ML16349A604
ML16349A604 | |
Person / Time | |
---|---|
Site: | Hope Creek |
Issue date: | 12/14/2016 |
From: | Carr E Public Service Enterprise Group |
To: | Document Control Desk, Office of Nuclear Reactor Regulation |
References | |
EA-16-184, LR-N16-0232 | |
Download: ML16349A604 (15) | |
See also: IR 05000354/2016003
Text
PSEG Nuclear LLC P.O. Box 236, Hancocks
Bridge, New Jersey 08038-0236
DEC 14Z016 LR-N16-0232
U.S. Nuclear Regulatory
Commission
Attn: Document
Control Desk Washington,
DC 20555-0001
Hope Creek Generating
Station Renewed Facility
Operating
License No. NPF-57 Docket No. 50-354 PSEG NttcleaT
LLC Subject:
Response
to Preliminary
White Finding in Integrated
Inspection
Report No. 05000354/2016003;
Reference:
Hope Creek Generating
Station Unit 1 -Integrated
Inspection
Report 05000354/2016003
and Preliminary
White Finding,
November
14, 2016. By letter dated November
14, 2016 (Reference),
the U.S. Nuclear Regulatory
Commission
(NRC) issued Inspection
Report 05000354/2016003
completed
on September
30, 2016. The inspection
report identified
a preliminary
White finding and associated
apparent
violation
of Title 10 CFR Part 50, Appendix
B, Criterion
V, "Instructions,
Procedures,
and Drawings,"
asserting
that PSEG did not adequately
implement
an adverse condition
monitoring
procedure,
specifically
for performing
monthly oil sampling
of the High Pressure
Coolant Injection
(HPCI) system. As stated in the November
14, 20161etter,
PSEG has the option to submit additional
information
regarding
the significance
determination
of this preliminary
finding.
Accordingly,
we are submitting
the attached
additional
information
supporting
our position.
There are no regulatory
commitments
associated
with this submittal.
If you have any questions,
please contact Mr. Thomas MacEwen at (856) 339-1097. Eric Carr Site Vice President
Hope Creek Generating
Station ttm Enclosure:
Additional
Information
Submitted
Pursuant
to Inspection
Report 05000354/2016003,
Preliminary
White Finding
LR-N16-0232 Document
Control Desk Page2 cc: Mr. D. Dorman, Regional
Administrator,
NRC Region I Ms. C. Parker, Project Manager-Hope Creek Mr. J. Hawkins,
NRC Senior Resident
Inspector-
Hope Creek (X24) Mr. P. Mulligan,
Manager IV, NJ Bureau of Nuclear Engineering
Mr. T. MacEwen -Hope Creek Commitment
Coordinator
(H02) Mr. L. Marabella
-Corporate
Commitment
Coordinator
(N21)
LR-N16-0232
Document
Control Desk Enclosure
Additional
Information
Submitted
Pursuant
to Inspection
Report 05000354/2016003,
Preliminary
White Finding Contents
1.0 Summary 2.0 Review of PRA Analysis
3.0 Review of Sensitivity
Case 4 4.0 References
LR-N16-0232
Enclosure
Additional
Information
Submitted
Pursuant
to Inspection
Report 05000354/2016003,
Preliminary
White Finding 1.0 Summary NRC Finding Summary The inspection
report describes
a self-revealing
preliminary
White finding and apparent
violation
because PSEG did not detect and act upon an adverse trend of water intrusion
into the HPCI oil system. Specifically,
PSEG did not adequately
implement
procedures
to perform monthly HPCI turbine oil analysis
for water contamination
with known steam leakage by the Steam Admission
Valve (FD-F001
). The NRC screened
the finding for safety significance
and determined
that a detailed
risk evaluation
(DRE) was required.
The DRE was performed
by a Region I senior reactor analyst (SRA) and concluded
that the condition
resulted
in an increase
in core damage frequency
(CDF) of low E-6/yr, or of low-to-moderate
safety significance
(White).
This result was obtained
using the NRC's Standardized
Plant Analysis
Risk (SPAR) model for Hope Creek. Hope Creek Response
PSEG agrees that the performance
deficiency
occurred.
Hope Creek did not adequately
implement
procedures
to perform monthly HPCI turbine oil analysis,
did not identify
significant
moisture
contamination
in the HPCI oil system, and thus did not take the necessary
response
actions.
As a result, the HPCI system was not able to perform its design function
for a period greater than the fourteen
days allowed by plant Technical
Specifications.
PSEG has performed
a Root Cause Evaluation
that identified
weaknesses
in the Adverse Condition
Monitoring
(ACM) process,
as well as in oversight
of the ACM process and in individual
performance
and accountability
to the process.
