peak neutron i
flux location are tabulated in Table I.
The mean value reference temperature istthe sum of the unirradiated reference temperature and the increase in
~
reference temperature resulting from neutrt.. irradiation at an irradiation temperature-of 500*F. The staff's estimate of the increase in reference-temperature-was estimated for_ the peak neutron fluence in.1990 at the inside 19 surface of the reactor v_essel.
The peak neutron fluence is 2.3 x 10 n/cm2 19 for the' upper shell plate, 2.05 x 10 n/cm2 for the' lower shell plate and 19 circumferentiel welds, and.38 x 10 n/cm2 for the axial welds. The neutron fluences were calculated by the licensee using a methodology documented in letters from G. Papanic, dr. dated January 22, 1986, October 28, 1986 and February-4,1987.
The staff review of the licensee neutron fluence calculation methodology is documented in a letter to the licensee dated March 10, 1987.
I The licensee is currently recalculating these fluences. The results of this analysis will not be available before October,1990.
. 7 10
- w.
I f
11.2.2.1 Upper Plate-U L
The licensee's estimate of the increase in reference temperature for the_ upper l
plate was derived from Yankee Rowe and BR-3 surveillance data, but did not
- 3 L
correct the'BR-3 data (irradiation temperature 525-540'F) to account for the 3
lower irradiation temperature (500'F) of the Yankee Rowe reactor vessel.
In addition, the licensee doubled the neutron fluence values reported for the Yankee Rowe surveillance data. The licensee did not include the.effect of f
lower irradiation temperature in its analysis because they claim that the coe,rse grain size of the upper plate surveillance material eliminates'the effect of irradiation temperature.
The licensee's coarse grain theory is based on an argument that irradiation-induced defects in a coarse grain structure are more stable than irradiation-induced defects in fine grain structures. Since the irradiation-induced defects are more stable in the i
coarssgrainstructure,thelicenseeconcludesthatthelowerirradihtion.
temiarature of its reactor vessel will not affect the BR-3 data. Because of very limited surveillance data applicable to the Yankee vessel, the staff does not consider that the-licensee has yet substantiated this theory.
A literature survey performed by the staff revealed three reports which indicate irradiation temperature has an effect on neutron irradiation embrittlement.
In Reference 1 (Stallman, ORNL),. irradiation temperature was.
found-to increase transition temperature by 0.5 to 1.5' degree per degree decrease in irradiation temperature from 550*F, for a heat of A 533-8 plate (the 02 plate from the ORNL HSST program). Odette (Ref. 2) has similarly found
'i a factor of I degree per degree using'a large data base of. surveillance data.
In addition, Lowe (Ref. 3) has found about 0.7 degree per degree change in irradiation temperature, for Linde 80 welds. Overall, these factors are
- probably dependent on the composition, processing history, etc. of the steel'.
Although, References 1 and 2 do not.specifically address coarse grain structures, the staff included the irradiation temperature effect in its evaluation because the licensee has not presented any Charpy data that shows the reference temperature for its plate material does not increase with a i
\\
H 31 f
'd'ecreaseinihradiationtemperature. The staff estimate of the reference I
temperature includes a correction for-irradiation temperature and is based on the analysis performed by Odette (Ref. 4).
11.2.2.2 Lower plate
)
i The licensee's estimate of the increase in reference temperature for the lower.
plate was derived from Yankee Rowe and BR-3 surveillance data, but was not corrected for lower irradiation temperature or the increase in the amount of l
nickel in the lower plate compared to the amount in the surveillance plate.
The lower plate has 0.63 percent nickel and the surveillance plate has 0.18 percent nickel. The licensee believes no correction is necessary because of the postulation that the coarse grain of the plate eliminates the nickel.and irradiation temperature effects.
To support the conclusion that the nickel effect n'ay be eliminated for coarse-grain structural material, the licensee reports the conclusions of a Maricchiols(Ref.5) study.
In this study, " Nickel was reported to reduce the damage, introduced by neutron irradiation up to a content of about 1.0 percent." This study appears to contradict the results from a statistical analysis of commercial US reactor surveillance data.' The results of the statistical analysis of base metal surveillance data-is reported Table 2 of RG 1.99, Rev.2, which is contained here as Table 2.
