ML20086G025
| ML20086G025 | |
| Person / Time | |
|---|---|
| Site: | Waterford  |
| Issue date: | 11/26/1991 |
| From: | Burski R ENTERGY OPERATIONS, INC. |
| To: | NRC OFFICE FOR ANALYSIS & EVALUATION OF OPERATIONAL DATA (AEOD) |
| References | |
| W3F191-0661, W3F191-661, NUDOCS 9112040136 | |
| Download: ML20086G025 (9) | |
Text
-
t AR%
E ENTERGY i
n,r,nur,.u W3F191-0661 A4.07 z
QA November 26, 1991 Mr. Thomas H. Ilovak, Director Division of Safoty Programs Offico for the Analysis and Evaluation of Operational Data U.S. Iluclear Regulatory Commission Washington, D.C 20555 Sunject:
Watorford 3 SES Docket lio. 50-382 Licenso tio.11PF-38 Comments on the Draft AEOD Special Report " Experience with Pump Seals Installed in Reactor Coolant Pumps Manufactured by Dyron Jackson" Dear Mr.11ovaki In a lottor dated September 27, 1991, you requested Waterford 3 to review a draft AEOD special report, "Experienco with pump Seals Installed in Reactor Coolant Pumps Manufactured by Byron Jac); son" and provido commento.
Attachmont 1 providos Waterford 3 comments on t!.e report.
As stated in tho report, the findings and conclusions documented in the report woro based on several factors including discussion with key staff and industry exports involved with the resolution of Generic Issuo (G1) 23, "Roactor Coolant Pump Seal Failures."
The 11RC prepared a draft Regulattry Analysis for GI 23, draft 11UREG-1401, and an approach for the resolution of G123, Draf t Regulatory Guido DG-1008.
On April 19, 1991, a Federal Registor Notico was published by the llRC announcir.g the availability of those documents. The Federal Registor 110tico requested cotaments on those documents as well as responso to a series of questions addressing rGactor coolant pump seal performanco, proceduros, analysos, and other factorn.
By letter dated September 27,
- 1991, Entergy Operations submitted the requested comments and responses to the specific questions for Waterford 3 and A110.
Watorford 3 believes this document will provido you with significant additional information portinent to the special report.
A copy of the September 27, 1991, submittal is provided in Attachment 2.
pI
- p' % C W o m, 3 a wo om m, f
<i a
Pbh' f
J
t s
Mr. Thomas M. Novak-Comments on the Draft AEOD Specia? Report
" Experience with Pump Seals Installed in Reactor Coolant Pumps Manufactured by Byron Jackson" W3F191-0661 Page 2 November 26, 1991 Waterford 3 appreciatos the oppo ni, to review and comment on Jr Robert J. Murillo if you the special report.
Ploaso contact have any questions regarding these comments.
Very truly yours, jf;/u l~ lUwwA R.
hurski 91 rector, Nuclear Sofoty j
RFD/IL7M/dc Attachment cct R.D. Martin, NRC Region IV D.L. Wigginton, NRC-NRR R.B. McGohoo N.S. Reynolds NRC Resident Inspectors Offico a
f
e s
l s
l 1
9 I
ATTACllME14T 1 WATERFORD 3 COMMENTS ON DRAFT AEOD SPECIAL REPORT f
" EXPERIENCE WITH PUMP SEALS INSTALLED IN REACTOR COOLANT PUMPS MANUFACTURED BY DYRON JACKSON" i
t 3
2 b.n+-
.'ew-w-.
y..
-,-,+,,w h,--
4.-m.--
.c y,,_.
_9n,,,..,,-
m-
.1,w,,
=
4 Waterford 3 comments on the draft AEOD special report are provided i
below.
The major headings correspond to the heading used in the i
report.
Listed under the heading is the pago number and paragraph whero the comment is mado, i
i EXECUTIVE
SUMMARY
Page vili, first paragraph Use of NPRDS data to evaluate failuros is not appropriato for this report.
NPRDS data includes incipient failuro data and not necessarily actual failures.
Additionally, the reporting of this data is subjectivo and not nocessarily consistent from ono utility to another.
As an example, several of the entries in Tablo 5 doncribe seal degradation.
Those occurrences did not constituto failed seals, yet they are included in Tablo 5 which is titled, i
" Continued Operation with Failed Seal Stages".
In those occurrences, the seal conditions wore identified to ensure increased ongineering observation of the seals.
(
pago viii, third paragraph The statomont that the Byron Jackson (BJ) pumps, particularly the original SU design, have a history of unrollablo operatinn is not correct for Waterford 3.