Corrective
actions to improve the ACM process and management
oversight
of the ACM process are being implemented.
PSEG appreciates
the opportunity
to present our perspective
on the facts and assumptions
used by the NRC to arrive at the significance
level of the finding.
PSEG does not agree with the characterization
of the finding as low-to-moderate
safety significance
(White) and concludes
the characterization
of the finding should instead be one of very low risk significance
(Green).
This conclusion
is based on a review of the SPAR model which identified
many conservatisms
and some inaccuracies
in the modeling
of plant equipment.
Performance
Deficiency
Characterization
PSEG performed
a risk evaluation
similar to the risk evaluation
performed
by the NRC senior reactor analyst.
A review of the risk evaluation
performed
using Hope Creek Probabilistic
Risk Assessment
(PRA) models found significant
conservatisms
in the modeling
approach.
An extensive
PSEG review determined
that failure to credit equipment
available
to safely shutdown
the plant, including
secondary
plant equipment,
FLEX equipment
and other defense-in-depth
equipment,
caused the unnecessarily
conservative
results.
After an extensive
analysis
to incorporate
this equipment
into the internal
events and fire PRAs, PSEG concludes
that the risk increase
associated
with the HPCI failure is much lower than that originally
calculated
by PSEG and much lower than described
in the referenced
NRC inspection
report. Following
review of PSEG models, a review of the NRC model was conducted
and found similar conservatisms
and some inaccuracies
in the modeling
of plant equipment.
PSEG is providing
those results to the NRC to better inform the risk evaluation
of the HPCI system failure,
and to enhance the accuracy
of the NRC PRA model. 1 of 12
LR-N 16-0232 Enclosure
Following
correction
of the unnecessary
conservatisms
in the PRA models, the increase
in CDF from both the internal
and external
events is 7.6E-7/yr,
or of very low safety significance
(Green).
PSEG is requesting
that the NRC use the PSEG risk assessment
methodology
and results when assessing
the significance
of the event. A more detailed
discussion
of the Hope Creek PRA models and comparison
with the NRC SPAR model is attached
in section 2.0, Review of PRA Analysis.
In addition,
PSEG reviewed
the HPCI system data from June 23, 2016, that was described
in Sensitivity
Case 4 of the inspection
report. The inspection
report described
a concern that water intrusion
could have affected
system operation
as early as June 23, 2016, despite the successful
system test that was performed
on that date. The result of that review is being provided
for NRC consideration
and is contained
in section 3.0, Review of Sensitivity
Case 4, which concludes
that the HPCI control system was able to perform its design functions
during the June 23, 2016 test. As a result, PSEG believes
the exposure
time is most accurately
identified
as being 44 days. 2.0 Review of PRA Analysis
2.1 Purpose The purpose of this section is to summarize
PSEG's position
on the risk increase
associated
with the unavailability
of the HPCI system in July and August 2016. In its inspection
report (05000354/2016003),
the NRC discussed
a finding that was preliminarily
determined
to be White under guidance
associated
with the Significance
Determination
Process (SOP). PSEG's initial risk calculations
were generally
consistent
with this determination.
However,
further examination
of the Hope Creek PRA models revealed
significant
conservatisms
in the modeling
approach;
further review of the NRC models revealed
similar conservatisms
and additionally
some errors. 2.2 Key Assumptions
and Boundary
Conditions
The following
assumptions
are applied for the HPCI degraded
- The SOP risk evaluation
was performed
based on the following:
o As part of the determination
process,
an application-specific
internal
events risk model (ASM), HC116A-ASM
was created based on the most recent internal
events PRA Model of Record, HC111A. Development
of this ASM included
several revisions
to better reflect the as-built, operated
plant. This is referred
to as the Full Power Internal
Events (FPIE) model through the remainder
of this document.
o As part of the determination
process,
an application-specific
fire risk model (ASM), HC114FO-ASM
was created based on the most recent fire PRA Model of Record, HC114FO.
Development
of this ASM included
several revisions
to better reflect the as-built,
as-operated
plant. This is referred
to as the Fire PRA (FPRA) model throughout
the remainder
of this document.
o Seismic and other external
events hazard contributors
were reviewed
in the Hope Creek Individual
Plant Examination
for External
Events (IPEEE).
2 of 12
LR-N 16-0232 Enclosure
- Upon discovery
that the HPCI system was inoperable
on August 6, maintenance
on the reactor core isolation
cooling (RCIC) system was prohibited
by Hope Creek guidance.
The RCIC system was expeditiously
protected
by the control room operators
and remained
protected
during the last 5 days of the 44 day unavailability.
