ML20209G007
| ML20209G007 | |
| Person / Time | |
|---|---|
| Site: | Crane  |
| Issue date: | 03/31/1999 |
| From: | BROOKHAVEN NATIONAL LABORATORY |
| To: | |
| Shared Package | |
| ML20209F261 | List: |
| References | |
| NUDOCS 9907160209 | |
| Download: ML20209G007 (15) | |
Text
.
SUBMITTAL-ONLY SCREENING REVIEW OFTHE THREE MILE ISLAND (TMI) UNIT 1 INDIVIDUAL PLANT EXAMINATION FOR EXTERNAL EVENTS (Seismic Portion)
August 1998 (Updated January 1999, Finalized March 1999)
Brookhaven National Laboratory E S 7
- I 8 E E E E 8 7>8 3 s v P
. 1.
INTRODUCTION 1.1 Purpose In response to the NRC issued Syplement 4 to Generic Letter (GL) 88-20, " Individual Plant Examination of Extemal Events (IPEEE) for Sevem Accident Vulnerabilities - 10 CFR 50.54(f)," the GPU Nuclear i
l Corporation (GPUN) performed an IPEEE for the Three Mile Island Nuclear Generation Station, Unit I (TMI-1) and submitted the IPEEE results to NRC {} J. Brookhaven National Laboratory (BNL), as requested by NRC, performed the submittal-only screening review to verify the technical adequacy of the seismic portion of GPUN's IPEEE submittal. As a result of this review NRC sent a Request for Additional Information (RAD to GPUN. GPUN responded with the Three Mile Island Unit 1 IPEEE-Response to RAI (April 24,1998)[2]. This Screening Review presents the results and conclusions of the BNL review and evaluation of both the original submittal and the licensee's response to the RAL BNL's methodology utilized for the review followed the guidelines provided in the document titled,
" Guidance for the Perfonnance of Screening Reviews of Submittals in Response to USNRC Generic Letter 88-20, Supplement 4," (Draft, October 24,1996), as amended by NRC.
1.2 Background
TMI-l is a single unit,2-loop B&W PWR with turbine generators provided by General Electric and with a rated power of 808 MWe. The commercial operation of Unit 1 started on September 2,1974. The site is located on Three Mile Island, which is situated in the Susquehanna River, upstream from York Haven Dam, l
in Dauphin County, Pennsylvania. The cooling water (service water systems)is drawn from the Susquehanna J
River via the River Water System. In addition, two hyperbolic natural draft cooling towers are used as the ultimate heat sink to supply the main turbine condenser with makeup and blowdown to the Susquehanna River.
The Safe Shutdown Earthquake (SSE) for the site is 0.12g, and the plant is binned in the 0.3g fo:used-scope review category.
1.3 Licensee's IPEEE Process and Licensee's Insights The original TMI-l PRA, including the seismic analysis, was submitted by GPUN to the NRC in December 1987. The study was completed by Fickard, Lowe, and Garrick, Inc. (PLG) with significant input from GPUN ' staff. The NRC and its. contractor reviewed the PRA and published a review report.
l.
GPUN has implemented a number of changes to the plant and the procedures as a result of the 1987 PRA.
GPUN also completed the update of the internal as well as the extemal event PRAs and used the results as the IPE and IPEEE reports.
The system analysis was performed using the PC software package RISKMAN, and the fragility analysis and the soil evaluation were performed by EQE Intemational.
I t
l l
The overall PRA process is considered to be consistent with the PRA methodology described in NUREG-1407. The risk quantification was performed using both the EPRI and the revised LLNL hazard curves.
l The listing of components, walkdown procedures as well as the evaluated results of component fragilities are extensive and well documented. However, the evaluation procedures for seismic fragilities of structures and components, as well as the soil liquefaction, are provided in a separate report made available for review in response to an RAI. A detailed relay evaluation was performed by GPUN based on the guidulines of EPRI 1
NP-4147.
In the system analysis, a tot'd of 21 seismic top events were developed, which are " complements," of the I
internal event top events. iddition, the independent offsite power failure was added to complete the list of seismic top events.