- Again, the use of NPRDS data is subjectivo and not necessarily quantitativo and could lead to misleading and inappropriato conclusions.
Beforo any meaningful dialoguo can be hold related to RCP seal "failuros",
a basic understanding of what constitutos a soal failuro must be established.
For discussions such as those in the draft report, it is considorod appropriate -to cont.idor throo
=possible reasons for unscheduled RCP seal replacement:
P 1) seal-performanco degradation requiring seal replacement (no significant increase in seal leak-off rates),
2) seal failuro resulting in seal leakago ratos within the normal charging - capacity of the plant (i.e. not challenging plant safety systems), and or catastrophic soal failure that results in seal 3) gross ~
rates beyond ~ the plant's charging capacity (i.e.
leakago IACA).
As -stated in the April Federal Register Notico for GI 23, "In recent years the rate of seal failures appears to have decreased by.
roughly 50 percent...Porhaps most importantly, during the past soveral years there have been no seal failures with a high enough 1
c leak rate to be clascified as a LOCA."
The propouod GI 23 resolution backfit analysis used pre-1986 seal failuro data and, even as the Fodoral Register statomont doos, lumps the throo possible reasons for unscheduled RCP seal replacement described above togother.
In its offort to examino more recent RCP soal experience, the subject draf t AEOD report maintains this grouping, stating "... failure of a multi-stage seal is defined as loss or damage of one or more stagno."
This is considered inappropriato since it results in the Staff's goal to reduce the probability of catastrophic RCp seal failure (and corresponding risk) being equated to reducing the probability that an RCp soal dogradation will not occur outsido of a regularly scheduled replacement cycle.
The lattor goal is an oconomic one which Watorford 3 has boon aggressively pursuing.
Page ix, first paragraph Tho brief probabilistic discussion is inconclusivo and does not appear to bo portinent to the subject matter of this report (i.e.
recent BJ RCP seal experience).
Watorford 3 recommends that only a brief reference to documents containing more detail on the core melt risk contribution of RCP seal gross failuro be mado.
Page ix, third paragraph The statomont "However, the loss of CCW, even for very short
- periods, is sufficient to causo seal damage."
has not boon qualified suf ficiently by this report.
While there have boon instances where seals have 1 caked or mis-staged, presumably as a result of a loss of CCW, the - number of seals which havo not exhibited those characteristics when exposed to the same conditions is far greator.
This observation indicates there are other influon:os beyond the loss of CCW in seal failure mechanisms.
Industry testing and experience has demonstrated soveral instances of loss of cooling ovents where seal operation was not affected or would not have affected future coal operation. This information is identified in the Entorgy Operations September 27, 1991, submittal (Attachment 2)-on the proposed resolution of GI 23.
The statomont " Changes in service water flow, or rapid cmall CCW l
temperature changes, are sufficient to cause degradation of the i
seals." noods qualification.
Subsequent inspections of the seals have shown that little if any damago to the seal f aces result from these oscillations, provided that the magnitudo of the oscillations is not too sovero.
Those transients do not always chuse degradation, and the degradation (i.e.,
pressure oscillations) usually mitigates itself in relatively short order.
Pago x, third paragraph 2
\\
The statomont "... plants with these pumps frequently operate with one or more failed seal stages."
is grossly incorrect for Waterford 3.
Waterford 3 has soldom operated with one or more failed seal stage.
1.1 Background
Page 1, first paragraph The conclusion in the statomont, "off normal conditions which can load to seal failuro includo loss of offsite power (station blackout),
loss of component cooling wator,...",
has not boon adequately demonstrated. Industry experience has shown that the IU seals are resilient to loss of cooling events.
Additional comments on this subject are provided in Attachment 2.
Page 1, fourth paragraph s
The statomont that " Operating data for Ik7 RCPS show that plants with those pumps frequently operato with one or more failed seal stagos is encorrect for Waterford 3.
Waterford 3 has soldom operated with one or more failed stages.
Pago 2, third paragrata Regarding the statement, "There are design dif ferences between the seals used in the different pumps that tend to support a loss conservativo approach for Byron Jackson pumps because of greator redundancy in the seal," the term "less conservative" in quito misleading in this instance.
The Westinghouse seal has nearly the entire pressure drop across the first stago assembly.
The DJ soal has equal breakdowns across the seal stages, and the significance of a stage opening has less significance than had the stage opened in the Westinghouse scal.
Comparisons of different seal designs for relative conservatism is not technically meaningful.
2.0 RCP Seal Operating Experience for DJ Pumps Page 5, first paragraph Refor to the previous comments regarding the use of NPRDS data.
