ML20106E399
| ML20106E399 | |
| Person / Time | |
|---|---|
| Site: | McGuire, Mcguire  |
| Issue date: | 10/29/1992 |
| From: | Mcmeekin T DUKE POWER CO. |
| To: | NRC OFFICE OF INFORMATION RESOURCES MANAGEMENT (IRM) |
| References | |
| NUDOCS 9211060429 | |
| Download: ML20106E399 (20) | |
Text
1 11-I Dube 1%uvr Coo.. pony
. T C Atcyavv
= hicCude Nadrat Generation ikperment -
lice President I?iW)flagers ferry Road (hfGulA)
(70118754M ll:intersnue, NC Mi&SM5 (704)$754809 ht DUKE POWER October 29, 1992 Document Control Desk U.S.
liuclear Regulatory Corranission Washington, D.C.
20555
Subject:
McGuire Nuclear Station Docket Nos:
50-369 and 370
Dear Sir:
On Ser mber 23, 1992, representatives from Duke Power Company and Westingnouse Electrjc Corporation met with the NRC Staff at the NRC's of fice in Rockville, Maryland. The purpose of the meeting _was to present the results of the inspections and analyses that_had been performed on the McGuire Units 1 and 2 steom generators. As a x
fo] low up to this meeting and-subsequent conference calls,.
additional evaluations and analyses were perf ormed _. in order to characterize-the probability of occurrence of large differential
- pressure events and the likelihood of a tube burst concurrent with-such events. A discussion of these evaluations -and analyses are provided in the attached pages.
It' is concluded from the information presentea herein' and on September 23, 1992, that operation of McGuire Units 1 and 2 until the end of their respecti e fuel cycles does not present an undue
.rlsk to the nealth and safety of the public.
E i
i Very truly yours,
/
s a 3/
b T.
C.
McMeekin 9211060429 921029 PDR ADOCK 05000369
',{j-l P
pop n., nn,
l U.S.
Nuclear Regulatory Commission October 29, 1992 Page 2 xc:
Mr.
S.D.
Ebneter Administrator, Region 11 U.S.
Nuclear Regulatory Commission 101 Marietta St.,
NW, Sutto 2900 Atlanta, Ga.
30323 Mr. Tim Reed U.S.
Nuclear Regulatory Commission Office of Nuclear Reactor Regulation Mail Stop 141125, OWFN Washington, D.C.
20555 Mr.
P.K.
VanDoorn NRC Resident Inspector McGuire Nuclear Station
End of Cycle Steam Generator Tube Integrity McGuire Units 1 Introduction Information is provided herein for the support of the continued operation of McGuire Units 1 & 2 through the end of their respective fuel cycles, representing periods of operation of 8.A and 12 months respectively, An integrity analysis of the tubes was performed based on expected conservative assumptions relative to the configuration of cracks which may be postulated to exist in a very limited number of tubes in the steam generators (SG). A probabilistic analysis was also performed.
2 Background
Results of tube integrity evaluations were presented to the NRC in a meeting held on September 23,1992,in Rockville, Maryland. The information presented supported the operation of Units 1 & 2 until the end of their respective fuel cycles. This was based on 100% detection of free-span cracking at a 50% penetration level, a 2.5% per month groMh rate, and an allowable depth of 81% based on the limits of RG 1.121 (draft).
Subser ant discussions with the NRC were held via telephone on P ber 9, and October 15, 1992. Of specific interest was the probability of oce
.nce of largo differential pressure (AP) events, and the associated likelihood or a tube burst concurrent with such events. The second discussion was held to appraise the NRC of the results of performing additional analyses. Information was provided on the results of probability and risk assessment (PRA) analyses for the McGuire Units, see Attach-ment 1, and the results of statistical simulations (Monte Carlo) of the integrity of the steam generator (SG) tubes. The Monte Carlo analyses were based on statistically.
sampling the crack depths, crack growth rates, end of cycle (EOC) crack morphologies, tube material properties, and the results of burst testing.
The purpose of this document.is to provide a summary review of the information presented and the analyses performed.
i l
l 3 Deterministic Analysis of EOC Tube Integrity The deterministic analysis of tube integrity was based on postulating the existence of j
cracks in the SG tubes at a detection threshold depth. The detection threshold is that l
depth for which a very high level of confidence exists that such cracks would be L
detected by eddy current examination (ECT). This was followed by establishing a
(
NRCJtFK.A20
-- 1 --
octoirr 20, Ms02
s conservative value for crack growth rate into the wall of the tube, projecting that growth to the end of the fuel cycle, and then determining the burst pressure.
3.1 Detection Threshold i
Information was presented based on the non destructive and destructive examination of eleven tubes which had been removed from the SO's of the two McGuire Units. The results demonstrated that 100% of cracks were detected when the depth was 50% or i
greater. For cracks in the range of 40% to 49% deep the detection rate was 42%, and j
for cracks in the range of 26% to 39% deep the detection range was 27%.
3.2 Crack Growth Rate For the determination of growth rate the available data were segregated into three j
categories consisting of Units 1 & 2 known grooved and cracked tubes, Units 1 & 2 known cracked tubes, and Units 1 & 2 cracked tubes based on the ECT evaluations that had been performed. Fr r most of the cracked tubes very little information was available in terms of previous ECT magnitudes upon which to base previous depths.