He results of the IPEEE identify no vulnerability due to seismic events. The calculated seismic CDF using the EPRI hazard curve is 3.21x10~5/yr, which is 23.6 percent of the total external event CDF. Using the revised LLNL hazard curves, a higher value of 8.43x10-5/yr was obtained. The dominant seismic sequences involve failures of offsite power, Class IE AC power, control room ceiling failure, and the seismic failure of emergency diesel generator air start receivers.
2.
REVIEW FINDINGS 2.1 IPEEE Format and Methodology Documentation The submittal appears to be consistent with the guidelines of NUREG-1407. The study addressed all the issues that are emphasized in NUREG-1407, including plant walkdowns, relay chatter, soil evaluation, nonseismic failure, human actions, and containment performance. In the submittal the completeness of documentation, the evaluation procedures for structural and component f: agilities as wdl as soil failure are referenced to a separate EQE report [3].
2.2 Seismic Review Team Selection The seismic review team (SRT) consisted of personnel from GPUN, who are familiar with the systems and operation procedures, and external consultants, including EQE International and PLG, Inc. Although the selection procedure for the SRT is not addressed in detail as a separate subject, it appears that the SRT selection meets the NUREG-1407 objectives.
2.3 Hazard Analysis The study used both the EPRI and the revised LLNL seismic hazard curves. According to the quantification scheme, four (4) discrete earthquake evems at 0.15g,0.25g,0.40g, and 0.60g were considered (where the EPRI approach was used). It is stated in the submittal that the upper bound of the hazard curves is 1.0g. No sensitivity study was performed to evaluate the effects of the use of a cutoffless than the 1.5g described in NUREG-1407. No description was provided in the submittal regarding the response spectra used in the seismic analysis. Additional information regarding the assumed seismic input was provided by GPUN in their response to th' e RAI. EPRI's uniform hazard spectrum (UHS) was used to represent the ground motion.
This procedure for determining the ground motion is considered acceptable.
2
2.4 Component Selection A total of 1445 safety-related components were initially identified based on the IPE study and the A-46 component list. Components categorized as " generic high capacity component" were screened out and a total of 328 components were screened in. The screening procedures and component listing are well documented in the submittal and consistent with the guidelines of NUREG-1407.
2.5 Plant Walkdown Approach l
The submittal states that a series of walkdowns were performed including the initial familiarization j
walkdown, USI A-46 walkdowns, seismic capacity walkdowns, and relay walkdowns. The walkdown team consisted of engineers from EQE International (structural) and GPUN (seismic and PRA engineers). The results of the seismic walkdowns were extensively used in the screening verification as well as to identify I
potential seismie interactions. This walkdown procedure seems to be appropriate.
2.6 Fragility Analysis 2.6.1 Structural Response Analysis No information was provided regarding the seismic structural analyses in the submittal. As part of the licensee's response to the RAI, however, the EQE fragility report was provided. Based on a review of this report, the procedure for the structural response analysis seems to be appropriate.
2.6.2 Structural Fragility Analysis The structural fragility analysis was cerformed by EQE International and described in their report (3].
Relatively high median structural cgacities of 1.lg (condensate storage tanks),1.7g (diesel generator building), and 1.2g (reactor buildings) are listed in pp. 3.6-3.7 of the submittal.
2.6.3 Component Fragility Analysis A median fragility value of 1.0g was used to screen out most of the equipment items. A total of 57 items were screened in for a detailed fragility evaluation. The overall impression one gets of the evaluated fragility parameters is favorable and they seem to be in line with existing fragility data. No information was provided in the submittal regarding the component evaluation procedures, however, as pan of the RAI responses detailed fragility calculations by EQE were provided for Train BES bus 480V, the CST, the BWST, and the decay heat exchanger. Based on a review of these calculations the procedure for the component fragility analysis appears to be appropriate.
2.7 Soil Evaluation The submittal states that a soil liquefaction analysis and a seismic soil settlement analysis were performed l
by EQE International, and the results are described in their report. These analyses concluded that, should soil liquefaction occur, foundation-bearing strength failures are not expected, but ground settlement on the order of 0.5 to 1.0 inch could be expected. For the structures founded on compacted backfill, including the Diesel Generator Building, the Borated Water Storage Tank, the Condensate Storage Tanks, and the Turbine l
3 l
l Building, the aaalysis concludes that it is unlikely that such a small soil settlement would lead to significs.nt structural damage.