This Table indicates that for.a particular amount of copper, nickel increases the chemistry factor, which results inanincreaseinthematerial'sreferencetemperature(damage),notadecrease as reported in the Maricchiols study.
Since the statistical analysis-performed to derive the chemistry factor in the tables in RG 1.99, Rev. 2 indicates that-
- @ re is'a nickel effect and the licensee has not provided any data from coarse-grain structure material that shows there is no nickel effect, the staff concludes there is a nickel effect.
The staff' estimates that an increase in nickel from.18 percent to.63 percent at 500*F irradiation temperature results in an 80*F increase in the reference temperature. This value is based on analysis by Odette (Ref. 4). The staff
-12 r
considers that it is important in order to determine whether longer term e
operation should be authorized to determine the effect of coarse grain for operating.temperattire and metal chemistry representative of the Yankee Rowe vessel.
11.2.2.3 Circumferential and Axial Welds The circumferential. weld is one of the critical materials. The axial welds are not because.they are exposed to only one-sixth of the peak fluence due to their azimuthal location relative to the core.
The licensee estimated the increase in reference temperature for the circumferential welds using the methodology recommended RG 1.99, Rev. 2 and a correction. factor for irradiation temperature. As disM sed previously, the chemical composition of the Yankee Rowe beltline welds is not known. The licensee used the chemical composition of a BR-3 weld to estimate the increase in reference temperature resulting from neutron irradiation. _ The licensee believesthattheamountsofcopper(.183 percent)andnickel(.70 percent),
reported for the BR-3 weld may be used as estimates for their welds because the.BR-S weld'and Yankee Rowe beltline welds were fabricated by the same vendor Babcock Wilcox, using the same process (submerged arc) and the same procedures (copper-plated filler wire with Linde 80 flux). However, this conclusion is not supported by industrial experience.
The B&W Owners Group-(Ref. 6) evaluated the' weld chemistry of_ Babcock & Wilcox fabricated Linde 80 welds. The reports indicates that the total copper concentration in the weld metal results from a combination of the amount of copper. plating and the base I
filler wire alloy. concentration. However, the principle source'of copper in L
the as deposited weld metal is the amount of copper plate. Reference 6 L
. indicates the amount of copper varies from heat of wire to heat of wire.
Until the licensee determines the chemical composition of the circumferential
- and axial welds, the amount of copper in the welds should be considered unknown and bounding values of copper should be used to estimate the effect of neutron irradiation on the weld metal's reference temperature.
i z
e k
13
+
4 lThe staff used two bases for estimating RT f r the circumferential weld..
NDT One method uses a set of data' compiled by Odette'lRef. 4) for 500'F,irradia-L
T Rev. 2 methodology, bounding values for copper and nickel, 0.35 percent and 0.70 percent respectively, and 50'F for the irradiation temperaturo effect.
This yields a value of 330'F for RT Figure 1(Figure 4fromReference NDT.
- 1) raports the increase in reference temperature for weld metals and base.
metals (plates)atirradiationtemperatureof500'F. The dashed line has been added to represent the increase in reference temperature for the circumferential weld using the RG 1.99, Rev. 2 bounding method with 50'F correction for the irradiation temperature effect.
Since this curve bounds all the existing weld L'
data in the Odette report, this method has been used to estimate values.of of reference temperatures for the circumferential and axial weld metal where
- the amount of copper is unknown and the weld metal is subject-to 500*F irradiation temperature.
The predicted value of the reference temperatures in 1990 for the circum-
. ferential weld and longitudinal welds are 330*F and 226'F, respectively.
These values are for high copper welds.
If the chemical analyses of these welds indicates that_the amounts of copper are significantly-less than 0.35 percent copper and 0.70 percent nickel, the reference temperatures will be significantly reduced.
For example, if the~circumferential weld had 0.20 percent copper and 0.70 percent nickel, the reference temperature would be' 262'F (212*F from RG 1.99, Rev. 2 and 50'F for irradiation temperature effect).
l-a p
Thus, the staff considers that it is important in order to determine whether longer term operation should be authorized to determine the actual chemical i
[
composition of the circumferential weld.
