ML20235T139
| ML20235T139 | |
| Person / Time | |
|---|---|
| Site: | Peach Bottom  |
| Issue date: | 01/26/1987 |
| From: | Kowalski S PECO ENERGY CO., (FORMERLY PHILADELPHIA ELECTRIC |
| To: | Morris B NRC OFFICE OF NUCLEAR REGULATORY RESEARCH (RES) |
| Shared Package | |
| ML20235T023 | List: |
| References | |
| FRN-52FR7950, RTR-NUREG-1150, RTR-NUREG-CR-4550 52FR7950-00027, 52FR7950-27, NUDOCS 8710090272 | |
| Download: ML20235T139 (12) | |
Text
{{#Wiki_filter:m , 4 1 PHILADELPHIA ELECTRIC -COMPANY < 2301 M ARKET STREET P.O. BOX 8699 r PHILADELPHIA A.' PA.19101
. . .u. m o . m ..c .....,;,,, ' '
- IAN 2 6 $87 i Mr. Billy M. Morris, Director Division of Reactor System Safety !
; U.- S. Nuclear Regulatory Conmission Washington, DC- 20555 ,
SUBJECT:
Conments on Draft NUREG/CR 4550, Volme 3 Peach Bottom Accident Sequence Characterization
Dear Billy:
The opportunity to work with your personnel dwing the J preparation of the report'of Peach Bottom Accident Sequences has been beneficial for us In understanding the current.lssues and the risks calculated for Peach Bottom.' Although the analysis appears to be realistic In nest areas, some extremely conservative assumptions have been used, particularly in the area of Station Blackout sequences. Because it is difficult to discern these conservatism, we believe the analysis should be corrected'to more realistically estimate Peach Bottom's core damage frequency.
- Our detailed conments (see Attachments 1, 2 and 3) have been informally given to your contractor during 'the course of our review.
The nest significant are described below: Corrnon Cause Battery Fallure The dominant accident sequence (TBUX) comprises about half of the Peach Bottom calculated core damage frequency. The sequence results from postulated cornron cause failure of at least five (5) station batteries (both units) coincident w1th loss of offsite power. Because of lack of data for contron cause battery failures, a fallure rate of 1.0 was used for batteries after two had been calculated to fall. If nore normal dependent fallure rates of approximately 0.1 (or even 0.5) had been used, this sequence would not be sigr:ifIcant. We belIeve that the analysis does not reflect our survel11ance testing program, the weekly T.est of every diesel-generator, and the separation of the batteries between two generating units. In effect, a new significant accident sequence has been asserted to exist without the necessary technical support. . 8710090272 B70925?.
' Battery Life
%O PDR ,
The battery life used in the analysis (6 hours) is conservative. Plant procedures direct load shedding during a blackout. , . Additionally, the HPCI and RCIC systems would not run ' concurrently, Indicating longer life for each 250 voit D.C. system. It is expected that the additional recovery tirre provided by extended battery life would be significant. Analysis with extended battery life should reflect the refilling of the Condensate Storage Tank (CST).
1
; % D
-. - x '-
( , t q< 3 j It is atticipatedkhattheDieselFirepumpandtheconnectionprovided l for thid rpecific purpose would be used to refill the CST. HPCI and ECIC suctii.,n wouldj then be transferred back to the CST, as directed by
, procedures, so tha". they would remain operable regardless of suppression pool te qerature..
y [ Offsite 4cser Recovery Theprobbbilityofoffsitepowerrecoveryduringblackout should [. reflect the plant-specific recovery feature s, particularly the eleven unit Conowingo Hydroelectric plant which supplies one of Peach Bottom's offsite feeds. Use of Cct,Wingo to feed Peach Bottom is , l supported by system-wide and plant.-specific procedures to assure rapid ; i return of offsite power from this relfable, blackstart source. 1 Our experience in dealing with your contractors on this project indicates we are all interested in a document of high technical quality which accurately characterizes Peach Bottom risk. A more realistic analysis in the area of statinn blackout will help to attain this goal and make the Peach Bottom enslysis more useful as support for decision-making and other projects such as your PRISIM application 1 which is underway. Verytrulyyfurs, f i S. d
. fowalski Vice-President Engineering and Research Department 3
- b' ARD/j et /1001 %01 t
Attachments Copy to: V. Stello '?RC R. Bernero - NRC F. Elttwila - NRC C. P.eed - CECO. F. Harper - Sandia A. Kolaczkowski - SAIC M. Ernst - NRC (Region II) i T. Johnson - NRC (PBAPS) ( ) jY
)
Ji_______._________._.___.______ __ _ _ _ ___ _
?