A more precise PRA calculation
would eliminate
the RCIC test and maintenance
term and lower any risk increase
calculations
by 2-3%. For the purposes
of this analysis,
no credit is taken for the operator
actions to protect the RCIC equipment.
All calculations
are shown for a 44 day interval.
- Repair and/or recovery
of the HPCI system are not credited.
Replacement
of the HPCI hydro-electric
governor
(EGR) is a simple task, but this is not credited
because of the uncertainty
associated
with the time necessary
to troubleshoot
the failure.
- Risk values in this document
are generally
presented
showing 3 significant
figures,
which allows a reviewer
to track exactly where in the ASM documents
the risk value comes from. The reviewer
should be aware that risks and changes in risk of the magnitudes
generally
discussed
are accurate
to one significant
figure. 2.3 PRA Modeling
PSEG made preliminary
modeling
results available
in time for NRC to incorporate
this information
into Inspection
Report 05000354/2016003.
Since that time, PSEG has undertaken
a major effort to update our Fire and FPIE PRAs. This section describes
the PRA model changes and then shows the best estimate
calculations
of risk increase.
The risk increases
are significantly
lower than those discussed
in the Inspection
Report. Initial review of the Hope Creek FPIE and FPRA models identified
conservatisms
compared
to the as-built,
as-operated
plant. The area that yielded the biggest risk reduction
was properly
crediting
shutdown
using the secondary
plant. The use of the main feedwater,
condensate
and turbine bypass systems was partially
credited
in the FPIE model and not credited
in the Fire PRA (i.e., the secondary
plant equipment
was considered
to be failed in all fire scenarios).
As part of this effort the control and power cables for the secondary
plant equipment
were modeled and found to be routed through different
fire areas than the RCIC control and power cables. The difference
in the cable routing contributed
to a significant
reduction
in the fire risk calculation.
This robust design is now reflected
in risk models. Other improvements
included
crediting
newly installed
FLEX equipment
and incorporation
of B.5.b. equipment
that was only partially
modeled in the last PRA updates.
Another modeling
area that contained
unnecessary
conservatism
was in the way RCIC failure to run, both from random failures
and support system failures,
was modeled.
Hope Creek models were revised to include:
- Credit for injection
from enhanced
control rod drive (CRD) system after 4 hours4.62963e-5 days <br />0.00111 hours <br />6.613757e-6 weeks <br />1.522e-6 months <br /> of RCIC operation.
- Credit for battery charging
from FLEX and B.5.b. diesels allowing
long term operation
of RCIC. These were credited
for station blackout
(SBO) scenarios,
as well as for SBO scenarios
involving
random failures.
A few numeric changes to basic event probabilities
were made, but the risk reduction
was not as significant
as changes made to properly
credit equipment.
The most important
basic events, which are the operator
failure to depressurize
using automatic
depressurization
system (ADS) and random failures
of RCIC, were reviewed
and not changed.
3 of 12
LR-N 16-0232 Enclosure
Model of Record (MOR) values for CDF and Large Early Release Fraction
(LERF) are compared
with their respective
ASM base case CDF and LERF values below in Table 1. These changes in risk metrics between the MOR and the ASM are the result of careful evaluation
of each model's conservatisms
and details.
Table 1 HOPE CREEK RISK MODEL COMPARISON
Category
Model Name CDF LERF HC111A 4.20E-6/yr
8.44E-7/yr
FPIE HC116A-ASM
3.31 E-6/yr 7.47E-7/yr
[DELTA] 8.90E-7/yr
9.70E-8/yr
HC114FO 2.18E-5/yr
3.08E-6/yr
FPRA HC114FO-ASM
6.80E-6/yr
1.78E-6/yr
[DELTA] 1.5E-5/yr
1.30E-6/yr
The baseline
CDF changes are significant,
especially
in the case of the FPRA. The FPRA is a relatively
immature
model. Prior to the analysis
associated
with this SOP, the FPRA had not been seriously
challenged
to identify
and remove conservatisms
such as those identified
below. Additionally,
the model benefitted
from recent NRC FAQs that were generally
created and resolved
by plants working on NFPA 805. The FPIE model also contained
conservatisms.
Most were discovered
by working with the Operations
Department
to ensure that the available
equipment
was properly
credited.
A FPIE model update is scheduled
for 2017. The model update had been delayed awaiting
complete
installation
of FLEX equipment
and publication
of NEI guidance
for incorporating
FLEX into a PRA model. Changes to both PRA models are addressed
under established
processes
governed
by risk management
procedures.