Information was available from expert ECT analyst review of previous outage data to determine when a crack's initial presence was indicated by the data. For the cracks found in the pulled tubes a scheme for estimating crack depth at a previous cycle was adopted based on the final depth of the indication. Specifically,if the indication had a final depth ofless than 50% it was assumed to be of zero depth at the beginning of the cycle prior to its' initial presence being apparent. If the final depth was greater than 50% it was assumed to be at one-half ofits' final depth at the beginning of the previous cycle. For the one month of operation of Unit 1 in December,1991 and January,1992 growth rates were not calculated. Over such a short time period such calculations would be subject to significant error since the length of time in the denominator of the rate determination would be so small. This short time period was conaidered, however, as an end point from the beginning of a previous cycle.
The average growth rate from the confirmed grooved and cracked tubes was found to be in the range of 2.3%/mo to 2.5%/mo with a standard deviation (SD) of 0.5%/mo.
The maximum value was 3.1%/mo for one cycle of operation. For all confirmed cracks the average growth rate was found to be from 2.3%/mo to 2.4%/mo with a SD of 0.7%/mo. Inclusion of the ECT results with the confirmed cracks data base resulted in a mean of 2%/mo with a SD of about 1.6%/mo. For the deterministic analysis it was decided to use the largest average crack growth rate since lower bound material properties would be used in determining the burst pressure.
A comparison of the value selected for the growth rate with available published data, extrapolated to the cold leg operating temperaturr, was made to verify that the rate was not in disagreement with test data. The selected value was found S lie in the middle of the extrapolated band.
NRe_RFK.A20
- Octater 20.1992
3.3 End of Cycle Crack Morphology
^
Initial analysis concentrated on a single crack morphology, i.e., that present in section 5 of the tube removed from R18C5 of SG A of McGuire Unit 2. - The selection was based on the fact that this was the largest crack that had not been called during the ECT of the tubes prior to removal of the tube for destructive examination. The crack was found to have a maximum depth of 73%, an average depth 'of 54%, and was 1.1" long. A series of tests were performed to determine the burst pressure of the crack profile as a function of depth. For a tube vith the lower bound material properties it was found that the maximum depth of the crack could be 81.5% and still retain a factor of safety of three relative to burst during normal operation and a factor of 1.4 -
relative to postulated accident conditions. It was also found that the maximum depth '
that would result in a burst pressure egnal to a postulated SLB differential pressure (2560 psi was used for conservatism) was 89.7%.
A second series of tests was subsequently performed using the morphology of the crack found in section 10 of the same tube. This crack had a maximum depth of 100% and was found to be 1.4" long. The purpose of the second testing program was to determine the sensitivity of the results of the evaluation to the length of the crack.
For the R18C5-10 crack morphology the burst depth values for normal operation and SLB were found to be 79.8% and 89.2% respectively, t
3.4 Lower Bound of Material Properties Burst pressure determinations are usually performed considering a 95%/95% lower tolerance limit on the flow stress (one-half of the sum of the yield stress and the ultimate stress) for a general population of tubes encompassing the heats of tubing used in the SG's for several plants. However,information relative to heats used in the i
fabrication of the McGuire Units' SG's was available. Using this information, a 98.6%
lower bound for the flow stress of the tubes in the McGuire Units was selected for the burst pressure evaluations. The value was adjusted to a cold leg operating tempera-ture value of 140.3 ksi based on the results of an extensive testing data base-comparing values at room temperature und at 650 F.
3.5 Margin for Operation To determine the margin for operation the burst pressure of the R18C5-5 and R18C5-10 type cracks was plotted as a function of time of operation considering the growth rate reported previously. Using the results of the burst testing, a tube with lower -
bound material properties, an initial crack depth of 49%, and a crack growth rate of 2.5%/mo would be expc' d to have a burst pressure greater than the RG 1.121(draft) limit for 13 months ofoperation for the R18C5-5 type of crack morphology, and slightly-more than 12 months for the R18C5-10 type morphology. For both types of cracks the burst pressure would not be expected to be less than the postulated SLB differential pressure for a period in excess of 16 months. Neither type of crack would be expected
~
to result in a r"ohire at t.ormal operating conditions at any depth, e.g., through wall.
N RC.RFK.A20
-- 3 --
october 20.1992
Additional considerations indicated that the growth rate would have to be on the order of 3.4"c/mo for a R18C5-5 type of crack for the tube to have a burst pressure less that the postulated SLB pressure at the end of12 months of operation. This growth value is greater than the maximum value considered for the determination of the crack growth rate to be used in the analysis.
4 EOC Probabilistic Analysis (Monte Carlo Simulation) 4.1 Beginning of Cycle Crack Depth For each simulated tube a beginning of cycle (BOC) crack depth was selected as the sum of the mean BOC crack depth plus the product of the standard deviation (SD) of the crack depths and a random variate distributed as a Student's t-distribution for the number of degrees of freedom (doO used in determining the standard deviation.
The mean and standard deviation for cracks less than 50% deep (the detection threshold) was determined from the crack population present in the tubes removed from the McGuire Units. Examination of the data shows them to be normally distributed with a mean depth of ~39% and having a SD of ~6%. A total of18 cracks were found at a depth ofless than 50%, thus the t variate was based on 17 dofs.
The performance of the simulation included rules for eliminating mathematically possible, but unrealistic, random initial depths ofless than zero.