- 11.2.3 Summary l-l The level of uncertainty is higher for the estimates of RT values for Yankee NDT l-Rowe than has been encountered for other reactor vessels. Therefore,. con-L sidering the uncertainty in weld chemistry and the effects of coarse grain, the staff believes the RT f r both the-lower plate and the circumferential weld NDT should be assumed to be 350'F 50*F.
l l
14
. e
,, <.o 11.2.4 Probabilistic Fracture Mechanics
.t Although the Yankee-Rowe reactor vessel beltline has not received any inservice volumetric inspection, other areas of the reactor vessel have b'een inspected.
These inspections report that the welds do not contain any flaws exceeding the
~
acceptance' limits defined by 10 CFR.50.55a and ASME Code Section XI.
-In developing the PTS rule, the staff used a " Marshall" distribution (Ref. 7) j of flaws. The "Harsha11" distribution, which was developed in the mid-seventies, characterized defects in a vessel entering service, including defects considered acceptable according to fabrication codes and undetected during inspection.
The Yankee Rowe reactor vessel beltline was fabriacted using methods and materials similar to other commercially operated reactor vessels except that the clad in the Yankee Rowe reactor vessel is spot-welded and the clad in all.
other commercially operated reactor vessels is fusion welded.
Hence, except for the effect of spot welding, the distribution of flaws in the Yankee Rowe reactor vessel should oe similar to the distribution in other commerically operated. reactor vessels.
During-the Summer 1990 refueling outage, the licensee ultrasonically examined the reactor pressure vessel closure head and upper regions of-the pressurizer, which contained spot-welded clad similar to the clad in the reactor vessel I
beltline. The staff inspector (Ref. 8) concurred with the licensee's evalua-tion of the ultrasonic data that there was no extension of previously observed cladding cracks into the base metal. This inspection supports the conclusion that postulated cracks in the spot weld in the reactor vessel beltline cladding.
would not progress into the base metal due to the operation of the reactor-vessel and the " Marshall" distribution appears to be applicable for the Yankee Rowe reactor vessel beltline.
However, until the licensee performs an inservice inspection of the beltline materials, the conditional failure prob-ability should be increased to account for the uncertainty in service-induced flaws.
G' 15 s..
1
.To assess the effect of cracks on the probability of failure given the occurrence of a transient event, the licensee utilized probabilistic fracture mechanics analysisc The staff guidance for estimating the conditional probability of reactor vessel failure is provided in Regulatory Guide 1.154 Thermal and stress analyses for the vessel wall have to be performed.
Input for this analysis includes the primary system pressure, the temperature of the
. coolant in the-reactor vessel downcomer, the fluid-film heat transfer coefficient adjacent to the vessel wall, all as a function of time, and the vessel properties.. Probability density distribution functions for flaw size, crack initiation fracture toughness, crack arrest fracture toughness, and i
either the vessel materials nil-ductility rearence temperature, or the vessel
?
. materials copper and nickel. contents, and fast neutron fluence have to be developed.
For each transient of interest, many deterministic fracture mechanics analyses have to be performed to determine the number of times the crack penetrates through the vessel wall per 100,000 runs (forexample)asa, l
result of the stress level, flaw size, toughness and other variables se heted for each run. The calculations are performed with a probabilistic fracture mechanics computer code based on the Monte Carlo simulation technique.
i 1
1 The licensee has performed a probabilistic fracture mechanics analyses for several transients. -For example, the licensee performed a sensitivity study that predicts conditional probability of reactor pressure vessel failure is k
approximately 10-3 given the occurrence of a;1.3 inch-diameter small break LOCA event, which they believe is the controlling event, and for the reference temperatures reported in Table 3.
The reference temperatures used by the licensee are similar to the values estimated by the staff except for the lower plate. The conditional failure probability for a small break LOCA event for the lower plate'with a reference temperature of 325'F is less that 10-5 This plate-has a. low conditional failure probability at these high. reference tem-peratures because only a small portion of the plate-is in the-beltline region.
Considering the results from the'325 F reference temperature analysis, a mean value of 355'F should not significantly change the conditional failures probability.
i
. i
. u
16 When evaluating the rt 41ts of the licensees sensitivity study one must con-sider the assumptions use,' in the analysis.
The licensee assumed a "Harsha11" distribution of flaws and that cracks would arrest according to the average crack arrest data (Ref. 9). The flaw density distribution function used by the licensee may not be rep *esentative of the Yankee Rowe reactor vessel because of its unique spot cladding on the inside surface of the reactor vessel.