Update Requirement
Evaluations
(UREs) have been created for both the FPIE and Fire PRA model adjustments
to ensure those changes are incorporated
in the next periodic
updates.
The following
describes
the major changes to the models and the results of the PSEG analysis.
The analysis
packages
are available
for NRC review. All changes have been made in accordance
with the PRA Standard
(Reference
1 ), PSEG Risk management
procedures
and industry
best PRA practice.
They are permanent
changes to the Hope Creek models. FPIE LlCDF and LlLERF Calculations
The HC116A-ASM
model features
the following
changes from the MOR: * Fault Tree Changes:
o RCIC success criteria
with CRD available
o Crediting
of some FLEX procedures
and equipment
o SACS heat exchanger
valves o MCC 108421 cross-tie
o Diesel generator
undervoltage
circuitry
o Additional
basic events 4 of 12
LR-N16-0232
Enclosure
- Data Changes:
o HPCI/RCIC
room steam leak event o Dependent
failure to operate high pressure
systems o Suction strainer
basic event calculation
method o SRV accumulator
leakage event Base FPIE HC116A-ASM
CDF = 3.31E-61yr
FPIE CDF with HPCI OOS(1l = 8.63E-61yr
FPIE f1CDF = [(8.63E-61yr)-
(3.31 E-61yr)]
- exposure
time = 5.32E-61yr
- [44 days I (365 dayslyr)]
= 6.42E-7 Base FPIE HC116A-ASM
LERF FPIE LERF with HPCI OOS(1l = 7.47E-71yr
= 1.15E-61yr
FPIE nLERF = [(1.15E-61yr)-
(7.47E-71yr)]
- exposure
time = 4.03E-71yr
- [44 days I (365 dayslyr)]
= 4.86E-8 Fire PRA f1CDF and f1LERF Calculations
The HC114FO-ASM
model features
the following
changes from the MOR: * Additional
model detail for hot short spurious
actuation
- Radwaste
area hoist scenario
refined * Restoration
of circulating
water pump house scenarios
- Fault tree, data adjustment,
and basic event additions
similar to the FPIE changes listed above. * Incorporation
of additional
cable data for the following
systems:
o Condensate
o Circulating
Water o Feedwater
o Instrument
Air o Instrument
Gas o 120 VAC Power Panels o Primary Containment
o Reactor Auxiliaries
Cooling * Revised probabilities
& calculations:
o Human error probabilities
o Non-suppression
probabilities
- Targets revised in the following
fire areas: o CD28 o CD29 o CD30 o CD31 (1) Set Basic Event HPI-TDP-FS-OP204
(HPCI FTS term) to TRUE via flag file 5 of 12
LR-N 16-0232 Enclosure
Base Fire PRA HC114FO-ASM
CDF = 6.80E-61yr
Base Fire PRA CDF with HPCI OQS(1l = 7.76E-61yr
Fire 6.CDF = [(7.76E-61yr)-
(6.80E-61yr)]
- exposure
time = 9.6E-71yr
- [44 days I (365 dayslyr)]
= 1.16E-7 Base Fire PRA HC114FO-ASM
= 1.91 E-61yr Base Fire PRA LERF with HPCI OQS(1l Fire 6.LERF Results = [(1.91 E-61yr)-(1. 78E-61yr)]
- exposure
time =1.38E-71yr
- [44 days I (365 dayslyr)]
= 1.66E-8 The totai6.CDF
is 6.42E-7 (FPIE) + 1.16E-7 (FPRA) = 7.57E-7.
The total 6.LERF is 4.86E-8 (FPIE) + 1.66E-8 (FPRA) = 6.52E-8.
2.4 Comments
on the SPAR model analysis
NRC used the Hope Creek SPAR model to evaluate
the internal
events risk and clearly described
their risk analysis
in the referenced
Inspection
Report. Using Sensitivity
Case 5, the NRC developed
a refined best estimate
delta CDFiyr of 2E-6, which is based on the sum of the internal
events risk analysis,
calculated
from the SPAR model of 9.92E-7 and the provided
fire risk increase
of 1.1 E-61yr. This section discusses
the conservatisms
in the SPAR model. The fire risk increase
is based on a preliminary
analysis
that was made available
to the NRC, as described
in Section 2.3. For sequences
in which RCIC failed to run, the NRC adjusted
the probability
of operator
failure to depressurize
the reactor from 5E-4 to 1 E-4. The adjustment
was intended
to account for the operator
action and the inherent
conservatism
in using a 24 hour2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> run time for RCIC. Given the simplified
structure
of the SPAR model and the simplified
nature of the SPAR-Human Reliability
Analysis
Method (SPAR-H)
being used to quantify
human error probabilities,
this approach
is reasonable.
