ML18107A315

Submittal-Only Screening Review of Salem Generating Station Individual Plant Exam for External Events (Seismic Portion), Rev 1

ML18107A315
Person / Time
Site:	Salem
Issue date:	04/30/1999
From:	BROOKHAVEN NATIONAL LABORATORY
To:
Shared Package
ML18107A309	List: ML18107A315 ML18107A319
References
NUDOCS 9905270215
Download: ML18107A315 (12)

Text

e*

SUBMITTAL-ONLY SCREENING REVIEW OF THE SALEM GENERATING STATION INDIVIDUAL PLANT EXAMINATION FOR EXTERNAL EVENTS (Seismic Portion)

Rev. 1 August.1998 (Finalized April 1999)

Brookhaven National Laboratory 9905270215 990521 .

PDR ADOCK 05000272 P PDR

- -- -- --*--~-- ....... -~-**---*-*._. ______.,. _____ ,,.------*----*----..-*-~------------,....-

. . . . .---* . .------=-*.-----***--.. . . . . -------,---*-* -~----**- ----------- .... ------------~----- ------- --- . **--------------------*- -..

• 1. INTRODUCTION 1.1 Purpose In response to the NRC issued Supplement 4 to Generic Letter (GL) 88-20, "Individual Plant Examination of External Events (IPEEE) for Severe Accident Vulnerabilities - 10 CFR 50.54(t). ", the Public Service Electric and Gas Company (PSE&G) performed an IPEEE for the Salem Generating Station (SGS) and submitted the IPEEE results to NRC [Reference!]. Brookhaven National Laboratory (BNL), as requested by NRC, performed a submittal-only screening review to verify the technical adequacy of the seismic portion of PSE&G's IPEEE submittal. As a result of this review NRC sent a Request for Additional Information (RAI) to PSE&G. PSE&G responded with the Response to RAI Regarding IPEEE: Salem Generating Station [Reference 2] in April of 1998. 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 RAI. .,__

BNL's methodology utilized for the review followed the guidelines provided in the document titled "Guidance for the Performance of Screening Reviews of Submittals in response to USNRC Generic Letter 88-20, Supplement 4" (Draft, Oct. 24, 1996), as modified by NRC.

1.2 Background The Salem Generating Station (SGS) is a two unit plant located on the southern part of the PSE&G reactor site located on the east bank of the Delaware River in Lower Alloways Creek Township, Salem County, New Jersey. Each unit employs a Westinghouse 4-loop PWR rated at a thermal power of 3411 MW. The NSSS is enclosed by a large, dry, reinforced concrete, steel-lined containment. The SGS is built on an artificial island, and most of the Seismic Category I structures are founded on a common lean concrete mat poured within the confines of a cellular coffer dam. The lean concrete mat was used to replace soil with lower bearing and shear capacities. SGS Unit 1 began commercial operation in June 1977, and SGS Unit 2 in October 1981.

The Safe Shutdown Earthquake (SSE) for the site is 0.20g, and the plant is binned in the 0.3g focused -

scope review category. For design dynamic time history analyses of Seismic Category I structures, the 1940 El Centro ground motion records, scaled to 0.20g peak acceleration, were used.

1.3 Licensee's IPEEE Process and Licensee's Insights A Seismic Probabilistic Risk Assessment (PRA) approach *was used to identify potential seismic vulnerabilities. The seismic analysis addressed the major issues described in NUREG-1407, i.e., plant

• walkdowns, human actions, relay chatter, soil liquefaction, and containment performance. The risk quantification was performed using both the EPRI and the revised LLNL hazard curves.

The overall PRA process appears to be consistent with the PRA methodology described in NUREG-1407.

The system analysis was performed using the NUS SEISMIC code to quantify the frequency of the seismic damage states (SDS), and the NUS PRA Workstation code to calculate the conditional core damage 1

• l'

, probabilities. The non-seismic failures were considered using the IPE internal event model.