4.2 Crack Growth Rates The same data base for the evaluation of crack growth rates was available for the probabilistic analysis as for the deterministic analysis. Once a mean and standard deviation are determined the crack growth rate for each simulated tube was determined as the sum of the mean and the product of a Student's t-distribution random variate and the standard deviation. The t-variate was based on the degree of freedom (doO used for the determination of the standard deviation.
The mean and standard deviation from the largest amount of available data were selected for inclusion in the simulations. Approximately forty-one (41) data points from the non-destructive and destruction examinations of tubes from both units were availuble for use in the simulation. The data were subjected to a robust estimation of location test with the result that four (4) data points were identified as outliers based on having residual to dispersion scale ratios ranging from 3.15 to 13. The associated 4
probability of occurrence of such values on a random basis ranges from 8.210 to 8 10 The three largest deviations occurred for values based on only three months 4
of operation. Determination of mean and standard deviation values for use in the analysis was therefore performed for two cases: the first being with the exclusion of the outliers, and the second being with the exclusion of all values for which the operating time was on the order of three months. For the first evaluation the mean unc_ar u m
-- 4 --
o arm.um
growth rate was found to be 1.7% per month with a SD of 1.1% per month. For the second case the mean growth rate was found to be 1.8% per month with a SD of 1.2%
per month. Both rates were simulated in independent Monte Carlo evaluations.
4.3 End of Cycle Crack Depth The EOC crack depth was determined simply as the sum of the BOC crack depth plus the product of the growth rate and the number of months of operation being simulated.
4.4 Crack Morphology Crack morphologies for the Monte Carlo analyses were the same as considered for the deterministic analysis, i.e., Unit 2, SG A, RISC sections 5 and 10. The crack in R18C5-5 was found to be 1.1"long, and in the R18C510 the crack was 1.4"long. Both configurations are considered to be representative of the more severe morphologies that can occur. For analysis purposes the basic profile of each crack was considered to orly vary in depth, thus the EOC profiles for the simulated tubes matches significant crack profiles observed on the removed tubes. The Monte Carlo analysis considered the probability of occurrence of each morphology to be equally likely. Thus, approximately half of the simulated cracks were 1.1"long and the other half were 1.4" long.
4.5 Material Properties The calculation of a burst pressure for a tube utilizes the sum of the yield stress and the ultimate tensile stress (twice the flow stress). The material properties simulated were based on the actual plant distribution of mill reported properties. The yield stress and ultimate tensile stress for about 90% to 95% of the tubes in each generator was available based on a data base of the material heat used for each tube. The distribution of mill reported properties had a mean of162.7 ksi with a SD of 6.1 ksi.
Random devues based on Student's t-distribution based on 100 dof's were used for the simulations. The actual number of different heats is more on the order of 200.
The sampling also considered the actual material properties of the tubes removed from the McGuire Unit's SG's. It was found that all of the removed tubes exhibited material properties in excess of those reported on the mill data sheets. The variation was found to have a mean of 1.07 with a SD of 0.06. This distribution was also simulated. Thus, the tubes simulated would on average have a higher value for the sum of the yield plus ultimate than the mill reported values. Given the magnitude of the SD approximately 17% of the tubes would also be considered to have a flow strength less than the mill reported values.
Each distribution was sampled independently to arrive nt a final value of the flow stress to be used for the burst pressure calculation.
A final adjustment to the sum of the yield plus ultimate properties was made to determine a value corresponding to the operating temperature of the SG. Previous testing on a large number of tube specimens has shown that the 3/4" diameter mill NRC.RFK.A20
-- 5 --
oa-20.um
4 amtealed tubing has a decrease in flow strength of 7% in going from room temperature to hot leg operating conditions. For cold leg operating conditions the reduction is 0%.
Since the SD of material properties at cold leg conditions is generally less than the SD i
i at room temperature conditions no simulation of any variation of the strength 4 -
reduction factor was necessary.
l-i 4.0 Burst Pressure Determination i
i t
i A series of tube burst tests was performed for each of the crack morphologies considered. In general the tests considered maxhuum depths ranging from 50% to 100% ihrough the tube wall. There is essentially no change in bmst resistance for these types of morphalogies once the crack depth reaches 93% ibrough the tube wall.
This b due to the burst mode changing from plastic instability to crack opening displacement (COD) dominated fracture. For each crack morphology a second order regression equation was litted to the burst data for cracks up to 93% deep. Hriving the crack depth and the flow strength for the material a determination of the burst prassare was made. More extensive burst testing programs have shown that the SD l
af hurst pressure about the burst curve is ~8%. For the Monte Carlo simulations n-value of 24% was selected as representing three SD's. Since the total number of hurst tests used to establish the SD value was high, variation about the burst curve.ms simulated using random Normal deviatea. This results in a difference in the final burst pressure of nbout 0.1% relative to the use of variates based on Student's t distri-bution.
4.7 CompaHson of Hurst Pressure to Steam Line lheak Pressuro l
The calculnted burst press"re for each simulated tube was then compared to the differentit '.nssure thm would be expected dur: w a steam line break (SLB). Specific analyses penbrmed established that two variata c s of SLB could be considered. The-first terminating in a differential pressure of 2335 psi, and the second terminating in a differential pressure of 2485 psi, see Attachment 1, Table 2. Burst pressure sorting from the Monte Carlo analyses indicate that the probability of hurst is slightly higher for the higher pressure, however the frequency of occurrence of the second event is several orders of magnitude less than for the first event. Thus, the joint probability L
of occurrence of burst coupled with a SLB evem is significantly lower for the higher L
pressure event, and only the lower pressurc event probabilitif a are considered Fignilicant for evaluation.