Since the numeric change is a rough estimate
based on SRA judgement,
there is no conclusive
way to quantify
the validity
of this adjustment.
However,
as discussed
below, this numeric change does fully approximate
the difference
in RCIC failure rates and the failure to depressurize
the reactor,
which often appear in the same cutsets.
The NRC calculated
a change in CDF of 1.86E-61yr
using this modeling
approach.
PSEG reviewed
the calculations
done by the NRC, and reproduced
the calculations
based on the NRC descriptions
of the analysis.
The NRC ran five sensitivity
cases: (1) Set Basic Event HPI-TDP-FS-OP204
(HPCI FTS term) to TRUE via flag file 6 of 12
LR-N16-0232
Enclosure
Sensitivity
1 (1.64E-6/yr):
SSW 'B' Train Unavailable
Due to Test & Maintenance
The SRA removed event SSW-SYS-TM-LOOPB
from cutsets as a sensitivity
case. This change alone reduced the change in CDF from 1.86E-6/yr
to 1.64E-6/yr,
or about 12%. This change should be part of the base case because it stems from an error in the SPAR model. This maintenance
event is modeled in SPAR to immediately
and completely
remove the possibility
of depressurizing
using the ADS valves following
a loss of offsite power (LOOP). The model is incorrect,
because the ADS valves would be functional
until battery depletion,
which would be over 4 hours4.62963e-5 days <br />0.00111 hours <br />6.613757e-6 weeks <br />1.522e-6 months <br /> in a LOOP and over 6 hours6.944444e-5 days <br />0.00167 hours <br />9.920635e-6 weeks <br />2.283e-6 months <br /> if an extended
loss of AC power (ELAP) is declared.
Opportunities
to charge the batteries
with 10 CFR 50.54(hh)(2)
equipment
(usually
called B.5.b equipment)
or FLEX equipment,
as well as the probability
of recovering
from the LOOP, are not credited
in the SPAR model. Sensitivity
2 (2.35E-6/yr):
Basic SPAR run This sensitivity
analysis
removes the improvements
made in the base case described
above and provides
no additional
information.
Sensitivity
3 (1.64E-7/yr):
Additional
Changes to Depressurization
Probability
This is a further adjustment
to the depressurization
probability
for the base case (depressurization
probability
= 1 E-4) to this case (depressurization
probability
= 7 .5E-5). The risk reduction
of a 25% reduction
in depressurization
probability
leads to a -12% reduction
in CDF increase.
PSEG understands
that the risk increase
is very sensitive
to the rare event probability
that the operating
crew fails to depressurize
the reactor when required.
PSEG reviewed
and did not change the depressurization
probability
in the Hope Creek PRA model. Sensitivity
4 (2.03E-6/yr):
Full Exposure
Time The NRC performed
this sensitivity
calculation
assuming
an increased
exposure
time, including
the failure to depressurize,
as well as including
failure to depressurize
human error probability
(HEP) changes but not including
SSW B train adjustments.
PSEG concludes
that the HPCI system was operable
on June 23, 2016, as discussed
in section 3.0, Review of Sensitivity
Case 4. Therefore,
this sensitivity
analysis
is not appropriate
for significance
determination.
Sensitivity
5 (9.92E-7/yr):
Changes to delete core damage sequences
in question
and adjust operator
depressurization
failure probability
for fast acting initiating
event (Medium Break LOCAs (MLOCA))
This sensitivity
case comes closest to structurally
matching
the PSEG analysis,
so it provides
the best case for discussing
the similarities
and differences
between the SPAR model and the PSEG PRA model. The NRC deleted cutsets that contain LOOP events with SSW train B in test or maintenance.
This should have been done for the base case and all sensitivity
cases because that event is modeled incorrectly
in the SPAR model, as discussed
under sensitivity
case 1. This unlikely
. maintenance
activity
is correctly
modeled in the PSEG PRA. The NRC revised the HEP for operator
failure to depressurize
event as was done in the base case, but not the rest of the sensitivity
cases. The PSEG model uses 3. 75E-4 as the probability
of failing to depressurize
using ADS following
or a LOOP. PSEG did not adjust the HEP for failure to depressurize
for the RCIC failure to run scenarios
but did model other relevant
success paths, such as crediting
enhanced
CRD for decay heat removal and inventory
control after 4 hours4.62963e-5 days <br />0.00111 hours <br />6.613757e-6 weeks <br />1.522e-6 months <br /> of RCIC success.
7 of 12
LR-N16-0232
Enclosure
The NRC increased
the operator
failure to depressurize
event probability
to 2. 75E-3 for MLOCA scenarios,
which resulted
in a slight risk increase.