2.8 Relay Chatter Evaluation l
The submittal states that most relays were screened out using the screening criteria in EPRI NP-7147,
" Generic Ruggedness of Relays"(GERS). It is further stated that "all relays that cannot pass any seismic screening criteria will be replaced during upcoming refueling outages with the exception of the Westinghouse l
COM-5 model relay." As noted in Sectin 2.16 of this report, an RAI response indicated that some relays i
l were replaced while others were found acceptable on the basis of a GPU calculation. A total of 87 relays were left for the fragility evaluation using the generic data from GERS. The recovery of relay chatter was considered in the seismic analysis utilizing the recovery split fraction rules. The evaluation procedures, as well as documentation on the relay chatter, are considered to be consistent with the guidelines of NUREG-1407.
2.9 Containment Performance The effects of seismic events on the containment performar.ce have been evaluated from two perspectives:
l containment structure seismic capacity and the fragility of containment isolation valves and signals.
The lowest median capacity of the containment building was estimated at greater than 11.0g, with failure occurring due to shear failure of the shell wall near the base.
Regarding the seismic fragility ofisolation valves and signals, the lowest component was identified to be the ESAS relays with a median fragility of 0.89g. It is concluded in the submittal that " recovery times for containment isolation failure allow sufficiem time for manual or automatic closure of the valves prior to core damage."
2,10 Nonseismic Failures and Human Actions i
The seismic model used ia the TMI IPEEE analysis includes a seismic pre-tree developed for the IPEEE and l
the event tree logic models developed in the IPE for the Level 1 internal event PRA. Seismically initiated events enter the seismic pre-tree where system failure due to seismic event is determined. The impact of seismic failure is then propagated through the Level 1 PRA models for sequence determination. In this way, both the contributions of failure due to seismic events and independent or random events are captured in the l
analysis.
l l
Modeling of recovery actions is part of the internal event Level 1 models and is included in the seismic model. However, recovery from a loss-of-offsite power, which is modeled in the internal event PRA, is removed from the seismic model. This is to account for the possible extensive damage caused by the seismic event on the offsite grid and plant switchyard. In fact, most seismic failures are assumed to be unrecoverable l
in the seismic analysis. In general, recovery is allowed for systems with non-seismic related failure and not l
allowed for systems failed by seismic causes.
Loss-of-onsite ac power due to relay chatter is considered in the seismic model, and recovery from relay chatter is added to the seismic model. Relay chatter recovery is only allowed if: no LOCA is occurring, the 4
1 l
l
L control rods have inserted, emergency feedwater and de power are both available, sufficient time is available for the recovery actions to be successful, and Class IE ac power has not been failed due to the seismic event.
The efrects of human action failure rates are evaluated in sensitivity studies. Human error rates are increased by a factor of 10 in the sensitivity studies, and results show a small effect on the total CDF (an increase of 0.6 percent from the base case). Sensitivity studies were also performed for recovery actions and relay chatter. The total CDF is increased by 11.5 percent if all recovery actions are assumed to be unsuccessful.
It decreases by 0.6 percent if the effect of relay chatter is ignored in the model, and increases by 6.7 percent l
if recovery from relay chatter is not allowed.
In summary, the treatment of nonseismic failures as well as human actions in the IPEEE and the discussion provided in the submittal seem to be reasonable, thorough and adequate.
2.11 Seismic-Induced Fires / Floods l
A plant walkdown on the fire cuppression systems was performed by EQE lnternational and GPUN personnel to exami e the following:
Seismic-induced fire initiation; Inadvertent seismic induce 3 actuation of fire suppression systems; and Seismic-induced degradation or diversion of fire suppression systems.
The walkdown did not identify conditions that may contribute significantly to a seismically induced fire scenario. It also fourx! that inadvertent actuation of fire suppression systems is not a problem because in areas where safety-related equipment is located, the fire protection systems are either manual or are preaction requiring redundant sensor / relay circuit closure to result in actuation. Furthermore, very few instances were l
noted where sprinkler heads were in close proximity to other plant features, and sensitive electrical equipment is equipped with sheet metal " hats" to protect it from the effects of water intrusion. The automatic water systems contain no mercury switches. Although the CO system includes mercury rocker switches, 2
they serve as the pressure switch for the compressor and activate fire dampers and are not used in the circuitry for the discharge valves. Although the effects of dust raised during an earthquake on fire suppression actuation is not addressed, the submittal notes that fire zones equipped with only ionization detectors require dual zone detection prior to actuation of the preaction system in the :ontrol valves. The l
effects of relay chatter on inadvertent actuation of the fire protection system are mentioned in the submittal for two cases, but no systematic discussion is provided.