The results of the IPEEE identify no vulnerability due to seismic events. The calculated ~ore damage frequency (CDF) using the EPRI hazard curves is 4.7 x.10-6/yr, and 9.5 x 10~6/yr using the revised LLNL hazard curves. The dominant sequences involve offsite power loss, non-seismic failure of the diesel generators, failure of the service water system, and the collapse of the control room ceiling.

2. REVIEW FINDiNGS 2.1

• IPEEE Format and Methodology Documentation The submittal appears to be consistent with the guidelines of NUREG-1407. Th~ study addressed all the

• issues that are emphasized in NUREG-1407, including plant walkdowns, relay chatter, soil liquefaction, nonseismic failure, human actions, and containment performance. The completeness of the documentation is adequate, although some additional information, obtained via an RAI request; was needed to complete the review of the seismic analysis.

2.2 Seismic Review Team Selection The seismic review team (SRT) consisted of personnel from PSE&G, who were familiar with the system and operation, as well as external consultants, including EQE International for the seismic walkdown, fragility analysis, and relay assessment, NUS Inc. for the system analysis and risk quantification, and Woodward-Clyde Consultants for the soil evaluation. 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, that were truncated at 0.8g and l.Og, respectively. To study the effects of the cutoff, a sensitivity study was performed by extrapolating the LLNL hazard curves up to a 1.4g level. The results of the additional analysis indicate an approximately 25 percent increase in the overall plant CDF due to the extension of the hazard curves.

However, no unique or new plant vulnerabilities were identified as a result of the sensitivity study.

For performing structural response analyses, the median uniform hazard spectra (UHS) at a 10,000 year return period, based on the EPRI hazard analysis, were used. The use of such a spectrum is acceptable for purposes of the IPEEE according to NUREG-1407, 2.4 Components Selection A total of over 300 components were selected for initial screening as listed in Table 3 .4 of the submittal.

The listing is extensive but a few important items are missing, including pipe support structures, batteries and racks, and control rod drive mechanisms (CRDM). After seismic walkdowns, a total of about 100 items were screened in for further evaluation. The components screened out include the reactor coolant pumps (RCP), steam generators (SG), and the reactor vessel and internals. A median capacity of 1.5g and a High-Confidence-Low-Probability-of-Failure (HCLPF) of 0.5g were used as part of the screening 2

.- -~** -- ......--- .... **--------*-----*-- *------------~----------------.-*-------- ....-~.._..------...-.-~*-~-~---*----*._,....,.. _____________ -

l '

' 'criteria. As described on page 3-25 of the submittal, three components with a median capacity of 0.9g were not included in the component list because they could be isolated by valves or were not needed for a safe shutdown. In table 3.6 of the submittal, most the components whose failure may potentially cause a LOCA event, were ~creened out largely based on a generic screening criteria. However, in the RAI response the licensee provided an estimate regarding the fragility parameters* used for deriving the small LOCA conditional probability in the IPEEE analysis and this estimate appears reasonable. In the RAI responses the licensee also substantiated the fact that the Regenerative and Letdown Heat Exchangers could be screened from the small LOCA analysis.

2.5 Plant Walkdowri Approach The submittal states that a* series of walkdowns were performed during refueling outages to develop a component list, to pre-screen components with high seismic capacity, to determine the failure modes, to identify spacial interaction, and to evaluate the likelihood of a seismic induced fire and the potential actuation of fire protection systems.

The walkdown team consisted of personnel from PSE&G (plant knowledge and PRA expertise),

Halliburton NUS (system analysis), and EQE International (structural analysis). For each of the components listed in the submittal, a Seismic Evaluation Walkdown Sheet (SEWS) was prepared to record the walkdown findings. The walkdown procedure seems to be appropriate.

2.6 Fragility Analysis 2.6.1 Structural Response Analysis New probabilistic soil-structure (SSI) analyses were performed for the containment building including internal structures, auxiliary building, and the service water intake structure. Variabilities in stiffnesses and damping of both structures and soil were considered in the analyses based on a Latin Hypercube Simulation. The calculated floor spectra are provided in the submittal. It is stated that a significant reduction of spectral values has been achieved from the original design spectra. It is implied that the amount ofreduction is at least 75 % of the design spectral values throughout the calculated frequency range.