It also appears that the licensce's analysis may not have adequately accounted for the low upper-shelf energy of the vessel material which affects the " arrest" of initiated cracks. Given these apparent deficiencies and others that have been noted to date, the staff d es not accept the licensee's estimate of the condi-tional failure probability of the reactor pressure vessel.
The staff and its contractor are continuing a detailed review of the licensee's analysis. The review of this analysis should be completed by the end of October 1990. The results of this review will be important in determining i'ture action in connection with this license.
In view of these uncertainties the staff is unwilling to accept the licensee estimate of conditional vessel failure prob.
ability of 1x10'3 given a specific size small break LOCA.
In the meantime the staff judges it would be prudent to assume the conditional probability of reactor pressure vessel's failure to be in the range of 10~I to 10-2 l
l PTS Conclusions As discussed above, the staff concludes that P ere are substantial uncertain-ties associcted with weld chemistry ed the effects of coarse grain plate-material on the shift in reference temperature. These uncertainties could result in reference temperatures significantly higher than the screening criteria specified in the regulations.
Recognizing these uncertainties, the staff concluded that a more conservative range of conditional failure prob-L ability (by a factor of 10 to 100 relative to the licensee's estimate) was appropriate. This range when coupled with estimates of likelihood of the occurrence of PTS events and consideration of the plant specific features at L
Yankee Rowe important to such events, leads the staff to conclude that opera-tion until the end of fuel Cycle 21 is acceptable from PTS considerations.
However, additional information to resolve these concerns is needed to determine whether to authorize longer term operation.
17 III. LOW TEMPERATURE OVERPRES50R12ATION (LTOP)
III.1 Systems Evaluation in addition to the PTS events described above, another class of transients that could induce fracture in a brittle reactor vessel beltline are low tem-peratureoverpressure(LTOP) events.
These events could occur during plant heatup when pumps are being started and there are possibilities for the mis-alignment of valves and controls following maintenance operations. The occurrence of such events has led to requirements comprising a low setpoint reliefvalveandcontrolcircuitryasdescribedinNUREG/CR-5186,(Ref.10).
For LTOP considerations analyses are divided into two general categories:
(1) n$s(water)additioneventsand(2)energyadditionevents.
In its July 6, 1990 submittal the licensee presented analyses of such events for the Yankee Rowe plant. The analyses were based upon industry wide historical data on LTOP events from 1980-1986 adjusted by consideration of Yankee Rowe specific features. The licensee concluded that the likelihood of vessel challenges from LTOP events was very low.
The staff review in this area emphasized the applicability of historical data to Yankee, impact of Yankee specific LTOP system features; and administrative controls used to minimize human errors.
111.2 LTOP Event Frequency For LTOP analyses the licensee used the method and data described in NUREG/
CR-5186(Ref.10).
Features important for Yankee relative to the generic data base are:
Feature A: TheRHR(ShutdownCoolingSystem)atYNPSisadedicatedsystem which is different from most plants. The system is connected to the Main Coolant system through dual isolation valves. The suction to the Shutdown Cooling pump is from the #4 cold leg loop. There are two pumps and heat
n p*.
j e
18 i
exchangers for redundancy.
Sere is also a relief valve on both the suction i
and return lines for overp m ure protection.
I Feature B:
The POR'l (in the low setpoint condition) and the shutdown cooling relief valves are required to be operable by Technical Specifications whenever the plant is in the Modes 4 and 5 and the system temperature is less than 300'F. The shutdown cooling relief valves are tested when the plant is
)
operating in Mode 1 and the shutdown cooling system is required to be isolated.
The PORV is tested when the plant is in Mode 6 with the reactor head removed.
Feature C:
Plant procedures require that power be removed by locking out the breakers for the Main Coolant pumps and the Safety Injection pumps prior to being in a water solid condition.
Power is removed from SI pumps below 200*F.
4 Feature D:
The safety relief valves of the shutdown cooling system cannot be automatically isolated once the system is placed into operation because the system isolation valves do not have any automatic isolation capability.
1 Feature E:
During water solid condition operations, a dedicated operator is stationed to prevent or terminate any pressure excursion.
During operation below 300'F, 2 shutdown cooling relief valves and 1 PORY are available to mitigate LTOP events.