Page 7, firpt paragraph Waterford 3 had one forced outage in its history as a direct result 3
I i
of inadequato RCP soal performance.
There have been other instances of seal replacomonts that woro performed as an electivo l
maintenanco measure during forced outageal however, the forced j
outago was not result of a soal failure.
Attributing the outage to i
the soal is not appropriato.
Waterford 3 has not had a forced outago due to RCP seals sinco October 1985.
Additional comments i
related to this are provided in Attachmont 2.
Page 10, first paragraph Regarding the statomont,
" Plants with SU RCP soals frequently experienced soal stage failures during power operations, and son 1 staging problems on start-ups after outages," the Waterford 3 experienco has boon that on occasion (i.e.,
110T frequently) this occuro.
[
Page 10, second paragraph Watorford 3 currently operatos with 3 tU 11-9000 seals and a AECL i
Call 4 seal.
2.2.1 Plant-Specific Experience with SU Typo Seals Page 12, first paragraph Waterford 3 currently operatos with 3 ID 11-9000 seals and a AECL CAli 4 seal.
Page 16, sixth paragraph Regarding the statement, "In a 30-minute loss of component cooling water test highlighted in - liUREG/CR-4 54 4 (Ref. 5), all four seal stages excooded 250 degroos F at 10 minutos into the test," the statement is inconclusivo in that it does not acknowledge the results of this test.
Also Section 3.1, which gives a more detailed discussion of the temperaturo sensitivity of SU seals, does not acknowledge this test.
Page 17, first paragraph Waterford 3 has not experienced soveral additional failures after 1989.
Waterford 3 had a problem with upper shaf t 0-ring on ono RCP which resulted in extremely minor external leakage.
The seal stages functioned properly and were unaffected.
The occurrence should not be considered a failure.
4 t
Waterford 3 currently operatos with 3 BJ N-9000 seals and a AECL CAN 4 seal.
1 2.4.1 Plant-Specific N-9000 Seal Experience Pago 21, third paragraph Watorford 3 currently operatos with 3 BJ H-9000 seals and a AECL CAN 4 seal.
3.0 Summary of Operating Experienco Page 23, first paragraph Regarding the statomont, " Operating data show that SU type soais have frequently experienced single seal stage failures, and that i
they are temperature sensitivo.
As summarized in Section 3.1, the loss of CCW, even for very short periods, is sufficient to causo I
soal damage," the conclusion has not boon demonstrated.
Industry experience has shown that the BJ seals are rosilient to loss of cooling events.
Additional comments on this rubject are provided
)
in Attachment 2.
]
i
~ APPENDIX Page A-26, events 10/23/88 and 12/7/88 Event dated 10/23/88 is not soal related.
Event dated 12/7/88 is an inspection not a failure.
Page A-27, event 1/26/89 y
Evont dated 1/26/89 is not seal related.
i i
t I
I l
i 1
ATTACIIMENT 2 COPY OF ENTERGY OPERATIONS COMMENTS AND RESPONSES TO SPECIFIC QUESTIONS CONCERNING Tile PROPOSED RESOLUTION OF GI 23, DATED SEPTEMBER 27, 1991 i
5 t
r. - -:---,.s_.w',n,'
nn
,..,,.,c-, -,
+.,nr.-.
,=,n.:.,,-,n,,-,,,--ennn.--_-n- - - -,.
.,....n,,,
.n..,-
,,.n.
,- u -
es.
+
llllI M "" """ * " "
..=:= ENTERGY
=--
John R. McGaha i
i I
September 27, 1991 Chief, Regulatory Publication Branch Division of F4eedom of Information and Publication Services U. S. Nuclear Regulatory Commission Washington, D.C.-
20555 ATTENTION:
Docketing and Service Branch Subject Solicitation-of Public Comments on Generic Issue 23,
" Reactor Coolant Pump Seal Failure"; and Draft Regulatory Guide; Issuance, Availability, Federal Register Volume 56, Number 76 - April 19, 1991 CNRO-91/00028 Dear Mr. Chilk NRC has prepared a draf t Regulatory Analysis for Generic' Issue (GI)-23, draft NUREG-1401, and an upproach for the resolution of GI-23, Draft Regulatory Guide DG-1008.
On April 19, 1991, a Federal Register Notice was published by the NRC announcing the
- availability of-these documents.
The Federal Register Notice requested comments on these documents as well as responses to a series of questions addressing reactor coolant pump (RCP) seal performance, procedures, analyses, and other factors.
As' stated in NUREG/CR-5167, the primary objective for the resolution of GI-23 is to reduce the risk of core-melt accidents associated with RCP seal failure (and consequential small-break LOCA).