/ iy
*1 P uch Bottom Accidsnt Ssquence ChNdeterization M
.tci ,
'p m e
, I Q [.
,y (( .ij Alt $ctment 1 ,
Comnents og IMEG/CR-4550, vol . 3 i .:/ , , , e 4, r [ 0 Page 1-4: The core damage JMpquency for WASH-1400 , shour cibe 2.5E-5. '{ 0 Page 1-6: Just prior to the sequence gmur,descrhr.lons, a t statement should be added referring the; reader to Section V.2 for- , more detailed descriptions, assumptions, and sensitivities of the j sequence' groups.
- Page 1-9: The statermnt found on shge 1-13 regarding corrrron cause dominated sequences should also be placed at the end of the-
- discussion of TBUX. . The dominance e f TBUX In the total CDF makes discussion of comnon cause extremelpf critical.
t j. Page 1-9: Calculations and plant tiests have confirmed that the a ECW pt.mp can function adequately witaout booster punps. ^ Discussion of the sequence group TB will change glven ECW punp , I operation without boosters In both the areas of success criteria and probability. Additional. text / analysis should reflect this fact. (I j
<J 3 .
a
" Page 1-10: As a note to the discussion of TCUX, both the New j G y York Pcwer Authority (NYPA) and ORNL have postulated TCUX. type ,
sequences with RCIC'and CRD operation delaying if not preventing * ,A of a core melt. Power is flow-controlled (i.e. poWor is dependent t on water level, the operator is instructed by tN# RIP procedures to lower level), therefo're, a TCUX sequence may rht necessarily '1 lead to core melt. Page 1-10: Discussion of the TBUP sequence Indicates that if- 1s power is not restoreci in 30 minutes, core damage will occur. It' is important to indicate at what time core damage wMl occur in'i order to Indicate the time available for power recovery or otherf preventative actions. The availability of power 2-3 hours aften initiation of the sequence would make a significant difference (~n > the assessed likelihood of recovery. o Page 111-2: Please delete any reference to Peach Sottom reviewing the results as this has not occurred.- 0 Page IV.2-1: The reference to Ma dger of Engineering should be changed to Superintendent of Plang Services. Page IV.3-6: An 10RV represents a breach of the primary syster ; and not containment. y i i
l .
~
Parch Bottom Accident Sequence Characterization o Page IV.4-3: Does a "BWR of this type" refer to the Mk-1 containment or BWR-4 NSSS7 Page IV.4.3: It is overly conservative to asstme that saturated pool conditions will cause NPSH problems to the point of low pressure gf ptmp failure, y Page IV.4-5 A recent Brookhaven sttdy has determined nere realistic nOnbers for downcomer and SRV pipe fallures. Page IV. d5: Switching HPCI and RCIC back to the CST is not as difficult as indicated. If the low CST level is cleared (i.e. CST'1111ed or refilled) RCIC may easily be swltched back. Similar!1y, HPCI may be switched if signals are cleared or Interlocks bypassed. PECo's station blackout procedure contains Instructions for such operations. o Page IV.4-7, IV.4-15, IV.4-24: SLC is noted to be ineffective i for large LOCA scenarlos since it flows out the break. Th i s ,l_s overly conservative for steam breaks. Page IV.4-7: Please provide a document ntmber for the GE analysis mentioned in note (7). l
.Page IV.4-15: Avallibility of the Power Conversion System (PCS) l should be included as a success criterla for late containment l overpressure protection for intermediate LOCAs, as it is for the j small LOCA.
l l Dage IV.4-34: Success of primary system depressurization (Event X) l Is indicated as 2 or more valvts open. With only 2 valves open l the reactor would not immediately depressurize to the point needed for low pressure injection. The time reautred for depressurization could be 10 min. o Page IV.4-44: CRD ptmps must be manually loaded on the diesels following a LOSP, Adequate core cooling can be achieved with 1 CRD ptmp in approximately 3-4 hours. O Page IV.4-57: A 6 hour asstnption for battery life is ovarly conservative. Changes in battery life can change the success l probability for operator actions such as venting. Battery usage j would be sequential, (i.e. use RCIC first until its battery is depleted then HPCI) thus greatly extending battery life and , changing the likelihood of venting success. Venting would be used { prior to core degradation and tht.s success likelihood wauld be l high (l.e. no radiation concerns). l 4 i i
Pecch Bottom'A ccident Sequence Characterization
- Page IV.4-57: PECo station blackout procedure E-28 Instructs the operator to switch HPCI and RCIC suction back to the CST and'also
- gives specific .ltsnpering instructions.