Some problems were noted in the submittal regarding the availability of the fire protection system following a seismic event. Examples include the low HCLPFvalue (about the SSE level) for the 480V load center that provides power to the motor-driven fire pump, the nonseismic masonry wall near the diesel-driven fire pumps (thejacket water expansion tank for the pump is mounted on the wall), and the lack of sufficient support for the starting batteries and the fuel oil day tank'. In addition, because the fire system provides protection for l
both safety-related areas inside the main power block and nonsafety areas outside the power complex (which include structures oflimited seismic capacity), the failure of the fire water lines in nonsafety structures may l
affect the ability of the fire water system to deliver flow adequate for the seismic demands for safety-related 1
1
'See Section 2.16 of this report.
i I
N areas. However, according to the submittal, the above is only a safety issue if a fire is initiated as a result of the earthquake, which does not appear to be a very probable event.
In conclusion,'the examination of seismic / fire interaction in the IPEEE and the discussion provided in the
. submittal seem to be reasonably thorough and adequate.
Besides the statement that "The walkdowns were consistent with the guidance provided in EPRI NP-6041,"
seismic-induced flood is not specifically discussed in the submittal. However, more information is provided in the licensee's response to the RAI. According to the response, potential internal flooding sources were evaluated during the walkdowns, and the walkdown team did not identify any potential hazard from seismic-induced internal flooding except for the potential failure of the piping and heat exchangers in the heat exchanger vault area. However, the walkdown team determined that area annunciation was adequate to allow the plant operators to respond long before flooding became a concern.
Also, according to the response, there are no significant storage volumes of water (such as major dams) upstream of TMI. Although the fire protection altitude tank was identified as a potential onsite external flooding hazard, a review by the seismic walkdown team confirmed that tank failure due to seismic activity does not have the poteuial to impact safety-related equipment.
2.12 LogicModels Logic Models: A seismic logic model was developed in the IPEEE to determine plant responses to seismically initiated events. As mentioned previously, the seismic logic model includes a seismic pre-tree model developed for IPEEE and the event tree logic models developed in the IPE for the Level 1 internal event PRA, Component fragilities were developed in the IPEEE and used in the constmetion of the seismic pre-tree and the determination of the split fractions for the top events in the tree. Of the event tree models (or event sequence diagrams, ESDs) developed fcr the internal event PRA (for general transient, steam generator tube rupture, and medium and large LOCA), only that associated with general transient is used in the seismic analysis. This is because the general transient tree with turbine trip as the' initiating event (i.e.,
caused by the earthquake) is the event tree that closely models the course of sequences occurring after a seismic event.. It should be noted that, in the IPE, the general transient event tree model is used for both transient and small LOCA initiating events.-
In the IPEEE seismic analysis, seismically it itiated events enter the seismic pre-tree where the seismic top event's success or failure is questioned. Impact from the success and failure of seismic top events are then propagated through the appropriate Level 1 PRA models for sequence determination. In this way, both the contributions of failure due to seismic events and due to independent or random events are captured in the analysis. Recovery from 'a loss-of-offsite power, which is modeled in *.he internal event PRA, is removed from the seismic analysis. This is to account for the possible extensive damage caused by the seismic event on the offsite grid and plant switchyard.