This decrease in risk was to be accomplished by reducing the probability of seal failure, thus making RCP seal failure a relatively small contributor to total core-melt frequency.
-The NRC's proposed-resolution is designed to address two separate areas:
(1) The potential for excessive leakage of
- reactor coolant across the primary boundary through the RCP i
seals as a result of seal failure during normal plant operating l-conditions; and (2) the potential for excessive leakage as a result.of loss of RCP seal cooling due to off-normal conditions si;n as station blackout or failure of the component covling-l water system.
The NRC stated that the proposed resolution of-
) 1%h*
G\\ ui Wl V ' ~
7q C i JGC09161.J19/JCLFLR-1 l
--.u_.,m-,
.--___,._.._,___._..a_._
- ~
l t
Mr. Samuel S. Chilk September 27, 1991 CNRO-91/00028 Page 2 of 4 GI-23 applies to pressurized water reactors (PWRs) only.
The purpose of this submittal is to provide the requested co.nments and responses to the specific questions for Entergy Operations' PWRs - Arkansas Nuclear One, Units 1 & 2 (ANO 1&2) and Waterford-3.
Entergy Operations has reviewed the referenced documents, and our comments are provided in Attachment 1.
These comments can be summarized as follows:
Improved RCP seal performance during normal operating e
conditions has accurred as a result of utility efforts to this end.
This improved performance has accomplished a reduction in the probability of seal failure at Entergy Operations' PWRs that is comparable to the reduction attributed to the proposed GI-23 resolution roquirements.
Entergy Operations believes that significant industry information does exist which can be used to reduce the uncertainty in the assetsment of off-normal RCP seal performance.
Informatioa available to Entergy Operations indicates that the potenti:1 for significant RCP seal leakage (ano the corresponding potential for core melt) due to a losn of seal cooling, is much smaller than that assumed in the generic cost benefit assessments of the proposed GI-23 resolution.
Fundamental design differences and Jmprovements provide considerably improved performance (normal and off-normal
, conditions) for ANO and Waterford 3 compared to the performance used in the NRC's cost benefit assessment.
Entergy operations believes that due to these design differences and improvements, our plants have already reduced RCP seal failure probabilities to an acceptable level.
The Station Blackout (SBO) Rule provides graded and flexible approaches to the resolution of a generic issue.
These approaches account for differences in technical bases and designs inherent in nuclear power plants.
We consider the proposed GI-23 resolution, however, to be an excessively broad based resolution for technical issues that are highly vendor and plant specific.
The resulting provisions of the GI-23 resolution, such as an 8 hour9.259259e-5 days <br />0.00222 hours <br />1.322751e-5 weeks <br />3.044e-6 months <br /> independent seal cooling system JGC09161.J19/JCLFLR-2
- _ ~....
. - -.... - -. ~. -.. _.. -
7 Mr. Samuel S. Chilk September 27, 1991 i
Ct:RO-91/00028 Paga 3 of 4 or the assumption of gross seal leakage rate adversely and inappropriately interact with the resolution of SBO.
These 1
provisions substantially complicate the approaches in j
resolution of.SBO and preclude the use of alternate ac power sources and coping strategies which utilities have employed, after extensive. evaluations and expenditures, for SBO.
These complications and constraints are introduced by the proposed 01-23 resolution without an appropriate, corresponding improvement in protection of public health and safety.
The responses to the specific questions are provided in for ANO and Waterford 3.
Question 6 requests comments on what method of imposition (e.g..
rule making, orders, or' generic letter) the NRC should use, if it is decided that additional-RCP seal requirements are necessary.
If the l
NRC determines the 1eed to establish new requirements and request plant mo.'fications, Entergy Operations believes the NRC must proceed on a case-by-case basis -- focusing only on those plants at which plant-specific evaluations demonstrate j
that some action may be appropriate to reduce the seal failure-induced risk.
The Commission should exercise its discretion to use means other.than generic rulo-making to accomplish this objective.
Recent judicial-interpretations highlight the discretion of federal agencies to choose case-by-case solutions where more generic actions are not appropriate.
Entergy Operations believes.that new requirements related to this: issue should not be imposed'on licensees unless_they are demonstrated to b3 necessary-and reasonable and show that the issuance of orders to individual plants 1s a viable means of
= addressing 1the GI-23 concerns at those plants which are shown to be vulnerable.
Entergy Operations has provided input to and fully endorses the comments being submitted separately by NUMARC, the CE Owners Group and the 3&W/N-9000 Owners Group, i
JGC09161.J19/JCLFLR-3
. = -.. -.