This approach
was already used in the PSEG model, so no changes were necessary.
The NRC set RCI-MOV-FC-FRO
(RCIC injection
valve fails to reopen) to FALSE because this valve (BD-F013)
would remain open with no automatic
closure signal during RCIC operation.
This valve is correctly
modeled in the PSEG PRA, so no changes are needed. Since this change corrects
a SPAR model error, all sensitivity
cases should include this adjustment.
Reasons for the Differences
between the PSEG PRA and the SPAR model NRC sensitivity
case 5 will be used to discuss differences
and similarities
between the SPAR analysis
and the PSEG analysis.
Case 5 was chosen because it includes
corrections
for errors identified
in the SPAR model, making it a better choice for the basic comparisons.
Case 5 lists the dominant
sequences
as: * Loss of condenser
heat sink, with failure to depressurize
and RCIC in Test and Maintenance
- Loss of Main Feedwater,
with failure to depressurize
and RCIC in Test and Maintenance
- Loss of condenser
heat sink, with failure to depressurize
and RCIC failure to run * Loss of condenser
heat sink, with failure to depressurize
and RCIC failure to start * Loss of Main Feedwater,
with failure to depressurize
and RCIC failure to start These scenarios
are essentially
identical
to those in the PSEG analysis;
the differences
are in the quantification.
The NRC calculates
a b.CDF of 9.9E-7/yr
and PSEG calculates
6.4E-7/yr,
resulting
in a 35% difference.
The major difference
is caused by the difference
in the probability
of operators
failing to depressurize
using ADS. The basic NRC Human Error Probability
(HEP) is 5E-4 while the PSEG HEP is 3. 7E-4, a difference
of 26%. This HEP (or a similar event) is in almost every cutset, so the difference
in b.CDF is almost proportional
to the difference
in HEP. When analyzing
HEPs that are relatively
rare events (probability
< 1 E-2), Human Reliability
Analyses
(HRA) routinely
vary by much more than the 26%. The SPAR-H HRA methods,
used by the NRC, and the EPRI HRA calculator,
used by PSEG, were benchmarked
with many other methods in a broad international
study completed
over the last decade. Numerous
examples
of the variation
between these and other methods can be found in "International
HRA Empirical
Study-Phase
1 Report: Description
of Overall Approach
and Pilot Phase Results from Comparing
HRA Methods to Simulator
Data" (NUREG/IA-0216,
Vol. 1.) and several subsequent,
related documents.
was reviewed
and no changes were made for this SOP evaluation.
is unchanged
from the latest formal peer review of the Hope Creek PRA, and is available
for NRC review. After the HRA differences,
the major differences
come from RCIC system reliability
data. The PSEG test and maintenance
unavailability
for RCIC is 7. 71 E-3 compared
to the SPAR unavailability
for RCIC of 1.095E-2,
a 30% difference.
The PSEG value is based on data collected
from PSEG plant specific
maintenance
rule records during the last PRA update. Other differences
include the SPAR models' use of higher failure rates for RCIC and no credit for using CRD injection
after about 4 hours4.62963e-5 days <br />0.00111 hours <br />6.613757e-6 weeks <br />1.522e-6 months <br />. Additionally,
the SPAR models do not credit the possibility
of using B.5.b or FLEX equipment
to charge batteries
and operate RCIC when the normal chargers
are not available.
These details are not normally
credited
in the SPAR models. 8 of 12
LR-N 16-0232 Enclosure
Conclusion
on the SPAR analysis
NRC Sensitivity
Case 5 gives similar results to the PSEG analysis
because this case includes
corrections
to identified
errors and conservatisms
in the SPAR model. The difference
in the .b..CDF values is clearly understood
to be a result of different
HRA models for a rare event, some differences
in equipment
reliability
data and some simplifications
in the SPAR model. None of these differences
invalidates
the SPAR model as an independent,
confirmatory
tool. In fact, the SPAR results confirm that the latest Hope Creek PRA results properly
model the condition
because the dominant
.b..CDF cutsets and scenarios
are very similar.
Summary of predominant
analytic
differences
between plant and SPAR model: * ADS is available
for 4 to 6 hours6.944444e-5 days <br />0.00167 hours <br />9.920635e-6 weeks <br />2.283e-6 months <br /> following
a LOOP (Battery
life). The ADS function
is being modeled as unavailable
if B SSW Loop is in Test or Maintenance.
- ADS is available
after 4 hours4.62963e-5 days <br />0.00111 hours <br />6.613757e-6 weeks <br />1.522e-6 months <br /> following
a LOOP because B.5.b and/or FLEX equipment
can be used to maintain
batteries.