- o. Page'IV.4-58: Explanations regarding h w power recovery is treated, the nonber of diesels required for safe shutdown, and temporary fixes such as portable diesels should be included in the sequence descriptions.
Page IX.5-3: Although Instrtinent N, le- a primary source to the-MSIVs and SRVs some subtletles exist between Instrtsnent N , the 2 acctsnulators, bottle backup and instrtsrent air backup for the ADS valves which are not properly reflected. 0- Page IV.5-10: The.assunption that HPCI is non-recoverable-If it falls to trip on low suction pressure or high reactor water level is overly conservative and not technically supportable, o Page IV.5-11: Asstsnption (13) - Depletion of the CST should not be included in cases where HPCI and RCIC operation is required for long periods of time since many additional sources of water exist to refill the CST. These include a connection to the fire protection system which is provided for the specific purpose of assuring makeup, even during blackout. Page IV.5-12: Both HPCI and RCIC wil,1 automatically switch on low CST level. Page IV.5-16: Opening doors should suffice for room cooling. Page IV.5-21: The CRD ptsnps normally take suction f rom the condenser hotwell. Page IV.5-23: The statement regarding loss of room cooling not causing failure of the CRD punps is correct. The CRD ptrnps are not in a room at all. They are located on the 116' level of the turbirm building in a large open area.
~
Page IV.5-31: The limit described for nitrogen supply pressure and the Interaction with the ADS valve operators should be trore fully explained. Page IV.5-32: The included diagram is not a precise arrangement; check valves in the system are not shcwn and the acctnulator is positioned differently. Page IV.5-33: Each ADS acetinulator supplies enough pressure for approximately 5 valve operations at attrospheric containment pressure. Page IV.5-33: Automatic ADS Initiation occurs with either a low-low water level signal with an 8-minute time delay or low-low water level and high-drywell pressure. Each also requires either 1 RHR pump or 2 core spray pumps running.
' Peach Bottom Accident Sequence Characterization Page IV.5-34: The diagran should indicate that Instrunent N7 is the "alr" supply to the ADS valves. A note should also be added indicating that the instrument N system 2 also has Instrunent air and the safety-grade N9 supply as backJps. Also on the same diagram, a branch should be added Indicating the existence of an accumulator in the gas supply of each ADS valve. .
Page IV.5-41: Assumption 8 states that the CST was not modeled as an alternate source of water for the core spray systen. This is extrenely overly conservative for long term sequences. Page IV.5-42: Suction for the LPCI pumps cannot be aligned from the CST. This connent 01so applies to pages 5-48 and 5-61. Page IV.5-44: LPCI pumps start long before the injection valves are denanded open (ADS logic needs pumps running in order to depressurize). Page IV.5-45: Opening doors would also provide roan cooling. Page IV.5-46: LPCI pumps are denanded to stop or prevented fran starting if the suppression pool suction valve or any of three SDC valves are not fully open. O Page IV.5-68: Not all modes / combinations of RHR System operation are considered (i .e. 1 loop in Suppression Pool Cooling and 1 in LPCI). Page IV.5-69: Diesel generator failure is assuned after 15 minutes without roon cooling. This is an overly conservative assunpt i on. Diesel conbustion air is drawn f rom the room and the intake of warner air resulting from no room cooling will not cause a shnrt-tenn catastrophic engine failure. Rather, an increase in combustion air temperature will nerely rcsult in a decrease In the continuous load carrying capability of the engine. This decrease does not exceed 5% for intake air temperatures as high as 130oF. ; The operators being aware of this situation would be expected to open doors or reestablish room cooling. Page IV.5-70: Diesels automatically start on total loss of offsite paaer, low reactor water level, or high drywell pressure coincident with low reactor pressure. I Page IV.5-70: Protective trips of a diesel are overridden only ) on a LOCA signal. l Page IV.5-72: Diesel generator weekly testing is not an I assumption and should be noved to the EPS Test and Maintenance section. Page IV.5-73: Battery feeds for the diesels are also not an assumption but a fact, this should be noved to the description of , batterles on page 5-72.
-4_
L _ _________._.__._________________________________
. Perch Bottom Accident Sequ:nce Characterization Page IV 5-77: The ECW pump autanatically trips 45 seconds after initiation if ESW punp discharge pressure is established. The operator is not required to turn the pump off. The same connent applies for pages 5-78 and 5-81.