Mr. Samuel S. Chilk September 27, 1991 CNRO-91/00028 Page 4 of 4 Entergy Operations appreciates this opportunity to express our views on the proposud resolution to GI-23 and the Commission's consideration of our comments.
Sincerely, c
( ('(, N '
YN/JGC/ba\\)"\\
lottachments
' ce t Mr. T. W. Alexion Mr. Ross P. Barkhurst Mr. Neil S. Carns Mr. William T. Cottle Mr.
S. E.
Ebneter Mr. Paul W. O'Connor Mr. Byron Lee, Jr.
Mr. R. D. Martin Ms. Sheri Peterson Mr. D. L. Wigginton.
NRC Resident Inspector Office:
Arkansas Nuclear One Grand Gulf Nuclear Station Waterford Steam Electric Station, Unit 3 Central File (GGNS)
DCC (ANO)
Records Center (W-3)
Corporate File ( 39 )
(All above with attachments) i l
t l
JGC09161.J19/JCLFLR-4
. ~.
, _ _ _ _ _ ~. _.. _
LICENSE 082 ATTACIOiENT 1 ENTERGY OPERATIONS C0F0iENTS ON Ti(E DRAFT REGULAT0EY GUIDE ACCOMPANYING REGULATORY COST BENEFIT ANALYSIS, AND ON SPECIFIC TECl(NICAL AREAS Before any meaningful dialogue can be held related to RCP seal "f ailures",
a basic understanding of what consititutes a seal f ailure must be established.
For these discussions, it is considered appropriate to consider three possible reasons for unscheduled RCP seal replacements (1) seal performance degradation requiring seal replacement (no significant increase in seal leak-off rates).
(2) seal failure resulting in seal leakage rates within the normal charging capacity of the plant (i.e. not challenging plant safety systems), and (3) gross or catastrophic seal failure that results in seal leakage rates beyond the plants' charging capacity (i.e. LOCA).
As stated in the April Federal Register Notice. "In recent years the rate of seal f ailures appears to have decreased by roughly 50 percent.....Perhaps most importantly, during the past several years there have been no seal failures with = a high enough leak rate to be classified as a LOCA."
The proposed 01-23 resolution backfit analysis uses pre-1966 seal failure data and, even as the Federal Register statement does, lumps the three possible reasons for unscheduled RCP seal replacement described above together.
This is considered inappropriate since it results in the staff's goal to reduce the probability of catastrophic RCP seal failure (and corresponding risk) being equated to reducing the probability that an RCP seal degradation will not occur outside of a regularly schedule replacement cycle.
The latter goal is an economic one (which Entergy Operations has been aggressively pursuing) and is not appropriate for generic regulatory action.
The prpposed resolution seeks to address two major RCP seal feilure categories:
failures associated with normal operation (including mechanical-induced and maintenance-indaced failures) and e
failures occurring during off-normal conditions (due to loss of seal injection cooling and/or loss of cooling water, including station blackout (SBO) induced losses).
ANO-1 has four RCPs that utilizes the BJ N-9000 seals, while ANO-2 uses the BJ SU seals in the four RCPs.
Three of the RCPs at Waterford 3 utilir.e the BJ N-9000 seals and the-fourth pump has an AECL CAN 4 seal.
c l
Section I of Entergy Operations' comments addresses only normal operation I
related seal failure improvement efforts.
Section II addresses of f-normal operation related seal failures.
Section III discur.ses the applicability of the core melt estimates presented in NUREG-1401.
l 1-1 l
LICENSE.082 Section Is Nonnal Operation Related Seal Failure Improverent Efforts A.
RCp Seal Improvel..ent Fjfort s and Result s Entergy Operations' PWRs (Arkansas Nuclear One, Units 1&2 (ANO-1&2) and Waterford 3) have participated in various industry and owners groups activities aimed at improving seal reliability and parformance.
These included participation in the B&W Seal Perf ormance Improvement. Program, CE00 RCP Seal Meetings, and input into the Byron Jackson (BJ) development of the N-9000 Seal.
Activities specific to one unit's owners group activities were applied when appropriate to the other unit at ANO.
As a result of the B&V Seal Performance Improvement Program, BJ initiated an improved QA program on their seal parts.
A detailed seal part inspection plan has been implemented which identifies critical part dimensions and provides verification of those dlinensions for each seal component.
Deviations f rom the sp<.cified dimensions are evaluated and dispositioned as appropriate.
The seal part inspection results are provided to the purchaser so that independent verification may be accomplished if deemed necessary.
Entergy Operations has previously provided QC personnel to witness the BJ verification.
Sustained satisfactory vendor performance in the quality of parts has been demonstrated by QC review and irnproved seul perfonnance.