Since the original design floor spectra were not provided in the submittal, it was not possible to confirm the implied reduction. The use of EPRl's UHS, which has a significantly lower spectral value at lower frequency ranges (lower than 10 Hz) than typical design response spectra, may have caused this large reduction in structural responses.

2.6.2 Structural Fragility Analysis The structural fragility analysis was performed by EQE International, and the results are tabulated in Table 3.5 of the submittal. Calculated median fragility values appear very high considering the design SSE level of 0.2g, but are probably consistent with the significant reduction in the calculated floor responses described above.

According to Figure 3 .12 of the submittal, the liquefaction of slopes due to lateral spreading appears to be 3

)

1 L

I

, initiated at about a peak ground acceleration of 0.35g. However, according to Table 5.6 (p. 3-:70) of the submittal, the HCLPF capacity of buried piping was estimated to be 0. 72g. It is not clear from the submittal why the lateral spreading does not affect buried piping.

2.6.3 Component Fragility Analysis As stated on page 3-15 of the submittal, a generic fragility capacity of components is estimated to be l.96g

• in terms of median ground acceleration based on the newly obtained floor response spectra. This would imply that components have a median seismic margin of about 10.0 over the design SSE level of 0.20g.

Regarding the fragility evaluation, requests for additional information were made to PSE&G in the following areas:

1) The estimated strength factor of Fem = 2.53 for components appeared to be too optimistic for application to a wide range of generic components, particularly to rigid components with a non-ductile failure mode.

2) The anchorage failure of some components (e.g., DFUELTNKll, ICA-EC/ED) was evaluated ba_sed on the PRA database. A plant-specific analysis/evaluation may be more appropriate for this type of failure mode.

3) The same fragility value was assigned to the Exhaust Fans installed at different elevations (64',

100', and 122 ') of the Auxiliary Building.

In the RAJ response, the licensee clarified that the safety factor Fem changed from 2.53 to 2.35, and resolved point (3) above, i.e., the issue of the fragility values of the exhaust fans. However, the RAJ responses did not resolve the concerns of points (1) and (2) and these still represent weaknesses of the analysis.

2. 7 Soil Evaluation The liquefaction potential was assessed by Woodward-Clyde Consultants using a probabilistic approach.

A HCLPF capacity of 0.72g was estimated, which was used to evaluate the fragility of buried piping.

According to Fig. 3.12 of the submittal, the lateral spreading due to liquefaction of slopes becomes significantly large at a peak acceleration of about 0.35g. Whether or not this information has been used for the fragility evaluation of buried piping is not clearly described in the submittal.

2.8 Relay Chatter Evaluation Approximately 100 potentially low ruggedness relays (LRR) were identified. Some of the identified LRR's have been replaced with higher seismic capacity relays, including the 4kV Phase A/B/C diesel generator differential relays (p. 3-31 of the submittal). All other relays were screened out because (i) LRR are not associated with safety shutdown or containment performance (ii) relay chatter is acceptable (iii) the LRR have high seismic capacity (section 3.1.5.4.3 of the submittal). This evaluation procedure appears to be acceptable as long as the identified LRR's have been, in fact, replaced by higher capacity relays.

The licensee did not incorporate relay chatter into the PRA model. Instead, a screening review at the 0.3g

• 4

level (which was the plant's review level earthquake) was performed. This approach may somewhat underestimate the core damage* frequency at higher g levels.

2.9 Containment Performance The effects of seismic events on the containment performance have been evaluated regarding structural capacity, containment isolation, and containment bypass. The evaluations were performed based on seismic walkdowns and capacity calculations of components. No vulnerabilities were found regarding any aspect of the containment performance as stated in Section 3.1.6.9 of the submittal.

The containment structure and appurtenant structures (e.g., reactor coolant system supports, main steam lines, nearby structures) were determined to have high seismic capacity. Containment isolation was evaluated by reviewing containment penetrations and isolation valves and piping. Again, no outliers were found. The containment bypass potential was evaluated by reviewing the seismic capacity of the isolation valves and lines and associated relays. No vulnerabilities were found.