For the growth rate and SD cx ponduq the exclusion of outlim from the data base, the probability of a single tube having a burst pressure greater that the SLB differential pressure (2d35 psi) at the end of 12 months of operation was found to be -
99.9E For the growth rate and SD corresponding to the use of all data for which the l
01)erating time period was in excesFor ~3 months the corresponding probability was found to be 99.5%.
[
NW,.RFKA20
-.0y-octAf 20.1W2
.ns~n.
...e in--c.,,,
-w,~n,-,,-,,--o-,,w a
.,.rw
-wa,,n,.-.-ww-.,--
,,'.m J-_,,.,-..,4
,e,v.vn,
,.w,
-s u m
,+ -
s-n.r w w nm r-w r-w
.-,n-m,. -, ->-
w-----
During the last outages fbr both units a total of ten tubes (five per unit) were plugged for. free span indications believed to be similar to those found on the tubes removed from Units 1 & 2, i.e.,1(18C25 from SG H of Unit 1, H46047 from SG D of Unit 1, and 1118C5 from SG A of Unit 2.
A like number of tubes remaining fbilowing the inspection outages may be ostulated. Po:.Jating the existence of two susceptible tribes in one (;enerator, the probability of them both having burst pressures exceeding the SLU differential pressure at the end of the operating cycle would be 99.8% For three postulated tubes this value becomes 99.6%. Postulating all ten tubes to be in a single SG, which w mid have a probability of 7.510*, results in a value of slightly less than 99% Using the growth rato and SD based on all data for operating times in excess of ~3 months results in the above probabilities being 99.8%,98.59, and 95.0%
respectively.
L 5 Conclusions A deterministic evaluation of the integrity of the McGuiro Units' SG tubes, paragraph 3., was performed which concluded that tube cracking would not be expected to progress to a depth such that the burst pressure of the cracked tube (s) would be less than the limits of itG 1.121 (draft) fbr twelve months of operation of the units. In i
addition, it would be expected that the burst pre:sure would be greater than the differential pre %ure associated with a SLB if the crack growth rate was significantly in excess of that used in the analysis (3.4G/mo versus 2.59hno).
A Pila was performed to quantify the likelihood of high AP events, see Attachment 1.
A probabilistic analysis of the integrity of the SG tubes was also performed, paragraph 1., which concluded that tho probability of tube burst during a postulated SLB event is unlikely through the end of the units' respective operating cycles. Thus, we conclude that. the combined probability of a tube burst and a SLB event is extremely unlikely.
It is noted that the end of the current cycle for Unit I will occur at approximotely 8.5 months of operation following the last inspection outage. Unit 1 inspection results will be avadable after 10 months of unit 2 operation, and alternate actions could be taken k
if deemed necessary.
1 It is concluded that the information provided herein supports the continued operation of McGuire Units 1 & 2 through the end of their respective fuel cycles.
NRc_RFK A20
-- 7 --
oM a 0,Um
l MIACHIGNi 1 Probabilistic itisk Persnective of Steam Generator Tebe j
1(upture af TicGuire Nuclear Stitii ni t
l iM01t2.
'I he McGuire PRA study done in response to Gl. 88 20 estimated the public health risk of tube rupture sequences The conclusion of the McGuite PR A was that the nsk was very low because of the multiple and diverse means of terminating the accident prior to core damage NUREG-0844 also analyzed the public health risk of SGTR and concluded that, " risk from steam ger'erator tube rupture events is not a significant contributor to total risk at a gisen site, nor to the total risk ;o which the general public is routinely exposed " Although the conclusion of both these studies was the same, the actual results show that there are differences m the calculated results. For example, the NURI.G estimated the total core damage frequency to be appiox. 4E 6/RY while the McGuire PRA estimate was approx IE.8/RY A couparison of the McGuise PRA to the NUREG reveted that there were many similarities but also some differences between the two reports. A summary of this comparison is included in.1.
One of the major differences between the NURl!G and the McGuire PRA is that the NURl!G evaluated the potential fin transients to induce a tube rupture, while the McGuire PRA study did not evaluate this possibility. To determine the elTect of this type sequence at McGuire, a screening study was performed to determine if the NUREG analysis hounds the McGuire plant I he conclusion of this review is that the likelihood of a transient induced challenge to the SG tubes for McGuire is either comre. raoie to or less likely than the NUREG-0844 anslysis Additionally, the difierential pressures for these transients would be less severe for the McGuire plant.
The frequency of transient induced it.be challen;;es for McGuire and resulting pressuret are discussec in moie detail below.
Another d ITerence between the NUREG and the McGuire PRA is the likelihood that a tube ruptuce sequence willlead to core.nelt. The NUREG estimated that for a single tube ruptua with a stuck open main steam safety vdve, the probability of core melt was lE-3. For a similar sequence the McGuire FRA estimated a probability of IE.6.
t.