Initiating Events: The initiating events used in the seismic analysis are the peak ground acceleration (pga) levels. They are based on the site seismic hazard study performed by EPRI and reported in NP-6395-D. The four initiating events used in the seismic analysis have pga values of 0.15g,0.25g,0.40g. and 0.60s, and the corresponding frequencies for the four events, determined from the EPRI hazard estimates, are 4.48E-4/yr, 3.25E-5/yr,1.44E-5/yr, and 3.76E-6/yr, respectively. The pga ranges covered by the four initiating events are 0.052g - 0.2g,0.2g - 0.3g,0.3g - 0.5g, and 0.5g - 1.0g, respectively. The upper bound of the frequency 6
1 range of 1.0g used in the IPEEE is lower than the 1.5g cutoff value described in NUREG-1407. According to NUREG-1407, sensitivity studies should be conducted to determine whether the use of a lower cutoff affects the delineation and ranking of seismic sequences. This was not done in the TMIIPEEE. It is stated in the submittal that for the EPRI seismic hazard curve "the upper bound ground acceleration at Three Mile Island is 1.0g" and that stmetures and equipment "that have median acceleration capacities greater than 1.0g are generally excluded."
Data and Quantification: Split fractions for the seismic pre-tree top events were developed in the IPEEE for the four seismic initiators.
Conelations of seismic failures are considered in the analysis. Two train systems thet use similar components and are located at the same elevation in the same building are assumed to have a correlation factor of 1.0. Systems that fall into this category include the Class IE power system (including the diesel generators), vital instrument power, decay heat river water pumps, and nuclear river water pumps.
Quantification of the seismic logic model was performed in the IPEEE using the RISKMAN computer code.
The total seismic CDF, as well as co,ditional CDFs obtained for the four seismic initiators (for the four seismic levels), are presented in the submittal.
As a sensitivity study, quantification results were also obtained using the revised LLNL seismic hazard curves (NUREG-1488). In the sensitivity study, the four seismic initiating events remain the same (i.e., with the same pga values), but the frequencies of the four initiating events are changed based on LLNL hazard curves. The frequencies used in the sensitivity study are 9.53E-4/yr,7.75E-5/yr,3.84E-5/yr, and 1.31E-3/yr for the 0.15g,0.25g,0.4g, and 0.6g initiating events, respectively.
2.13 Accident Frequency Estimate Accident sequence quantification was performed using the RISKMAN computer code. The total core j
damage frequency (CDF) from seismic initiating events is 3.21E-5/yr. Total unaccounted frequency from the summation of all sequences that were truncated below a cutoff frequency of IE-12 is 4.56E-8/yr, or 0.01 percent of tctal CDF. Contributions from the four seismic initiators are also provided in the submittal.
They are 5.78E-6/yr (or 18 percent tual CDF),1.04E-6/yr (32 percent),1.22E-5/yr (38 percent), and 3.71E-6/yr (12 percent) for the four seismic initiators with pga of 0.15g,0.25g,0.40g, and 0.60g, respectively. The conditional core damage probability (i.e., given that the seismic event occurred) for the four seismic levels
{
are 0.013,0.32,0.85 and 0.997, respectively. This indicates that core damage is almost assured for a seismic event with a 0.6g pga. The top 40 sequences are presented in the submittal.
The total CDF obtained from using the LLNL hazard curves is 8.43E-5/yr. The contributions from the four initiators are 1.26E-5/yr (or 14.9 percent of total CDF),2.61E-5/yr (31.0 percent),3.25E-5/yr (38.6 percent),
and 1.31E-5/yr (15.5 percent). The top 12 sequences are presented in the submittal. They are,in general, consistent with those obtained in the base case (i.e., using the EPRI hazard curves).
Sensitivity studies were performed in the IPEEE for the following cases: (1) increasing human action failure rates, (2) no recovery actions, (3) no relay chatter, (4) guaranteed failure of relay chatter with no recovery, (5) no failure of a penetration pressurization air tank (which was found not to be seismically restrained ud its failure would not affect CDF but could change the release mode),(6) no excessive RCS LOCA due 'o seismic failure of RCS piping or major component, (7) no seismic failure of control room ceiling, (8) control 7
i j
f room ceiling failure results in total failure of Class IE power,(9) reweld of 480V load centers IP, IR, IS, and IT, and (10) use of NUREG-1488 hazard curves. Except for the last three sensitivity studies (sensitivity cases 8 to 10), the effects of the sensitivity cases on total CDF are small. As shown above, the use of the NUREG-1488 hazard curve results in an increase of total CDF from 3.21E-5/yr to 8.43E-5/yr, an increase of 163 percent. However, sequences and importance rankings are similar to the base case. The sensitivity l
studies also show that the total CDFis reduced by 55 percent with load center I P, I R,1 S, and 1T gusset weld modification and increases by 4:l percent ifit is assumed that control room ceiling failure results in a station blackout. The latter is believed to be overly conservative because the ceiling tends to fail at the walls and the damage would be localized and recoverable..