Improved RCP neal performance at ANO and Waterford is also rttributed to increased engineering involvement in resolving reliability issues and root cause analysis.
An example of this engineering involvement is the increased awareness of RCP shaf t aligtunent ef fects on rqa1 reliability.
!L _Sy,ee i fi,
N0 9
In addition to the various industry / owners group ef forts, ANO has had t
independent. contractors (MPR Associates, Inc. and B&W) review the seal maintenance activities / procedures and perform spare parts inventory inspections.
These independent review efforts resulted in the following Improvements that have contributed to improved seal performance at AN9:
1)
Two dif ferent rotating f ace materials (Carmet R-099 and Kennametal l
KZ-162B) were In use with the BJ SU and SU Hodified seals.
Past experiences at B&W plants indicated the Carmet material may be more susceptible to cracking and seal face fracture.
Therefore, the i
Carmet mt.terials were removed from the spare parts inventory and stocking now includes only the Kennametal rotating face material.
l l
With the development of the N-9000 seal design, materials have continued to be improved, as evidenced by the absence of heat checking on the seal faces.
2)
Improved seal lifting hoists have been purchased and utilized which provide finer controls for lifting seals and helped correct a clearance problem with the pump shaft during seal installation.
Additionally, the use of a seal lif t plats was implemented.
These improvements minimize the possibility of damage to the seal cartridge during transport and insta11atien and have effectively eliminated real " infant mortality" problems at ANO.
l 1-2 1
LICENSE.082 l
1 3)
Procedure improvements were accon.plished in conjunction with the improved seal handling techniques discussed above.
Additional, more detailed procedural steps were also added for installation of the N 9000 Scal.
Training of maintenance personnel with regard to RCP seal handling and maintenance has also improved.
Training on the RCP Seal maintenance procedures is now accomplished using an actual seal cartridge in s
" hands-on" c1cssroom training environment.
Waterford Speciffc The following operational experience and design improvement resulted in the improved performance of the N-9000 seals at Waterford 3.
The initial seal performance with the N-9000 seals for the first few months was very i
good.
The N-9000 seals were installed in March 1988. but after about six months, excessive controlled bleedoff leakage occurred, and seal cavity pressures began to mis-stage.
Inspection showed that the cartridges had accumulated a significant amount of rust-colored solid particulate matter, and the carbon f aces were heavily worn.
One cartridge showed evideaca of excessive shaft runout as there was indication of heavy contact between the shaft sleeve and the wtationary pcrts.
Subsequent investigation that included testing and analyses hat shown the design of the N-9000, as installed at Waterford 3 was not suited for operating in the reactor coolant water without seal injection.
Radial slots in the carbon face, designed to enhance lubrication in the gap between the dynamic sealing surfaces, allowed abrasive solids to enter the gap and cause excessive wear.
1h6 design of the secondary seal on the balance sleeve resulted in increased friction when tha particles accumulated betwecn the sleeve and the secondary seal.
The friction caucen loss of axial flexibility of the spring loaded carbon f ace.
As a 'esult of the investigation, improvements have been made to the seal design to overcome the causes of the malfunction.
Radial slots wt.re climinated from the carbon face with additional revisions made to maintain the required lubrication in the sealing gap.
The secondary seal configuration was changed to reduce the friction between the secondary seal and the balance sleeve.
A major factor in the improved performance of the Waterford 3 seals is improved maintenance techniques and training.
Waterford 3 has continued to work with AECL and BJ to enhance maintenance techniques and training.
Improved courses and lesson plans have been developed which cover RCP seal disassembly.
reassembly, and testing including RCP seal shaft reuioval had replacement.
The training involves both classroom instruction as well as intensive hands'on training.
1-3
LICENSE.082 Some of the objectives which are cover ed, for example, by one lesson plan are the following e
Basic understa sng the function of face type shaft seal design ramiliarization with training manual i
Disassembly and assembly of all subassemblies Build up of preassembled components on lower and upper shaft sleeve o
e Measurement of axial movement 1
Spare parts ordering procedure Lesson plans addreas general factors which relate to seal performances such as seal qualities for high reliability, long lifetime, high seal integrity 1 seal maintainability like good design features, quality assurance of components, and component refurbishingt RCP seal ins t rumentation: and common seal assembly problems, such au grease on seal f ace. 0-ring cuts, projections and broken welds, hidden dirt, and split ring seating.
Lesson plans also address specialized maintenance areas, such as principles, techniques, and equipment used for flatness measurementi equipment and techniques used in lapping of seal parts:
RCP seal post-service examinations precautions, inspections, critical vieasurements, and doctunentation requirements for RCP seal cartridge rebuilding; and criteria for post-maintenance testing.
i B.