Page IV.5-77: Fallure of ESW will not cause failure of the low pressure punps. The RHR and core spray punp roons are expected to reach equipnent design temperatures within about 4 hours without room cooling. RHR pump seals are designed for 160 F fluid temperatures which is strictly a function of torus temperature. CS motor bearings are designed for cooling water temperatures up to 165 F. These temperatures do not necessarily represent operability limits. Page IV.5-78: The ECW punp will function adequately without booster punps. This will change the nelntenance contribution of the MOV-0498 valve as well as the diesel B and C failure impact on the ESW/ECW systen. This comnent also pertains to the description found on page 5-83.
- Page IV.5-81: The operator is not required to switch to the energency heat sink node. The EMS mode will be automatically aligned after 90 seconds if both ESW pumps fall to provide discharge pressure.
C Page IV.5-82: Makeup in the emergency cooling towers is not modeled because " low pressure reduces the snount of evaporation." This is a meaningless statenent. O Page IV.5-97: The Instrument air systen is a backup to the instrument nitrogen systen. Page IV.5-99: The air compressors do not trip when a reactor trip occurs. Page IV.5-101: The TBCW pumps trip on loss of offsite power and RBCW automatically supplies cooling water to essential loads through air operated valves 2352 S 2354. O Page IV.5-102: Instrunent N is m re important and isn't 2 described or modeled. o Page IV.5-104: " Venting success during an ATWS requires 2-18" vent l paths as a minintm". The power level that is assumed should be l stated. A loa power ATWS sequence will have additional tine j and/or una11er vent paths would be required for success. ' Page IV.5-104: The purge and vent valves are used for inerting/delnerting during each outage requiring drywell access. 1 The 18" butterfly valves are covered by containment isolation i valve Tech. Specs, and are logged and limited to 90 hrs of operation l per year. The ILRT lines are used during each ILRT (i.e. - at I least 3 times in 10 years per 10CFR50 Appendix J).
- Page IV.6-2: The 6 hour battery depletion tine is extrene;y conservative. Actual battery life is expected to be substantially longer, with the duration of operability of the 2-250v systems being additive.
-3 l - - _ _ _ _ _ _ _ _ - - _ _ _ _ _ _ _ - _ _ _ _
'. Peach Bottom Accident Sequence Characterization 0 Page IV.6-4: The Peach Bottom staff will recover room cooling by opening doors to critical areas which are nornelly locked in accordance with plant procedures. Reference to Tech. Spec.
limitat!ons is inappropriate e. Page IV.7-7: NUREG/CR-4696 (Draft) study of Peach Bottom venting conservatively indicates success of .6 .7 for ATWS sequences. O Page IV.7-9: Note (c) - The perceived reluctance to inject river water into the core is inappropriate. PECo's Implementation of the EPG's (TRIPS) directs the operator to inject river water, if required. C Page IV.7-10: Success or recovery of the CRD system is not dependent upon the operator aligning two punps. One punp autanatically injects without operator intervention. A substantial snount of the flow capability of two punps can be obtained from one punp by opening the flow control valve from the
. control roan.
Page IV.9-6: Unit 2 and Unit 3 batteries should be considered as separate systens. Page IV.10-1: The TRIP procedures are based on the generic BWROG Emergency Procedures Guidelines (EPG's) Revision 2. o Page IV.10-6: A 10-14% chance of falling to defeat ADS with an inhibit switch and procedures directing Inhibiting ADS for a controlled blowdown is much too high. Comment also applies to Page 10-38. Page IV.10-29: ATWS with MSIV closure analyses for Brown's Ferry were used for the Peach Bottom ATWS event tree. These analyses are not necessarily appilcable to Peach Bottom due to possible differences in contalnnent paraneters, system trips, and procedures, all of which can have an effect on timing. O Page IV.10-36: The frequencies of transients listed in the table are high. Area specific grid data indicates a loss of offsite power of .05 or approxinetely 25% better than the number found in the table. Attachnent 2 discusses LOSP and reactor trip induced LOSP. In addition, NUREG/CR-3862 initiating frequencies for Peach Bottom Indicate much lower MSIV closure and LOFW frequencies. Page IV.10-45: Manual control of HPCI during an ATWS sequence is highly probable, thus high or oscillating power levels would be mitigatea. Page IV.12-3, 4: Sensitivities on nelntenance unavailability of MOV-0498 and ESW pump failures show less than a factor of 2 variation of overall core damage freauency. Recent Informat ion regarding ECW puno operability without boosters will lower the CDF by a factor of two as determined by the sensitivity runs.