2.14 Dominant Contributors The top 40 sequences au presented in the submittal. The top ten sequences account for over 51 percent of total CDF, and all of them involve seismic failure of Class IE power. The leading sequence, which contributes 12.8 percent total CDF, is a sequence with seismic failure of Class IE ac power with offsite power availablo This is followed by a loss-of-offsite power (LOOP) sequence with control room ceiling failure and failure of Class 1E power (8.7 percent). The third leading sequence is a sequence similar to the leading sequence but with a different seismic initiator (7.7 percent), and the fourth sequence is another LOOP sequence with failure of Class IE power (6.6 percent).
System (or top event) and seismic component Fussel Vesely contributions for seismic initiating events are provided in the submittal. For system importance, seismic failure of offsite power contributes to 51 percent of the total CDF, seismic failure of Class lE switchgear with offsite power failed contributes 47 percent, seismic failure of Class 1E switchgear with offsite power available contributes 48 percent, and emergency feedwater failures occur in sequences accounting for 5 percent of the total CDF2. For seismic component importance, the leading contributor is the failure of Class IE ac power load centers with offsite power available (44 percent). This is followed by the contributions from the failure of Class IE ac power with offsite power failure (15.7 percent), the failure of offsite poiver insulators (12.1 per cent), the failure of control room ceilings (7.3 percent), and the failure of emergency diesel generator ai. start receivers (5.9 percent).
Recovery action importance is also discussed in the submittal. They are in general small. The leading contributor, according to the Fussel Vesely importance measures obtained in the analysis, is that from relay chatter recovery for the 0.40g seismic initiating event (2.54E-3). This is followed by relay chatter recovery for the 0.25g seismic imtiating event (1.27E-3). The contribution from independent system failures (i.e., not from seismic) as indicated by th:: Fussel-Vesely measures obtained in the IPEEE is also small. The leading contributoris independent failure of the Class IE ac power train A (7.60E-3).
Importance measures were also obtained in the sensitivity study for the LLNL hazard curve cases. The rankings are, in general, consistent with those obtained in the base case (i.e., using EPRI hazard curve).
i 2There are some minor inconsistencies between the values cited in the discussion (p. 3-141) and those presented in the table (Table 3.1-14). The values used here are those provided in the discussion.
8
/
e
e l
1~
2.15 Unresolved Safety Issues and Generic Issues USI A-45 Shutdown Decav Heat Removal Reauiremente The licensee concluded that no seismic weaknesses of the decay heat removal systems were discovered in the seismic evaluation. (Section 3.2.1 of the submittal)
GSI-131 Potential Seismic Interaction Involving the Movable In-Core Flux Manoine System Used ja.Eestinehouse Pinate
]
GI-131 is only applicable for Westinghouse plants.
GSI-156 Systematic Evaluation Pronram (SEP)
The seismic-induced settlement of foundation was addressed in the submittal (Section 3.1.1.6); the potential dam failure and site flooding were addressed in the mbmittal (Section 5.2); seismic design of structures, systems, and components was addressed throughout Section 3.1 o'the submittal.
GSI-172 Multiole System Resoonse Pronram (MSRP)
The seismic II/I spatial interactions were addressed in the analysis. (Section 3.2.2 of the submittal)
- Fire suppression actuation was discussed in the submittal. (Section 3.2.2 of the submittal)
Seismically induced fires and seismically induced actuation of fire protection systems were e
addressed in the submittal. (Section 4.8.1 of the submittal)
Seismic induced flooding was discussed in the RAI responses, specifically in the response to the fifth seismic RAI.
Failures related to human errors were discussed in several pans of Section 3.1.5 of the submittal.
Seismic induced relay chatter was addressed in Section 3.1.4.1 of the submittal.
The evaluation of earthquake magnitudes greater than the safe shutdown earthquake can be considered as being covered by the information in Sections 7 and 8 of the submittal. Section 8 notes that the analysis identified no vulnerabilities, and the analysis assumed one of the improvements listed in Section 7, i.e., enhancement of the emergency diesel generator day tank supports, to be in place.