In this temperature range, and with no credit for human intervention during an event, the licensee estimate of vessel 1
challe.:ge event frequency (events where mitigation systems fail) is 6.5X10-5 j
per reactor year. NUREG-5186 reports a frequency of 2.5x10-3 per reactor year
]
using generic data. The difference is attributable to 2 factors:
(1)the availability of an additional relief path at Yankee relative to generic data assumptions;and(2)apowerlockoutrequirementforMCPandSIpumpsatRowe which precludes energy addition events such as were reported in the generic data base.
The staff judges that the specific featurcs of Yankee Rowe would reduce the likelihood of the vessel challenges from LTOP events in the operating range when the PORV is reset to the lower setpoint and the SDC system SAVs are
19 l
available. An event frequency of 1x10-3 per reactor year was therefore chosen as a conservative screening value to assess the importance of LTOP events in this temperature range relative to PTS events.
Between 300'F and 330'F the SDC system is isolated, and above 380'F and 450 psig the PORY is reset to 2500 psig. For all temperatures greater than 180'F a pressurizer bubble is required.
In the range of 300'F to 450'F a dedicated operator is required whose only responsibility is LTOP protection (by maintaining a 400 psi margin to the Appendix G curve).
Power is also removed from 2 of 3 safety Siection pumps at these conditions and all SI pump switches must be in pull to lock.
IntN ertent.9 (which could cause a maximum pressure of 1550 psig) would theirefore require a spurious SI signal plus failure to have the SI pumps in pull to lock.
In addition, the auto safety injection signal is blocked until 1800 psig. The licensee concluded that the most probable LTOP challengeinthisrange(Tgreaterthan300'F)isacharging/letdownmismatch.
A charging / letdown mismatch involving all 3 pumps could allow 100 gpm injec.
tion. This rate would allow 10 minutes for operator action to preclude viola-tion of the Appendix G curve in the event of a PORY failure to open.
- However, even without credit for operator action, the licensee's frequency estimate for an event that would challenge the vessel is about 1x10-5 per reactor year.
This estimate assumes a PORY failure rate of about 10'I per demand, a mismatch frequency of 10-2 per reactor year, and the fraction of time the plant would ha operating in the temperature range per year (6 hours6.944444e-5 days <br />0.00167 hours <br />9.920635e-6 weeks <br />2.283e-6 months <br /> in 600 shutdNn hours) or 10-2 per reactor year.
In view of the liceic,ee's analysis and the historical data regarding challenges to systems with a pressurizer bubble and PORY (zero events), the staff considers that the screening value of 1x10-3 per reactor year discussed above for LTOP below 300*F is also conservative in the temperature range above 300'F.
Above 380'F and 450 psig the PORY is reset to'2500 psig.
However in this range the vessel temperature is high enough that brittle fracture is of negligible concern.
l 1
20 111.3 LTOP Materials Evaluation The licensee did not discuss materials aspects of LTOP events in their reports.
The staff calculated the conditional probability of vessel fracture based on the peak pressure for the Yankee Rowe vessel using the methods set forth in Reference 11 and assuming RT is 320*F. An LTOP peak pressure in the range NDT 1000-2000 psig has a conditional probability of vessel fracture in the range 10'3 to 10-2,
111.4 LTOP Conclusion Based upon a conservative screening value of 1x10-3 per reactor year for LTOP event frequency and a conditional vessel failure probability for LTOP events of 10-2 to 10-3, the staff concludes that PTS events are bounding for brittle fracture considerations.
IV. UPPER-SHELF ENERGY EVALUATION IV.1 Background Reactor vessel beltline materials are required by Appendix G to 10 CFR Part 50 to have adequate fracture toughness. Specifically, beltlino materials are requiredtohaveCharpyupper-shelfenergy(USE)nolessthan50ft/lb throughout the life of the vessel. Otherwise, an analysis, approved by the staff, to demonstrate the existence of margins of safety against fracture equivalent to those of Appendix G of the ASME Code is required.