The NUREG appears to base on!y given credit for depressurizing the primary system using the other steam generators At McGuire there is also the capability of opening the primary system PORVr as a means of reducing pressure. Both secondary depressurization and pr; mary depressurization are veiy reliable actions given the long time available before core damage would occur. The NUREG appears to use conservative estimates ihr the likelihood that a tube rupture sequence will icau to core damage 1
t~
Conclusinn - The following conclusions can be drawn from the combination of the McGuire PP A and NUREG-0844 analysis of SG1R risk-1)
Considering the design basis transients and accidents, the frequen:y of events which are expected to indt.cc large stresses (frorn DP loadings) is small (1E-4/RY for DP= 2485psid and lE-3 for DP= 2335 psid). It should also be noted that the consequences of a high energy line break inside containment with a 1
tube ruptute are expected to be much less severe thar. if the break is outside containment.
2)
For anticipated transients with a frequency of 0.1/RY - 2.0 /RY, the pressure loadings are in the range of normal operating pressure loadings.
3)
For McGuire, the plant features of 3 PORVs and 3 safety relief valves on the pressuriter limit the severity of pressure loadings.
4)
For accident sequence'. leading to a core melt condition, both the frequency and risk from a Cngle SGTR event are very small 5)
The consequence results of NUREG-0844 have been compared to resuhs for McGuire for two sequences. The fission product release fractions and consequence results are a factor of 5 to 10 less th tn those presented in ;he NUREG, when water is asaumed available to cover the rupture as was done for the NURiiG analysis The risk contribution from these sequences is relatively srnall fraction of the risk calculated in the McGuire IPE.
6)
For the complete spectrum of SGTR related core melt sequences, NUREG.
0844 estimated a latent fatality iisk of approx. 2E-3/RY. This is equivalent to approx. 25 person-rem / RY exposure to the public. If this value is reduced by the factor ofi-10 for more realistic values for McGuire, the risk is 2.5 5 pe son remi RY exposure to the public. In contrast, an additional steam l
l generator tube inspection would result in plant personnel exposure of approx.
30 person-rem. The SGTR risk is a conservatively calculated number while t%
worker exposure during the outage is virtually certain.
(
Annivsis of Scournees Which llave the Potential to Sublect the Steam Generator l
Libes to 1.arce Differential Pressures l
l SGTRs can potentially occur as a result of plant transients or accidents when loadings -
i on the steam generator tubes are increased above normal operating loads. These
!~
transients car cause significant pressure differences across the SG tubes by either decreasing the secondary pressur.e, increasing the prinury pressure, or a combination l
of both. NUREG 0844 identified four types of erents which have the potential to l
create this type of transient. These include ATWS, steamline or feedline breaks, LOCAs and transients which result in a stuck oper secondary safety valve. An 2
'I h
4 attempt was inade 10 estimate the frequency of these events and the resulting pressure differentials for the McGuire Nuclear Stelion.
A TWS - The ATWS sequences analyzed in NUREG 0844 are very similar to the analysis performed for the McGuire PRA. The sequences consist of an initiating tranwent followed by a failure of the reactor to SCR/ A The McGuire PRA frequene) of the initiating trensient (1.5/RY)is very similar to that of the NUREG-0844 analysis (1/RY) Also, the probability that the reactor will fail to SCRAM is very sim:lar: 1.5E 5 for McGuire, compared to 3E-5 for the NUREG analysis. The pressures resulting from this sequence i
depend on th' -frect of the rooderator temperature coeflicient (MTC). Since the MTC changes with time during a fuel cycle, the probability of achieving dilTerer,t pressure levels depends on.he fraction of time that the MTC is at ditTerent levels. For the NUREG analysis, the MTC was divided imo threc categories which could be called favorabic (0.5), moderate (0.49), and unfavorable (0.01) The McGuire PRA analysis only looked at a single MTC category which cmM be considered unfavorable and which was considered to exist 5% of the tinw. For the purpose of compariscn, the McGuire analysis has been modified using the NUREG values for the pressure associated with the other two categories. The resuhs of this comparison are presented in the Table
- 1. As can be seen, the estimated frequency of challenge to the steam generator tubes for McGuire from an ATWS sequence is very similar to ihat calcukted for the NUREG-0844 analysis.
Secondary Line Ilreak The secondary line break sequence analyzed by the NUREG was a steamline break resulting in a 2600 psi differential pressure across the SG tubes. The McGuire PRA considers two cecondary system initiating events which have the potential to cause significant difTerential pressures. These are secondary system lugh energy pipe breaks inside containment and a steamline break outside containment.
The resuhing pressure difTerential foi these events depends on the actions taken by the operators in response to the accident. For both events, primary system chrinkage caustd by the overcooling of the primary system willinitiate safety injection. The ECCS will refill and repressurize the primaly system at the same time as secondary system pressure is dccreasing. Emergency procedures require that safety injection be terminated as soon as pressurizer level and -
subcooling are regained. The resulting pressure differential across the SG tubes will depend on when the operators accomplish this action.
If the operators fail to terminate ECCS prior to the pressurizer going solid, the pressurizer PORVs should prevent the pressure differential from exceeding 2335 psid. The differential pressure across the tubes could only reach the maximum pressure of 2485 psid if the operators fail to. terminate ECCS and all-three PORVs fail to relieve pressure. For the steam line break outside-containment the reliability of the PORVs is expected to be very good.
Ilowever, the secondary line break inside containment will cause containmem f
3 aw s
aro.a, w
-., - ~,
-g-w,gv v
>g ".y v
i i
isolation which cuts offinstniment air to the POItW. The operators are instructed by the procedures to reestablish air to the PORW, but the possibility that this action will not be performed reduces somewhat the reliability of the PORW.