Also, no vulnerabilities were found with respect to containment hatches (the plant does not use inflatable

• seals on the hatches), containment isolation actuation and containment pressure suppression and heat removal. Fan coolers were found to be apparently more fragile than other components (the wording in the submittal is confusing), but they were not found to be important in the analysis. Loss of support systems (instrumentation, control and electric power) were found to be the most important contributors to. failure of containment performance.

• Early containment failure occurs with a relatively low frequency, .because over 50 % of the core damage sequences result in a long term station blackout (SBO) (6 hours6.944444e-5 days <br />0.00167 hours <br />9.920635e-6 weeks <br />2.283e-6 months <br /> to well over 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> after the seismic event) due to such delayed effect failures as room cooling failures or failures of the diesel generator transfer pumps. Even for sequences in which the SBO occurs immediately, containment failure is not expected to occur for several hours.

No containment failure vulnerabilities were found by the licensee. It should be noted that no fragility

• parameters or HCLPF values for containment related components were provided in the submittal (except for the containment structure).

2.10 Nonseismic failures and human actions Nonseismic failures and human actions were incorporated into the model from the internal events IPE PRA.

The submittal states that diesel generator mission time was increased to 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br />. No power recoveries

• were allowed within the first 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br />. However, most human error probabilities were kept at the same values as in the IPE model, due to ample time being available (such actions dealt mainly with restoration of rooin cooling). Procedures will also be updated for such actions, and a limited sensitivity study was done. However, there was no discussion of any accessibility problems. The human error probability (HEP) for initiation of feed and bleed cooling was also kept the same as in the IPE, even though the time scale is much shorter (30 minutes), and the explanation given in the submittal for not changing the HEP value was not very clear.

In general, the HEP treatment seems to be somewhat optimistic.

5

"---*-----. - .,-... ----------------~---~-~~-*~.~~-.,..., ----~---------------~- ....----*----------- -

• 2.11 Seismic Induced Fires/Floods This* issue was* handled as part of the walkdown.

The licensee did identify several items of concern, but there were always mitigating considerations to screen the item out. For example, inadequate separation between redundant pressurizer power operated relief valve (PORV) cables was screened out based on lack of fire sources that could be ignited by a seismic event. Radiant shields in electrical penetrations were determined -to be rugged and there were no seismically induced fire sources, so that lack of separation was not a concern. There were some concerns about waste oil spillage from the reactor coolant pumps (RCPs) in the vicinity of the PORV cables.

However, the autoignition temperature of waste oil is 200°F higher than the RCP motor design temperature, and spillage was unlikely to occur due to the seismic capacity of the spillage sources.

All the potential flammable sources (gases, liquids) were screened out based on high seismic capacity, although, again,, no HCLPF values were provided in the submittal.

The seismic capacity of fire protection equipment was also evaluated during the walkdown; however, such equipment would not be needed as there would be no seismically induced fires.

Similar conclusions were reached with respect to flooding and spray interactions. Fire protection sprinklers were evaluated for interactions with nearby objects, and the seismic flooding potential from tanks was reviewed._ However, seismic flooding interactions from the various tanks, and the component cooling water and service water systems, are not discussed in detail in the submittal. Also included in the evaluation was chatter of relays causing inadvertent actuation of the fire protection equipment. No vulnerabilities were found. There was no discussion of any external flooding due to river water, etc .

.caused by seismic events.

It seems the licensee considered most items of interest in this section, with the exceptions noted above.

2.12 Logic Models The analysts did not credit recovery of electric power within the first 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br />. Recovery of room cooling by some alternate operator action is considered, but this action is not described in detail (such action should not depend on service water, for instance). Unique seismic failures were considered in the logic model.

These included failure of the control room ceiling causing loss of instrumentation and control and therefore leading directly to core damage; failure of batteries incapacitating the diesel generators; and failure of the diesel generator fuel transfer system, etc.