Peach Bottcm Accident Sequence Characterization O Page V-12: Contributors to core damage will change given the argtnents found in the cover letter and recently provided information (i.e. - MOV-0498 maintenance unavailability, Diesel 2 ,, and 3 failure, ESW ptmp failures, and DC cormon rrode fallures). o Page V-19: The conclusion of comon trade battery failure affecting Unit 3 batterles is not substantiated with data or reasoning. Same corment applies to Page V-20. Page V-19: Conowingo Dam would supply power to Peach Bottom 1f a loss of grid occurred. This dam has blackstart capabilities. Page V-22: Sensitivity case 1 for the TBUX sequence shows the , . impact the comen rrode beta asstmption has on the sequence and thus the total CDF. Investigation of rrore realistic cormon mode failure rates based on maintenance, surveillance, and the separation of the batteries between the two units should be undertaken. The batteries and associated emergency equipment needed for Unit 2 operation during LOOP ls listed in Attachment 3. Page V-25: Diesel failure is stated as 1.13E-2 for the listed cutsets whereas Peach Bottom specific failure is 6.7E-3. Why is the ntnber different than the Peach Bottom specific ntmber? Page V-26: ECW punp operation significantly affects the asstmed loss of cooling to diesels 1 and 4 due to comen trode fallure of diesels 2 and 3. - Page VI-3: Discussion regarding the contribution of TBUX and TB to overall CDF should mention or refer to the source term / consequence report. The consequences of the two station blackout secuences are vastly different. GAK/cmv/10028602 9 gg j ,-
'\
h - Peach Bottom Accident Sequence Charac,terization . - l e . h, ' N) 8[l 1 h E y kk
,'u tn gi .
E!
% c,5i s
E f E
% eE8 A
) E 2 f
1 1 I I E E T M ! !
f i
{% .6lI o l 8[ il y
.: % m,, , -
si , , , . s E. : . i g A. F "" E o E g
@s s (: 5I
~< ._.) ez a ---
- s, c .E -
e h ' ig s i eg
, ' % P: % s- %
s ng ng g ""
,, z; Z Z .[ I,
% *
- e.e 9 c' a
EI x t . s Z Z .x t
~m 1 % % yy T <~
I ~ - Z Z ! a t nb 4 4- E - s s e a < Y.Id E
@!* 6! * !
I .$ 4 IE, 2
-{
l=$ t s E 4 I$* C 6i '2 '2 ! Es !s 8 l . - - - ta ne o Er Er E
\ fr Er E
'4 ds ;! ;! :-
e,s
. .as, DRAFT n.s-n
4, Peach Bottom Accident Sequence Characterization Attachment 2 The loss of offsite power initiator should be derived using local grid-specific (PJM) data since it is influenced by local factors such as climate, transmission system design and operating procedures. This was done for the Limerick PRA study with a result of .05 per reactor year. In addition, Cesnowingo Dam is a blackstart facility with the capability of supplying Peach Bottcm with power given a grid failure. Specific procedures exist, both for Peach Bottom and the PJM grid, on restoration of offsite AC and system restoration following a loss of grid respectively. Estimates of offstte power restoration at Peach Bottem from Conowingo given fallure of the grid is between 40 and 60 minutes. LOSP resulting from a turbine trip is approximately a factor of 10 higher than would be expected from nuclear operator experience data. Since no recorded cases of LOSP Induced by a nuclear plant trip in trore than 470 reactor years of experience through 1982M have occurred, an estimate of probability can be made. 1 Probability = (470 Rx-years)(9 Transients /Rx year) Probability = 2.4 x 10~ per Reactor year 9 transients per year is taken from EPRI NP-2230 Including first year data. q peration, 7.3 transients /yr) would MMore recent Indicate data Cl.e.of1000 a probability 1.4 Rx-yrsj/R x 10 x-yr. GAK/ jet /08048601 l l
. Peach Bottom Accident Sequence Characterization Attachment 3 Batteries needed for Unit 2 system / component operation following a loss of offsite power event Unit 2 Unit 3 Diesel 1 2A Diesel 2 2B Diesel 3 3C Diesel 4 3D HPCl 2B and 2D RCIC 2A and 2C ADS /SP.Vs 2A or 2D Conman node battery failure for sequence TBUX would therefore need to include as a mininun, failure of batterles 2A, 2B, 2D, 3C, and 3D.
1 J . _ _ _ _ _ - _ - _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ - _ _ _ _ _ _ _ _ _ _ .}}