IV.2 Upper-Shelf Energy Events - Material Evaluation in a letter dated May 1, 1990, the staff informed the licensee of the results of analyses that indicate that the USE for the Yankee Rowe vessel could be as low as 35.5 ft/lb. The staff specified the regulatory requirements that had to be met for vessels with USE below 50.ft/lb and provided the USE evaluation criteria based on current developments of the ASME Code. At present, these criteria have only been developed for ASME Code Service Levels A and B, e.g.,
j 2.1 Normal and Upset loading conditions. The staff believes that Service Level C and D, i.e., Emergency and Faulted condi. ions, criteria are unnecessary because, except for-PTS and ATWS transients, Service Level C and D loads do not exceed level A and B loads.
PTS events are discussed above. With regard to ATWS, the staff reviewed results of ATWS analyses which the licensee has submitted in 1974. The peak pressure estimated for a loss of feedwater A1WS was estimated to be 2820 psig. Since the licensee's Charpy USE analysis assumed an RCS pressure of 3437 psig the staff concludes that ATWS events are reasonably bounded by the licensees USE analyses.
The licensee performed an USE analysis for Normal and Upset loading conditions, i.e., ASME Code Service Levels A and B, using the ASME Code criteria now in preparation. The ASME code criteria now in preparation will require margins of safety against fracture equivalent to those required by the regulations. Based on a preliminary review of the licensee's analysis, it appears that the licensee's analysis satisfies the ASME code criteria for Service Levels A and.B and provide margins of safety against fracture equivalent to those required by Appendix G of the ASME Code. The licensee also performed a low USE analysis for two of the PTS transients. The effects of low USE on crack arrest will also be considered in the PTS analysis being evaluated by the staff's contractor.
V.
CONCLUSION In order to address several NRC concerns with respect to the requirement for reactor vessel fracture toughness for protection against pressurized thermal shock events, the Yankee Rowe licensee has provided an analysis of the potential events leading to a challenge to the reactor vessel. That analysis addressed both the probability of the initiating events as well as the probability of a pre-existing crack propagating through the vessel wall. The licensee also estimated the likelihood of challenges to the vessel from low temperature overpressurization events.
As discussed above, there are a number of areas in which the staff concludes that additional safety margin or con-servatism in the analysis would be appropriate; and that additional infomation l
l
'a o
22 to fully resolve the areas of concern is needed in order to determine whether longer term operation should be authorized. Actions required of the licensee during the next operating cycle are specified below. However, in the interim, the staff concludes that reasonable assurance of the public health and safety is provided since the potential for reactor vessel failure is very unlikely.
VI.
FUTURE ACTIONS In order for the licensee to demonstrate that longer term operation can be carried out without undue risk to the public health and safety, the licensee should provide the NRC, within 60 days after restart, a detailed plan of l
action. The following elements should be included in the plan:
VI.1 Short Tem (Completed within 3 months) 1.
Peer review of YAEC 1735
" Reactor Pressure Vessil Evaluation Report for, Yankee Nuclear Power Station."
2.
Revise fluence calculations.
e VI.2 Longterm (CompletedpriortoCycle22startup)
[
1.
Develop inspection methods for the beltline welds and each beltline plate from the clad to 1 inch from the clad / steel interface to determine if +he metal contains flaws.
1 2.
Perform tests on typical Yankee Rowe base metal (0.18-0.20% Cu) to 19 2
determine the effect of irradiation (f = 1-5X10 n/cm),austenitizing temperature (1650'F-1800*F)andnickelcomposition(0.18-0.70 percent)on emb'rittlement at 500*F and 550'F irradiation temperatures.
3.
Determine composition of the circumferential weld metal in beltline by removing samples from the weld.
7
j o"
23
^
In addition, the licensee should install surveillance capsules in accelerated irradiation positions. The capsules are to include materials representing the beltline circumferential weld metal and upper and lower plates.
\\
VII.
REFERENCES 1.
F. W. Sta11mann, " Curve Fitting and Uncertainty Analysis of Charpy Impact
)
Data," USNRC NUREG/CR-2408, January 1982.
2.
G. R. Odette and G. E. Lucas, " Irradiation Einbrittlement of LWR Pressure 1
Vessel Steels," EPRI NP-6114, January 1989, 1
3.
A. L. Lowe, "An Evaluation of Linde 80 Submerged-Arc Weld Metal Charpy Data Irradiated in the HSST Program," ASTM STP-1046 Vol. 2, 1990.
t 4.
G. Robert Odette, Acting Dean, College of Engineering, UCSB "1990 Shift, Estimates for The Yankee Rowe Vessel," July 30, 1990.