RCP Seal _ inspection and Specification Verification improvements vs.
i
{roposed QA Requirements The RCP seals are classified as'a Q (saft.ty related) component at ANO and Waterford since certain portions of the seal perform a reactor coolant pressure boundary function.
Seal components which are purchased Q include seal flange, pressure breakdown devices, and seal stubout piping ahd flanges.
Other seal parts which are not related to the reactor coolant pressure boundary (such as elastomer o-rings and u-cups, and stationary and rotating faces) are purchased as augmented quality components with procurement and manufacturing in accordance with BJ Technical Manuals.
10CFRSO Appendix B requirements are not placed on the vendor for augmented quality components purchased by ANO and Waterford.
While some RCP seal components do not have 10CFR50 Appendix B
requirements placed on them, this does not mean that inadequate quality components are installed.
To the contrary, as a direct result of the Seal Performance improvement Program, BJ has implemented improvements in its QA program for seal components.
Improvements include an inspection plan for each ceal component.
This plan specifies critical dimensions and attributes for individual seal parts and requires the as-found 1-4
~-
LICENSE.082 j
i condition of the supplied part to be documented.
Deviations from the required dimensions and attributes are evaluated on a case by case basis with appropriate justifications required for continued use of the component.
These inspection results are provided to the purchaser of the RCP seal for verification as required.
Improvements in the quality of parts supplied have been observed at ANC BJ's QA program for sedi and Waterford as a result of parts.
For these reasons the implementation of a 10CFR50 Appendix B QA program on RCP eals will not provide seal comronents of substantially better quality than those currently supplied to ANO and Waterferd by BJ.
NUREG/CR-51ti/ and Draf t NUREG-1401 indicate that the nuclear indstry will observa a cost during refueling outages by impicmenting on Appendix B QA progra RCP seals.
RCP seals are changed at predetermined refueling outages at ANO and Waterford as a preventative measure and, as a result, have reduced forced outages and improved equivalent availability go..ls for next operating cycle.
Additionally, past seal operating expeilenc., and seal performance during the previous cycle provide inputs to the det as to which RCP seals get changed during a refueling outage.
slon it is our opinion that the presence (or absence) of a 10CFR50 Appendix B The re fo re,
QA program will have no significant ef fect on th decision as to whether a seal gets changed during a refueling outage.
As such, no reduction in the numbot of seals which are changed during an outage will be observed as a result of the implementation of a 10CFR50 Appendix B QA program.
This essentially eliminates the proposed cost savings for operation, averted property cost, and operational occupational exposure areas of the cost benefit analysis, and leads to a cost /bunellt analysis that does not favor the proposed resolution.
C.
Irmroved Seal Inst rumentation and MonitorinR The RCP seal instrumentation ins.alled at ANO and Waterford is consider to be fully adequate to monitor RCP seal performance.
The ins t rumentation installed is summarized in the response to Federal Register Question 3.2 (Attachment 2).
The installed RCP seal instrumentation at ANO and Vaterford supports the appropriate operational and maintenance decisionu.
The presence of additional seal monitoring instrumentation and the regulation of seal monitoring requirements would provide no substanti.s1 added oenefit and would have no significant effect on the frequency of kCP acal changeouts at ANO or Waterford.
Therefore, with proper reflection of site specific
- benefits, the cost / benefit analysis would not favor the proposed resolution.
1-5
..- ~
LICENSE.082 Section 11: Off-Normal Operation Related Seal Tailures For of f-normal conditions (due to loss of seal trjection cooling End/or loss of cooling water, including station blackout induced losses) the major concerns that the NRC has involve seal failures due to adverse temperature effects on secondary seal elastomer materials, and e
on perf ormance instabilities at the primary seal f aces related to e
coolant flarhing and two-phase flow.
The NRC censiders current RCP secondary seal materials ta be susceptible to unacceptably accelerated degradation when seal cooling is lost and seal temperatures approach normal RCS operating temperatures.
Item 3 of the proposed 01-21 resolut ton proposen an independent cooling water syatem for RCP seals to preclude failure of the RCP seals (due to loss of cooling) that could progress to a small-break LOCA and, with additional failure, a subsequent core melt.
The NRC resolution presumes that a failures of the seal will result for any loss of cooling water event of significant duration.
The various NUREGs on this subject provide analysis of judustry sponsored loss of seal cooling test results and analysis of NRC Sponsored testing for the specific purpose of GI-23 re m19 tion.
In addition, generalized estimates of core melt frequency are made in the NUREGs for loss of cooling water and station blackout induced small-break LOCAs.