_Offsite power recovery via the gas turbine is not credited. The gas turbine is connected through the ring bus and major transformers, therefore failure of the ceramic insulators will .disable this means of recovery.

The. seismic event tree was not described in detail, however, its structure seems to be logical.

Treatment of passive failures and systems interactions was considered and included, but there is not much

.discussion, or examples given, of problems found.

6

• -*- - **--~-.*--~--..--..--*- *~------.-.-----_....,.~~~~-----------~--.---------- ---...,.......--------*--- ... --*---*----. ---~-~ -*

l .

The treatment of human error probabilities (HEPs) seems op~imistic.

The mathematical treatment of the quantification is not discussed in the submittal. It is not clear whether seismic bins were used, or whether somehow all seismic acceleration levels were combined for the analysis. The results are not broken down into seismic bins.

It is not clear what the cutoff for cutset quantification was. The cutoff for the seismic damage state quantification (whose results are input into the internal IPE model for final quantification) was l .E-7/yr.

Depending on the conditional core damage probability of the scenario in question, this may be optimistic (i.e., underestimates the core damage frequency for station blackout, for example).

It is not clear what missiori time was used for the overall analysis. The submittal states that the mission time for the diesel generator was extended to 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> but this mission time may not coincide with the overall PRA mission time). In another statement, some core damage sequences lead to SBO in "well over 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br />".

The construction of logic m*odels seems to have been generally correct, with some minor concerns as noted above.

2.13 Accident Frequency Estimate The seismic core damage frequency is 9.5E-6/yr, using the LLNL seismic hazard curves.

An uncertainty analysis was not performed.

Sensitivity analyses were performed with respect to EPRI vs. LLNL hazard curves, and extending the upper PGA cutoff to 1.4g. A HEP sensitivity was also performed. None of these analyses resulted in a large impact on the core damage frequency.

Using the EPRI hazard curves reduced the CDP by a factor of 2. Extending the LLNL hazard curves to 1.4g (which the licensee states has no geotechnical basis) increases the CDP by 25 %. The HEP sensitivity analysis increased the CDP by about 30%.

2.14 Dominant Contributors The four dominant sequences, representing 78% ofthe seismic CDP, are the following:

1) loss of offsite power combined with random failures of the diesel generators and associated support systems, CDF=2.9E-6/yr;

2) loss of offsite power and seismic failure of the service water system, resulting in diesel generator failure, CDP=l.3E-6/yr;

3) loss of offsite power and battery trains A&B (due to block wall failure), leading to failure to start of diesel generators A and B and of the fuel oil transfer pumps; the latter failure leads to eventual failure of diesel generator C, CDP=2.0E-6/yr; 7

I ---** --- **- - ..... .....---.---*-*---~*--*~* ... ------~ *------.... ~-*---*-*------------- ... --. ~-----**.*-*--*--,-----*-*-*-*-*---*--*- *-~------------

4) loss of offs.ite power and seismic failure of instrumentation and control due to ceiling collapse in the main control room, CDF=l.2E-6/yr.

No importance analys~s seem to have been performed.

2.15 Unresolved Safety Issues and Generic Issues USI A-45 Shutdown Decay Heat Removal Requirements No seismic vulnerabilities of the decay heat removal systems were found in the analysis. (Section 3.2.1)

GSI-131 Potential Seismic Interaction Involving the Movable In-Core Flux Mapping System Used in Westinghouse Plants Potential seismic interaction involving the Movable In-Core Flux Mapping System was examined as part of the IPEEE study, and no seismic vulnerabilities were found. (Section 3.2.7)

GSI-156. Seismic Evaluation Program CSEP)

The seismic induced settlement of foundations was addressed in the submittal (Section 3 .1.1.1); potential dam failure was not discussed but the site flooding was addressed in the submittal (Section 5.5); seismic design of structures, systems and components was addressed in the submittal (Section 3.1.2.2).