S.
Maricchiolo, C., Mile 11a, P. P., and Pini, A..." Prediction of Reference f
Transition Temperature Increhse Due to Neutron Irradiation Exposure;"
Radiation Embrittlement of Nuclear Reac'cr Pressure Vessel Steels:
An International Review (Second Volume), ASTM STP-909, L. E. Steele, Ed.,
t American Society for Testing and Materials, Philadelphia, 1986,iages96-105.
6.
B&W Owners Group Report BAW-1799, "B&W 177-FA Reactor Vessel Beltline Weld Chemistry Study," July 1983.
7.
W. Marshall, An Assessment of the Integrity ?f PWR Pressure Vessels, Uniteif f.ingdom Atomic Energy Authority, October 1976, d.
Letter from H. Kaplan and J. O'Neil, " Yankee Rowe Feeder-Ultrasonic Examination of Pressurizer and Ree: tor Vessel," August 15, 1990.
6 20 l
9.
F. A. Simonen, et al., "VISbil - A Computer Code for Predicting the Probability of Reactor Vessel Failure," Battelle Pacific Northwest
)
Laboratories, USNRC Report NUREG/CR-4486, April 1986.
i 10.
B. F. Gore, et al., PNL, "Value-Impact Analysis of Generic Is!,ue 94,
" Additional Low Temperature Overpressure Protection for Light Water Reactors," NUREG/CR-5186, November 1988 l
11.
C. Y. Cheng, Chief. EMCD memorandum to Robert C. Jones, Chief, SRX8,
" Conditional Probability of Vessel Fracture from LTOP Events," August 9, j
- 1990, i-12.
FAX, Jane Grant, Yankee, to Pat Sears, NRR, 8/8/90 - Material Properties j
Answert to Questions at 8/7/90 Meeting.
L J
13.
FAX, Jane Grant to Pat Sears 8/10/90 - Answers to Quastions at 8/7/90 L
Heccing.
14.
FAX, Jane Grant, Yankee, to Pat Sears, NRR, 8/10/90 - Answers to Questions at 8/7/90 Meeting.
l-15.
FAX, Jane Grant, Yankee, to Pat Sears, NRR, 8/10/90 - Answers to Questions at 8/7/90 Meeting.
16.
FAX, Jane Creat, Yankee, to Pat Sears, NRR, 8/14/90 - Ansurs to Questions at 8/7/90 Meeting, t
l 17.
FAX, Jane Grant, Yankee, to Pat Sears, NRR, 8/14/90 - Answers to L
Questions at 8/7/90 Meeting.
- 18. Yankee letter dated 8/3/90 - Pear Review of Reactor Pressure Vessel Evaluation.
l-
- 19. Yankee letter dated 8/2/90 - PfS Sensitivity Study.
's o'*
25
- 20. FAX, Jane Grant Yankee, to Pat Sears, NRR, 8/14/90 - Updated Table 5.7 of 7/5/90 submittal.
21.
FAX, Jane Grant, Yankee, to Pat Sears, NRR, 8/17/90 - Answers to Questions by G. Kelly, NRR at 8/16/90 Telecon.
- 22. FAX, Jane Grant, Yankee to Pat Sears, NRR, 8/17/90 - Answers to Questions by G. Kelly, NRR at 8/16/90 Telecon.
23.
FAX, Jane Grant, Yankee, to Pat Sears, NRR, 8/27/90 - Fracture Mechanics Results.
?
i
- O o
b;.*
TABLE I LICENSEE AND STAFF ESTIMATES OF REFERENCE TEMPERATURE, RT FOR THE YNPS BELTLINE MATERIALS IN 1990 NDT YNPS Increase in Ref.
Beltline Unirradiated Temp. Resulting Ref. Temp.,
Material Ref. Temp.(*F) fromIrrad.(*F)
RT in1990('F)
NDT Staff Licensee Staff Licensee Staff Licensee Estimate Estimate Estimate Estimate Estinate - Es$1 mate Upper Plate 30 10 245 180 275 190 Lower Plate 30 10 325 173 355 183 Axial Welds Y0 10 216 131 226 141 Circum.
ferential Weld 10 10 320 219 330 370 229
.c O
e
.O TABLE 2 TIMISTRY FACTOR FOR 8ASE METAL, 'F C
Nickel, Wt e W.