2.16 Vulnerabilities /PlantImprovements According to Section 3.1.5.6 of the submittal, a vulnerability is defined as in the IPE,i.e., as any core damage sequence greater than 1x10d/yr or any containment bypass or large early containment failure greater than 4
1 x10 /yr. As a result of the seismic PRA analysis, no vulnerabilities have been identified. However, several plant improvements are assumed in the risk quantification or planned for the fuwre.
The plant improvements that were assumed to be in place in the seismic PRA analysis include:
. All relays that cannot pass any seismic screening criteria will be replaced during the upcoming refueling outage (Section 3.1.4.1); and Strengthening the supports for the Class 1E emergency diesel generator day tank.
1 I
9 4
e 1
In response to an RAI on the status of the relays to be replaced, the licensee provided the following:
1 "The IPEEE relay review identified the possibility of a seismic event starting the reactor building spray j
pumps with the pump suction and discharge valves closed. The chatter of contacts in the engineered safeguards relays could cause the pumps to stan. These relays, Clark Model SU 12-1 1, were evaluated in 1996 as being seismically acceptable per Farewell & Hendericks, Inc., Repon No. 50090.8, Rev. 0 and GPU Nuclear Calculation, C-1 101-900-5320-025.
i Regarding the acceptability of other relays, TDR 1185, " Relay Repon for Three Mile Island", dated December 28,1995, was generated to document the methodology, documentation, and results of the relay evaluation ponion of the USI A46 program. The latest status of the identified relays is as follows:
EG-Y-0001A l
Relay: K1
)
Manufacturer:
Westinghouse Model:
MD101 Location: AKM/CNPL Building:
DGB,305-0 1
Relay: K2 Manufacturer:
Westinghouse Model:
MD120 Location: AKM/CNPL Building:
DGB,305-0 Status:
Relays found to be acceptable per GPU Nuclear Calculation C-1101 -900-5320025, Rev. 0. No need to replace relays based on this evaluation.
EG-Y-0001B Relay: K1 Manufacturer:
Westinghouse Model: MD101 Location: AKM/CNPL
{
Building:
DGB,305-0 Relay: K2 Manufacturer:
Westinghouse Model: MD120 Locationi AKM/CNPL Building:
DGB,305-0 Status:
Relays found to be acceptable per GPU Nuclear Calculation C-1101-900-5320025, Rev. 0. No need to replace relays based on this evaluation.
10 i
I
c IC-P-0001A Relay:
62X/IC-P-IA 7
Manufacturer: Agastat Model: 7022 Location:IA ES MCC Bu.ilding: CB,322-0 Status:
Relays found to be acceptable per GPU Nuclear Calculation C-1101-900 5320025, Rev. 0. No need to replace relays based on this evaluation.
IC P-000I A'
' Relay:
62X/IC-P-IA Manufacturer: Agastat Model: 7022 Location:IA ES MCC Building: CB,322-0 Status:
Relays found to be acceptable per GPU Nuclear Calculation C-1101-900-5320025, Rev. 0, No need to replace relays based on this evaluation.
Invener I A/IB/IC/ID Relay:
DC UV Sensing RL6 Manufacturer: GO Model:
R.V.-20 Location:
Invener I AllB/lC/ID Building: CB,322-0 Relay:
DC OV Sensing RL7 Manufacturer: GO Model:
R.V.-20 Location:
Invener I A/lB/IC/lD Building:
CB,322-0 Status: RL6 and RL7 replaced in October 1993 per EER 92-0436 Relay:
SCR Shtng Brd RL1
. Manufacturer:
Potter Brumfield Model:
- KHAU-17D 1 1 Location:
Invener I A/IB/IC/lD Building: CB,3224 Status:
Relay found to be acceptable per GPU Nuclear Calculation C-Il01 900-5320025, Rev. 0. No need to replace relay based on this evaluation."
11
r o
i A number of plant improvements were identified in the submittal (Section 7.2) as possible enhancements being evaluated, but were not accounted for in the IPEEE seismic analysis. These are:
1.
Load Centers IP, IR, IS, and IT gusset weld reinforcements, 2.
Modification of the control room ceiling, 3.
Restraining of the penetration pressurization tank PP-T-1 A to prevent potential impact on the purge line isolation valve; 4.