The event sequences described above are presented in the attached figures, and a comparison to the NUREG analysis is presented in the attached table The analysir. for McGuire rhows that when operator action is considered, the frequency of a secondary line break resulting in very large differential pressures across the SG tubes is reduced by an order of magnitude.
LOCA NUREG 0844 considered primary system LOCA to be a potential sequence which could result in a large pressure ditTerential across the SG tubes.17or this sequence, the pressure would tend to c.ause tube collapse rath:r thar burst. The concern identified in the NUREG was related to the effect of steam binding from the tube rupture causing delay in core retlood. This would primarily be a peak cladding temperature concern. Several significant conservatisms were included in the analysis and resulted in an event sequence frequency of 2.$E-7/RY, Comparison to McGuire does not seem to bc necessary for this case.
Stuck Open f afety Valve - NUREG+0844 concluded that the frequency of stuck open safety valve sequences for Westinghouse type reactors was low enough that it was bounded by the steam line break analysis. Potential transient initiators were reviewed for McGuire to determine if this statement could be supported. Two sequences were identified that could challenge the safeties with sufficient frequency to be a concern. These initiators are loss of olTsite power (LOOP) and loss ofinstrument air which were estimated by the McGuire PRA to have frequencies of 0.07/RY and 0.33/RY respectively. The probability that a safety valve will stick open upon demand is in the range of IE 2 so that the frequency of this sequence is on the order of 4E-3/RY.
The plant response following the stuck open safety valve would be very similar to the secondary line break inside containment sequence discussed above This is because both the LOOP and the loss of air would resuh in failure of air to the pressurizer PORVs and would require the opera rs to align nitrogen from the accumulators 'o the PORVs. This action is considered to be as reliable as l
the one for restoring instrument air following the secondary line break inside containment.
t L
Although the stuck open safety valve sequence may have a frequency l
comparable to the secon'dary line break sequence, it is less likely to result in large differential pressures across the SG tubes. The rate of pressure drop on i
the secondary side of the tubes for a stuck open safety would be slower than
. for a pipe break, such that even if the primary systern goes water solid the l
l-4 p
l r
w
- srcondary pressure will probably be greater than atmospheric pressure. Since the primary system pressure can not exceed the pressurizer safety vahes' set point presswe of 2485 psig, the differential pressure would be 2485 less the g
secondary pressure Since the likelihood of this sequence is comparable to the secondary line break frequency and the differential pressures are expeued to be less, we agree with the NUREG conclusion that the stuck open secondary safety valve sequence is bounded by the secor:dary line break analysis.
i Additional Transients - In addition to the transient discussed above, the McGuire FSAR Chapter 15 analysis was revie.ved to determino if any other transients had the potential to cause significant differential pressures across the i
SG tubes. Table 3 presents the estimated piessure loads and frequency of the-Chapter 15 transients As can be seen, the sequences discussed above bound all other transients by either pressures or frequencies.
8 i
The analyses presented here have c emonstrated that the frequency of transients which cause larec differential pressures across the McGuire steam generator tubes is,:ither lower than or comparable to the analyses presented in NUIEG 0844.
i v
I h
i-l r
- j l
l=
i-1 i
l 1
+
l';
.5 y
e.
ta
- v'w m-
++ w"e r-t'
+-+w w
s
- +w e r e mu
-se v
i
+-
p b
iry
.i=,g--
m e
v.Mre v-e*+T-t r-t r
Table 1 - Seaaences Which Mav Result in Substantial Pressure Differentials Across the Steam Generator Tubes (ATWS)
{ NUREG-NUREG-lNUREG-McGuire l McGuire McGuire 1
0844 0344 3&44-Sequencei
' Sequence 2 Sequence 3 SecuenceI
, Sequence 2 Sequenc= 3 Sequence Loss f Main Loss of Main ! Loss of Main LOFW or i LORV or LOFW or
! Turbine Trip Turbine Trip I
Initiator Feedwater =
Turbine Trip ! = 1.5/RY 1/RY 1/RY 1/RY
= 1.5/R'(
= L5/RY Action 1 Fa:Iure to l Failure to Failure tc
' Failure to l Failure to Failure to SCRAM = -
SCRAM =
SCRAM =
SCRAM =
l SCRAM =
SCRAM =
3E 3E-5 3E-5 1.5E-5 i 1.5E-5 L5E-5 I
Unfavorable Action 2 Favorable -
Moderate Unfavorable Favorable l Moderat:
MTC = 0.5 MTC = 0.49 MTC = 0.01 MTC = 0 48 MTC = 0.47 MTC = 0.05 Total I.IE-6/RY i
Sequence 1.5E-5/RY l 1.5E-5/RY
, 3.0E-7/RY 1.1E-5/RY 1.lE-5/RY 1
Fregt ency l
l Pressure l
6 I
Across 1800 psid 2I50 psid g 2650 psid 1800 psid l'30 psid 2350 psid Tubes l
1 i
i
- >,i
,l l
r
{li
- l i
j}
-!l[!!
f:!