0 0.20 0 40 0 to 0.80 1.00 1.20 0
20 20 20 20 20 20 20 0.01 20 20 20 20 20 20 20 0.02 20 20 20 20 20 20 20 0.03 20 20 20 20 20 20 20 0.04 22 26 26 26 26 26 26 O.0$
23 31 31 31 31 31 31 0.06 28 37 37 37 37 37 37 0.07 31 43 44 44 44 44 44 0.08 34 48
$1
$1
$1
$1
$1
-0.09 37
$3
$8
$8
$8
$8
$8 0.10 41
$8 65 65 67 67 67 0.11 45 oJ 72 74 71 77 77 0.12-49 67 79 83 86 86 86 C.13
$3 71 85 91 96 0.14
$7 7$
91 100 10$
106 106 0.15 61 80 99 110 115 117 111 0.16 65 M
104 lit 123 12$
12$
0.17 69 88 110 127 132 135 135 0.18 73 92 11$
134 141 144 144 0.19 78 97 120 142 ISO 154 154 0.20 82 102
' 12$
149 159 164 165 0.21 86 107.
129 15$
167 172
' 174 0.22 91 112 134 161 176 181 IM 0.23 95 117
. 138 167 IM 190 194 0.24 100 121 143 172 191 199 204 0.25 -
104 126 148 176 199 208 214 0.26 109 130 ISI 180 20$
216 221 0.27 114 134 1$$
IM 211 22$
230 0.28 119 138 160 187 216 233 239 O.29 124 142 -
=164 191 221 241 248 0.30 129 146 167 194 225 249 257 0.31 134 151 172 198 228 2$$
266 0.32 139 IS$
17$
202 231 260 274 0.33 144 160 180 20$
234 264-282 0.34 149 164 iM 209 238 268 290 0.3$ -
133 168 I87 212 241 272 298 0.36 158 173 191 216 24$
275 303 0.37 162 177 196 220 248 278 308-0.38 166 182 200 223 250 281 313 0.39 171 185 203 227 254 285 317 0.40 175 189 207 231 257 288 320 i
.r-9
. o
,D
- o
)
TABLE 3 CDNPARISON OF REFERENCE TEMPERATURES ESTIMATED BY THE STAFF AND YALUES USED BY THE LICENSEE IN ITS SENSITIVITY STUDY MEAN VALUE REFERENCE-TEMPERATURE USED IN REFERENCE TEMPERATURE LICENSEE"S SENSITIVITY-MATERIAL ESTIMATED BY STAFF STUDY UPPER PLATE 275 280 t
LOWER PLATE 355 325,
AXIAL WELD 226 222 CIRCUMFERENTIAL WELD 330-370 360 I
~~
4 L[.,*js o
I 500 i
iiis i i ii i
i i
iiiin l
1 maaW VAtut'l#REMrk Aar.
m e,roncia. wee
-zD@,E
.._300f4t w
a r +
.. :. e p,.,0
-ge
=
+
m g
100
+,
..f
?
2 50 O.1 1
10 a:(10'S n/cm )
8 Cu/NI/P Prod. Form Ref.
Cu/Ni/P Prod. Form Ref.
i.
+o 0.19/0.55/0.011 B
9
$0.17/0.12/0.016 W
7
@ 0.23/0.56/0.013 '
W 9
e 0.20/0,18/0.011 B
10
$ 0.30/0,56/0.021 W
9 e 0.26/0.28/0.012 B
10
@ 0.26/0,56/0.020 W
9 m 0.21/0,54/0.016 W
8
@ 0.32/0.67/0.017 W
9 s 0,35/0.66/0.014 W
8
@ 0.19/0.55/0.011 W
9
@ 0.22/0.60/0.015 W
8 A 0.15/0.09/0.025 B
7
@ 0.42/0.60/0.018 W
8 6 0.19/0.07/0.017 B
7
@ 0.40/0.59/0.011 W
8
+ 0.24/0.25/0.024 B
7 e 0.18/0,18/0.011 B
1
- a 0.21/0,17/0.033 B
7 (YR Plate) l Tigure 4 Shift data for 500110*Firradiations versus fluence and preliminary recommended trend curve.
049447D FIGURE 1
..