. GSI-172 Multiple System Response Program CMSRP)

GSI-172 issues were addressed in the IPEEE submittal as follows:

• Seismic induced spatial and functional interactions were addressed in the analysis. (Section 3 .2.3 of the submittal)

• Seismically induced fires were addressed in the submittal. (Section 3 .1. 7 and Section 4. 8 .1 of the submittal) *

• Seismically induced actuation of fire protection systems were addressed in the submittal. (Section 3 .1. 7 and Section 4. 8 .1 of the submittal)

• Seismic induced flooding was discussed in the submittal. (Section 3 .1. 7)

• Failures related. to. human errors were discussed in the submittal. (Sections 3.1.5.3.2 and 3.1.5.6.3)

• . Seismic induced relay chatter was addressed in Section 3.1.5.4.3 of the submittal.

• Hydrogen line ruptures were considered as part of the issue of seismically induced fires in Section

4. 8 .1.1 in the submittal.

2.16 Vulnerabilities/Plant Improvements No definition of vulnerability was found in the submittal. The submittal does state that as a result of the seismic PRA analysis, no vulnerabilities have been identified. However, a few plant improvements were assumed and credited in the risk quantification.

8

'l

• A procedural change to ensure long term alternate ventilation for rooms in the Auxiliary Building, was scheduled for implementation by Ju1y 1, 1996. (Section 7.1 of the submittal).

• Some of .the identified low ruggedness relays were replaced with higher seismic capacity relays, including the 4kV Phase A/B/C diesel generator differential relays (p. 3-31 of the submittal).

• An eight-foot tall masonry wall in the 4kV switchgear room will be reinforced, although it was judged not to be a credible source of seismic interaction (Section 7 .1 of the submittal).

3.0 OVERALL EVALUATION AND CONCLUSIONS The submittal appears to be consistent with the guidelines of NUREG-1407 in applying seismic PRA methodologies. The study addressed most of the major issues that are emphasized in NUREG-1407, including plant walkdown, relay chatter, liquefaction, nonseismic failure, human action, recent developments in seismic hazard evaluation, and containment performance.

The completeness of the documentation for most of the major issues appears to be adequate.

The treatment of the seismic hazard curves seems to be adequate. The study used both the EPRI and the revised LLNL hazard curves, and a sensitivity study was performed for hazard curve cutoff.

The PRA modeling seems to have included most issues of concern and the dominant sequences and contributing failures seem reasonable.

Two significant weaknesses in the analysis remain:

• The estimated strength factor of Fem, although reduced to 2.35 from 2.53 as a result of an RAI, still appears to be too optimistic for application to a wide range of generic components, particularly to rigid components with a non-ductile failure mode.

• The anchorage failure of some components (e.g., DFUELTNKll, ICA-EC/ED) was evaluated based on the PRA database. A plant-specific analysis/evaluation would appear to be more appropriate for this type of failure mode.

Some minor weaknesses in the submittal are that relay chatter was not included in the PRA model, and that seismically induced internal and external flooding was not discussed in detail~

However, overall the licensee appears to have satisfied the objectives outlined in the Generic Letter with respect to the IPEEE.

9

. *--*-* ~----- -~*-*- *------*-------*--~- ****---*.-...--'--------.. . . .....--.---.. . ~-.......---....,....~---~,..--===-====-= -*- - -- -** *------~-*---**..----- ----*------* -----*- --*------ --*~ --------* ---- ---

4.0 REFERENCES

[1] Salem Generating Station Individual Plant Examination for External Events, Attachment to Letter dated Jan. 29, 1996 from E. Simpson, Sr. Vice President, Nuclear Engineering, Public Service Electric and Gas Company, to USNRC.

[2] Response to Request for Additional Information IPEEE, Salem Nuclear Generating Station, Units 1and2, LR-N980175, Attachment to Letter dated April 9, 1998 from E. C. Simpson, Sr. Vice President, Nuclear Engineering, Public Service Electric and Gas Company, to USNRC.

10

I r Attachment 2 Salem GENERATING STATION INDIVIDUAL PLANT EXAMINATION OF EXTERNAL EVENTS (IPEEE)

TECHNICAL EVALUATION REPORT FIRES