I >
i:l s
t I
7 n
S l
t r
Y k
a ee ya eY e
s n
m mR s
i i
id r
t t
r a
r e
i r c a e o
em r
n/
e n
r s l
s t
i n d
soi3 a or uo uV p
u i 3 nB e a -
t t
n r t o s
m3 Gqu t
oed cnimE r t sR 1 E
5 uir0 nl t
l s t e
S c
eO0 A
9 4
s r s o3 i
r )s p ae ae 1 e ei i
- 5. S SLInC=
OFT =
OFRInAPP =
N 9
2 r
e e
T r
t I
f u
6 n
S s
t r
rl Y
k eY i
e ya n
s s
n e
eiae l
ee D
d r
iF v e4 R
R r
e z
z i
rc a e t
r i
r n/
o i
o en r s r
er
/
s n
i a
me uViuE 7-p e F Gu m B.
i 3 t
t ea aoi3 a ouo r
uV r
l m3 sR 9 nR e s
t ed t r
o3 r
r t E
5 c c cninE t
el e
t r c R
r s o3 piae0 ps s 8
r irO0 uO 1
2 4
e ei MS SLInC=
OFT =
ORInAPP =
PP oP =
1 2
u m
l; I
t ss e
t I
e t
5 n
S k
t r
r Y
en ya eY e
s n
e e
d r
s s
m R.
t r
e z
z R
r r
P y
e r
a r i
a i
se n/
o o e m i
i e r
/
s in n
r s u e 4B 3
uV wVv u 4
p t
t S
ea aoi3 a ouo r
G ;u m3 r
%inC=
t r t sR 9 uRises0 E
5 s
d einE el r t l
et s t
r l e1 3
c n s o3 piae r 0 pe si eO0 eOe 9
3 a
y MSu r
r r
OFT =
O R :n A P P =
PPRP =
8 2
5LI i
r tn a
t 1
l n
5 a
f 4
k t
n e e ya cY e
Y d
a e mR s
i t
r c r_
r n/
o a R
s r
s o
ia i
p n
/
i i 3 c
Ga mB u c e a -
tai7 3
5 l
m6 E-3 ed nE t
r u
e c g c n is o3 pe0 A
A 0
2 e
r 3
e ei S
S MS SLInC=
OT=
N N
2
(
jI l
n s
t I
e 3
n S
s i
k eY e
eiae d
rl Y
ec y a t
r e a e mR s
b r
r r en/
o a
iF v e t-R.
t z
i i
n r
e r s
n i
d u
u ua nBdi3 aoi3 uVl uE 8
p t
i a -
s Gu oe s nE el i
m3 sRe s3 E
5 t
r t T
s s
r e
Re1 A
3 4
c o c nt 8
u nI piae0 rOoP e
e ei MS SLOC=
OFT =
r Pf
=
N 4
2 R
r t
l' o
v t
t i
2 n
S a
a k
r ee ya eY e
e Y
mR s
d r
r c r
r i
R t
a e
-M o
a i
s e i
r en/
s n
d i
e ue nBdi3 n
r e r taoi3 uVv u 4
p u-n Gu oes nE el i
s m3 sRes0 E
5 r t s
r e
h c q c n piae0 t
e eiuoI roe e1 A
J 33 h
e My SLOC=
OFT =
PPRP =
N 3
2 r
c G
i h
t I
W m
n S
k ya eY e
Y' d
mR s
i t
t r
t r en/
o a R
s a
r e ae r
p i n d
n
./
s e
u e nBdi3 ai7 4
5 t
a -
Gu ue s i
r m6 E
3 tnE e
t e
m nt 3
u oI per0 A
A 7
2 c
c S MS hiLOC= OT=
N N
6 n
e Al llil l
e h n
t k
ya d
o s
G a
r rY Y
e is e
s E
aBR R
p l
S o
R4
/
/
c e3 3
0 r
U4 eiE A
A A
E-0 c n S
6 c
NG SLI N
N N
2 A
2 I
I
-l j'
e l
y e
ec b
t r i
2 3
c n e
nc ne r.
a e :a o
o o
n n
n s
eu s s a
l s oe ui auq s
T i
i i
qi t
t t
t qe erb t
c c
c o er r c u en SI A
A A
TSF PAT
,,.l j
[
[j;
- I i ;
4!
1 Table 3 Segutncts.. Width May Result in Substantial Presurc Differendals Across flig.Stenin Generator Tuber (Chanter 15 Seonences)
Differential Pressure I'S AR Chapter 15 E3tirnaled 5equence
,Across SG Tubes S3uence Description I'requeng 24R5 psid (pressure is tiniited 15.1.5 Steamline Greak IE-4/RY (See Table 2) i by the three prewuriier safety and valves) 15.2.8 Feedline litcak 15.l.4 Spurious Secondary Ikmled tiy St.li Frequency System Valve Opening
'I otal Categor y li' 4/RY 2335 psid (prenvie is limited 13.1.5 5teamline D eak 1.2P 3/RY (See Tabic 2) by the three pressori/cr and PORV4 15.2.8 Feedline tireak 15.1.4 Spurious Secondu/
Ikionded by St.Il Frequency Sptem Valve Opcning Total Cutimory 1.2 E-03/l* Y 2150 2335 psid (pressurc 15.1.5 Steamline lireak 2.7Fi-1/RY (See Table 2) deiender.t on time of Si and termination) 15.2.8 Feedline lireak 15.1.4 Spunous Secondary llounded by ELil Frequency System V:dvc Opening 15.8 ATWS Ibes riot upply to period under consideration.
c Total Category 2.7E 3/RY 1800 2150 psid 15.8 Al WS Does not apply to period under consideration.