Modification of the supports for the fuel oil tanks and batteries for the diesel-driven fire pumps.
- 5.
Modification of the anchorage for the decay heat service heat exchangers, and 6.
Anchoring of the air receiver pedestals to the floor of the Class IE emergency diesel generator.
In the response to an RAI the licensee described the status of these improvements as follows:
1.
IP, IR, IS, and IT load center gusset welds - Load Centers 1R and IT are located in the screenhouse and are the basis for the IPEEE calculated core damage frequency (CDF) values.
Load Centers I P and IS are located in the control tower and their fragility values are judged to exceed the fragility values ofIR and IT as described below. A more complete visual inspection of the gusset welds was performed after the completion of the TMI-I IPEEE. The welds were evaluated by the GPU Nuclear civil / structural engineering staff and verified to be the weak link of the load centers as stated in the TMI-l IPEEE. Although these welds continue to be the largest single contributor to the calculated seismic CDF, a rewelding of the gusset welds to make another element the weak link for the above load centers would only lower the seismic CDF from 3.21 E-5 to 1.44E-5/ year. This is a reduction in total CDF of about 10% when both the intemal and external event contributions are considered (a chaage from 1.78E-4 to 1.60E-4/ year). Based on the engineering evaluations described above, the load center welds in the control tower are judged to be more seismically rugged than the load centers in the screenhve Since the seismic capability values for the load centers in the screenhouse were used in 9 nEEE, the CDF values above are judged to be conservative. For these reasons, it has been demined that a reweld of the load center gusset welds is not required.
2.
Control room ceiling modification - This was completed on October 7,1997.
3.
PP-T-IA seismic restraint - A restraint has been designed and was installed on August 27,1996.
4.
Diesel driven fire pump battery / oil tank seismic restraint - At present, there are no plans to upgrade these components. These are not major contributors to the seismic core damage frequency (<!% CDF). The motor driven fire pump will be available after the majority of seismic events.
5.
DHCCW heat exchanger seismic restraints - At present, there are no plans to upgrade these components. This has been concurred with by the system' engineer. The heat exchangers contribute to approximately 2% of the seismic CDF and the modification would incur a substantial cost and produce an obstmetion to maintenance on the heat exchangers.
6.
EDG air receivers seismic restraint - These modifications were completed on June 28,1996.
12
3.
OVERALL EVALUATION AND CONCLUSIONS The study addressed all the major issues that are emphasized in NUREG 1407, including plant walkdowns, relay chatter, liquefaction, nonseismic failure, human actions, recent developments in seismic hazard evaluations, and containment performance. The licensee appears to have satisfied the objectives outlined in the Generic Letter with respect to the IPEEE. Among the reasons for this finding are the following:
The study is consistent with the guidelines of NUREG-1407.
Detailed evaluations of relay chatter were provided.
Extensive identification and listing of components and equipment based on a series of walkdowns was carried out.
The evaluated component fragility values seem to be reasonable.
The selection of seismic top events are extensive and considered to be reasonable.
Detailed sensitivity studies were performed including the use of both the EPRI and the revised LLNL hazard curves.
Recovery actions (including those for relay chatter) were modeled in the risk quantification.
Containment performance was evaluated.
A correlation factor of one (1.0) was assumed for similar components in a given system.
The dominant sequences and contributing failures seem reasonable.
Lessons learned from the earlier seismic PRA studies were reflected in the submitted IPEEE study.
4.
REFERENCES
[1]
TMI Unit 1 Individual Plant Examination for External Events (IPEEE), December 1994, Attachment to Letter of December 29,1994 from R.W. Keaten of GPU Nuclear to USNRC.
l
[2]
Individual Plant Examination for External Events - Response to Request for Additional Information Attachment to Letter of April 24,1998 from J.W. Langenbach of GPU Nuclear to USNRC.
i i
[3]
Seismic Fragilities of Civil Structures at Three Mile Island Unit 1, EQE International, April 1994.
l l
l 13 1
I
I.
l Appendix B 1
I Fire Submittal Screening Review - Technical Evaluation Report: Three Mile Island Nuclear Station, Unit 1, Individual Plant Examination for Externs.1 Events, Dated September 1998 l
l 1
l L.
.