- RCIC system reliability
uses the actual plant reliability
values in the plant model * RCIC injection
valve (F013) failure to reopen should be removed from the base case, because this valve remains open following
RCIC initiation.
- No credit is taken for CRD injection
after 4 hours4.62963e-5 days <br />0.00111 hours <br />6.613757e-6 weeks <br />1.522e-6 months <br /> of successful
RCIC operation.
to restore RCIC batteries
and maintain
injection
capability
after 4 hours4.62963e-5 days <br />0.00111 hours <br />6.613757e-6 weeks <br />1.522e-6 months <br />. The HRA model and equipment
reliability
parameter
calculations
in the Hope Creek model were done in accordance
with the PRA Standard
and have been subjected
to a peer review with no relevant
Findings
& Observations.
Therefore,
the latest PSEG internal
events PRA model and fire PRA model should be used for input into the significance
determination.
2.5 Seismic and Other External
Hazards Hope Creek does not maintain
seismic,
external
flooding,
or other external
hazard PRAs. Seismic,
external
flooding,
high winds, and other external
hazards are discussed
in the IPEEE (Reference
2). A seismic risk study (PRA that falls short of current standards
but provides
clear, NRC reviewed
insights)
was performed
for the IPEEE. The top five core damage sequences,
labeled seismic damage states (SDS), are discussed
in the IPEEE. The seismic risk is dominated
by loss of instrumentation
distribution
panels. Two SDSs are relevant
given a HPCI failure:
- SDS 26 is a seismic-induced
LOOP followed
by a failure of high pressure
injection
and random failures.
The random failures
are dominated
by RPV depressurization
failures
and EDG failures
resulting
in an SBO. Given a HPCI failure,
this SDS would become more significant,
as there would be limited high pressure
injection
capability.
However,
station FLEX capability,
which is not considered,
should be able to effectively
mitigate
the SBO scenarios.
This SDS contributes
-5% to seismic CDF. 9 of 12
LR-N16-0232
Enclosure
resulting
in core damage. Random failures
are dominated
by EDG failures
resulting
in an SBO. Neither random failures
of high pressure
injection
nor failure to depressurize
were the dominating
failure in this SDS. Given a HPCI failure and SBO, RCIC is available
on batteries
for injection,
and additional
would provide electrical
backup for RCIC as well as low pressure
injection.
The IPEEE review concluded
that external
hazards are not a significant
risk contributor.
The analysis
provided
also did not include newer station capabilities
to mitigate
external
events with B.5.b and FLEX equipment.
Seismic,
high winds and external
flooding
risk would not be significantly
impacted
by HPCI being unavailable.
2.6 Conclusions
For the base case with a 44 day exposure
time, the total
is 7.57E-7.
(6.42E-7
(FPIE) + 1.16E-7 (Fire PRA)) and the total
is 6.52E-8.
(4.86E-8
(FPIE) + 1.66E-8 (Fire PRA)). Thus,
is <1 E-6 and
is < 1 E-7, representing
a finding of very low risk significance
(i.e., Green). Table 2 SUMMARY OF HOPE CREEK HPCI SOP RISK CALCULATIONS
(BASED ON 44 DAY EXPOSURE
TIME) Case FPIE PRA Fire PRA Total Metric
Results 6.42E-7 1.16E-7 7.57E-7 < 1E-6
Results 4.86E-8 1.66E-8 6.52E-8 < 1E-7 PSEG performed
three sensitivity
analyses
to evaluate
differences
between plant and SPAR models and to evaluate
the benefit from FLEX equipment.
The sensitivities
were performed
using the FPIE because the fire PRA is not the dominant
contributor
to the total risk increase.
The three analyses
were: * Increase
the depressurization
probability
to the SPAR model probability.
- Increase
the RCIC failure to run probability
from Hope Creek's calculated
probability
to the SPAR model probability.
- Remove credit for FLEX equipment.
None of these sensitivity
analyses
increased
the delta risk to the thresholds
for a White finding.
3.0 Review of Sensitivity
Case 4 In Sensitivity
Case 4 of the Inspection
Report, the NRC discussed
a concern that the data from the June 23, 2016, HPCI test show the control valve opened to around the 80 percent position
on initial pressurization,
which was further than observed
on previous
tests, and that it achieved
a position
of about 95 percent under the ramp generator
control.
This is greater than previous
tests in which a control valve position
of 40-55 percent was observed.