1450 1800 psid 15.8 ATWS l.1E.5/R Y 1400 1450 psid 15.4.1 Zero Power Control Screening value 1F.-2/RY llani. Withdrawal 15.4.2 Control llank Screening value IE 2/RY Withdrawal at power 15 A.8 ihl Ejection Some wry low value e Total Cateyory 2E.2/R Y 1335 - 1400 psid 15.13 Excessive Loa.1Incrm lloundal by spurious Si frequency 15.5.1 Spurious Safety 0.2/RY In}0ction Total Catreory l 0.2/11Y k
1 i
l
l Differential Pressure FS Alt Chapter 15 Estimale<l Sequente Across SG Tutws Sequence Descrirlion Frequency 1,ess Than Normal Operating 15.1.1 Petdwater l eaip.
Frequencies for this group I)P Reductioti were not estimated since the preasures are Imnded by die 15.1.2 increase in Fced Flaw nonnal pressures 15.2.2 Loss of lead 12.23 Turbine Trip 15.2.4 MSIV Closure Transients 15.2.4 through 15.2.5 Loss of Contenser 15.2.6 havt the potential to Vacuum cause a ctuck coen secccilary safety valve. These sequerres 15.2.6 Loss of 6900 v Iluses should be tvunded oy the SLil as discu: sed earlier.
15.2.7 Loss of Feedwater 153.1 Single RCP Trip 153.2 Com1 ete Loss of Flow l
1533 Loc ked RCP h(or 15.1.4 RCI' Shaft lireak 15.4.1 Control Rnd Errors 15.4.4 Statt Up on inacuve RCP (5.4.6 Boron Dilution 15.6.1 Stack Open Pressuriwr IMV or Safety 15 6.2 Instrumentation Line 13reak 15.63 SGTR 15.6.5 LOCA l
1 m
__m__m___
m_.m_m._____. _ _ _
____________,_.______m__mm___. _._.
__m____._n__.__
_m_
h'
ATTAct! MENT 1.1 1111taluallon of AfcGuire S1cand;enerator Tube Inteeritv Issue for End of Cvele Onerritin NUREG-0844 is a comprehensive, state-of the. art analysis of the probabilities, consequences and risks for steam generator tube ruptune (SGTR) events. The report contains analyses of care melt and non-core melt events.
We agree that the NUREG analysis is a conservative analysis of the probabHities and consequences of SGTR related accident sequences.
Duke's N1NS (McGuire Macicar Station ) IPE/PRA contains the analysis of SGTR e*:cnt sequences leading to core melt conditions There are similarities and differences between the Duke analysis and the NUREG analysis.
We have attempted in the last few lays to evaluate the frequencies of die transient-induced high delta P SG tube stess scenarios, which have a probability of leading to j
SGTR. (See attached event trees.)
\\
For the non cure melt sequences, the similarities and differences are as follows:
The initiating event frequency fc,r most of the events are about the same. (For example, Duke uses 3x10-3 or a high energy line break insid: containment f
while the NUREG uses lx10-3 )
i The type of events of interest are about the same, except perhaps the LOCA +
a SGTR combination. (The 1000F or so adverse impact on cladding peak clad temperamre calculations arising trom a str.all impact - on post LOCA refill /reflood phenomenon due to multiple SGTR condi ion - should not be t
considered a significant contributor to core melt.)
1 For overcooling cvents (such a SLB), operator action to control primary side repressurization could be comidered as a mitigating fa tot. Also, the PORVs limit the pressure to 2350 psiinstead of 2600 psi.
l Even with operator failure. the maximum realistic delta P during a SL3 it l
l approumately 2500 psi instead of the 2600 psi used in the NUREG, I
L For ATWS events, the fractions of the time the delta P_is greater than or equal to 2500 psi are less than those estimated ir. the NUREG because of the presence of three pressurizer SVs and three pressurizer PORVs at MNS.
e.,
l l
Page I of 2
. ~. _.
I A
)
ATTACHMENT 1.1 Duke's PRA ass.essment does not have a number for the conditional probability of tube failure during a SLB or for the probability of multiple tube failures.
The conditional probabilhy of 0.25 for SG overfill during a SGTR event is very consersative.
Using realistic source terms (the hlNS coolant activity is very low), the consequences of these non core melt sequences is expected to be much less severe than those calculated in the NUREG.
For core melt sequcnces, the similarities and differences are as follows:
The initiating event frequency for a single SGTR is about the same.
Duke does not have a number for the frequency of a multiple SGTR event.
The core damage frequency for a single SGTR initiator is two orders of magnitude lower in the Duke MNS PRA Malysis versus the NUREG analysis.
The ability to der,ressurize the primary side and the plant specific reliability of DifR systems are probably the major causes for this difference.
The sequences modeled in the NUREG which result in depleting the RWST aner long periods of time were modeled as part of the MNS PRA event tree.
However, thev were not quantified because it was decided that if credit is given for throttling ECCS Fow to match decay heat, or refilling the RWST, then this sequence would not result in core damage.
In the consequence an.dysis, the MNS specific fission product telease fractions and site consequence results are simi!ar to NUREG analvsis results.
The results unply that frequencies of accident sequences are low and that the risk from -
SGTR related sequences is also acceptably low.
r Page 2 0f 2 m:
,,w n.
y
-, e