The SRA expressed
concern that this response
created uncertainty
in the length of the exposure
time and therefore
uncertainty
in the increase
in risk. However,
as shown below the HPCI pump was able to 10 of 12
LR-N 16-0232 Enclosure
perform its design functions
during this test so there should be no change to the assumed exposure
time of 44 days. The HPCI start sequence
is described
in the EPRI NMAC Terry Turbine User's Manual, as follows:
Once the auxiliary
oil pump is started,
the turbine oil relay hydraulic
system will pressurize
first. The turbine governor
(control)
valve will start to open. Next the governor's
hydraulic
system will pressurize
and the turbine governor
valve will start closing again. Then the hydraulic
oil pressure
will develop at the turbine stop valve's hydraulic
cylinder
and the stop valve leaves its closed position.
The magnitude
of the initial governor
valve opening and the overall time period is dependent
upon the drain down condition
of the turbine's
oil system. Once the stop valve leaves its closed position,
the ramp generator
signal and signal converter
(RGSC) ramp circuit will be initiated
and the voltage output will be increased
in a positive
direction.
During the HPCI System Start-up
on June 23, 2016: * Aux oil pump started * The indicated
position
of the governor
valve showed that the valve was open greater than expected
- The Pilot valve drove the governor
valve towards the closed position
in response
to the remote servo and EGR as expected
and lAW with EPRI NMAC Terry Turbine User's Manual * At this time, flow indication
and therefore
turbine speed was still at zero prior to the governor
valve moving towards the open position.
(reference
figure 1) * Governor
valve then began to open in response
to the demand of the RGSC as part of the normal start-up
sequence
The June 23, 2016, start-up
sequence
is consistent
with the operation
description
from the EPRI Manual. During the fall 2016 refueling
outage, a visual and dimensional
inspection
of the HPCI pilot valve under was completed.
The pilot valve was found to be in overall good condition,
with light wear, and was reused. The pilot valve's top, middle, and bottom control lands were inspected.
The control land corners have light wear but are still sharp and free from burrs and nicks. Outside diameter
measurements
of the control lands were taken with a micrometer
and met EPRI manual requirements.
The lower control land had minor wear. The middle control land had approximately
20 minor score marks, which were lightly stoned to be removed.
The top control land had very minimal wear. The bore of the pilot bushing was observed
in good condition
with minimal oil residue and no corrosion
build up identified.
No erosion or pitting was identified.
The inside corners of the control ports were sharp and free from burrs and nicks. A swab was used to clean out the bushing bore. The inspection
pictures
show score marks on the pilot relay which are consistent
with the anomalies
observed
in the governor
valve stoke trace data from June 23, 2016. 11 of 12
LR-N16-0232
Enclosure
As discussed
above, the June 23, 2016 test results are consistent
with the expected
system response.
HPCI pump 1ST G-23-16 * -8.13821
o8tU38671-----,-------,-------,
1"1""00-,1""oo,-----,to=oo,-1'"'2o"'o
.-------1
- HC.STR212
<> 2.!U1275
80.74072
%
I
5650.3*1:250
GPM
<>J-IC.STR211
605.18588
> 1\. -----r----.........-.J
I
I { 61231201610:06:00
AM 5/23!201610.07:03.66181
612312016
10:08:30
AM * HPCJ STEAM ADM. VALVE POS. (FV-4880)
o HPCJ CONTROL VA!.VE POS. (FV-4879)
+ HPCI PUMP DISCHARGE
HOW <> HPCJ PUMP DISCHARGE
PRESSURE
Figure 1: HPCI Test Data from June 23, 2016 3.1 Conclusions
A review of the EPRI summary of system operation
shows that turbine governor
valve will start to open, close, and then open again and the magnitude
and time of this opening is dependent
on system conditions.
On June 23, 2016 the governor
valve did open more than expected;
however the data trace from June 23, shows it reopening
in response
to the RGSC control signal prior to turbine/pump
rotation.
The control system demonstrated
that it was able to take control and respond normally.
The plot of the HPCI starting
sequence
above shows this governor
valve movement.
The oil sample taken on that day had water content higher than the EPRI recommended
limit, however from all of the parameters
monitored
it is concluded
that the HPCI control system was able to perform its design functions
during the June 23, 2016 test. 4.0 References
1. ASME/ANS
RA-Sa-2009,
"Standard
for Level 1/Large Early Release Frequency
Probabilistic
Risk Assessment
for Nuclear Power Plant Applications,"
Addendum
A to S-2008, ASME, New York, NY, American
Nuclear Society,
La Grange Park, Illinois,
February
2009. 2. Hope Creek Generating
Station,
Individual
Plant Examination
for External
Events, Submittal
Report, July, 1997. 12 of 12