This section of conenents addresses of f-normal conditions related seal failure susceptibillty for B.1 RCP seals (i.e. N-9000 and SU used at ANO and WLerford)
A.
Elastomer Related Seal Failure Susceptibility NUREG/CR-4948 st ates in Section 4.2.1.3,
" Conclusions on the Degradation of Polymers and Stability of Seals", that BJ static o-rings and polymer secondary seals are n_ o_t, expected to fail due to high temperature t
extrusion.
This conclusion is then dismissed based upon uncertainties associated with their ability to withstand multiple loss of cooling events. etc.
This dismissal doesn't recognize ANO and Waterford's practice to closely monitor and trend seal performance. such that any consequential degradation would be noted and dispositioned with a seal replacement if warranted (i.e. seals will not be subjected to multiple events to the point that - their integrity is significantly degraded).
This dismissal also does not recognize that, as described later, actual i
plant expericoce has demonstrated a resistance of BJ elastomers to high temperature extrusion, adding credibility to the original NUREG conclusion.
l l
1-6
1.! CENSE.082 NUREG/CR-4948 also states in Section 4.2.1.3 that BJ seals are susceptible to instability due to embrittlement. of nitrile U-cups.
The noted embrittlement is apparently based on NUREG/CR-4821 findings.
This l
report findirg was based on test results using U-cups that were cut and spliced to fit the ti st rig.
Wen tested, failures occurred at the splice, causing flui4 to leak away from the U-cup and exposing the elastomer to dry but (resulting in embrittlement).
This is not representative of inservice U-cups since they do not contain spliced joints.
Without this type of f ailure the U-cup would remain wetted with either steam or water and not be subjected to the dry test conditions that lead to embrittlement.
Therefore, the noted embrittlement susceptibility is not considered to be founded on valid tests and not representative of BJ RCP seal performance.
Therefore. NRC test results indicate a lack of susceptibility to high temperature induced elastomer failures.
Act ual plant experience (discussed later) has also shown BJ RCP seals to be relatively insensitive to high temperature induced elastomer degradation.
B.
Coolant FlashinR and Two-Phase Flow Induced instabilities NUREG/CR-4948 based on input from AEC1.
(NUREG/CR-4077 and NUREG/CR-4821),
considered the effects of increased.
off-design temperatures that could bc experienced during station blackout conditions.
These increased temperatures can cause degradation of polymer seal components and also affect the pres e distribution of the flow through the gap between the seal. rings.
Increased pressure loadings between the seal rings is stated to have the potential to cause the primary seal to open to a large gap. or to pop open completely, resulting in increased leakage.
NUREG/CR-4077 summarized testing performed to estimate leakage rates and seal stability during two phase flow conditions across the seal f ace.
For the eleven tests performed, one of the tests resulted in the seal f aces popping open and this was initiated externr*
- by operating valves in the wrong order resulting in no backpi sure on the seal.
NUkEG/CR-4821 concluded that BJ seals have a higuer balance ratio than Westinghouse or Bingham International seals and are the least susceptible to instability.
It also concluded that unstable behavior would be predicted for the second and third stage seals during SB0 conditions and that under certain conditions with U-cup hardening. the first stage seal could become unstable.
Ilowever, the seal testing summarized in the above documents is based on the Westinghouse seal design.
The differences between the BJ seal design and the Westinghouse design are significant.
The seal popping tests addressed in the above documents do not adequately correlate test results or prove by experimentation the applicability to the BJ seal design for the following reasons 1-7
4 LZCENSE.082 1)
Insufficient information was utilized f or the tested secondary seal polymer to establish its equivalency ard applicability to the DJ U-cup m1terial. design, or clearances.
As previously noted. the observed U-ct p hardening occurred due to f ailure of a s; ced joint is not representative of inservice BJ RCP seals and is therefore not considered valid.
2)
Seal tests were accomplished using Vestinghouse type hydrostatic seals with a high seal leakoff rate.
The BJ seal utilizes hydrodynamic type seals and a seal face leaka2e rate which is orders of magnitude less than the tested seal.
3)
An individual seal stage was tested vice the tandem staging design
- ised in the BJ seal design.
The possible dampening effects of tandem stages utilizing equal pressure drops and staging interaction was not evaluated.
4)
Seal backpressure was fixed at atmospheric pressure vice actual pressures which would be experienced.
The presence of pressure breakdown devices in the BJ seals and the backpressure they provide were not considered or evaluated by test.
5)
Seal inlet conditions were established using low margins to saturation.
While this information was useful in detertnining the critic.a1 balance ratio with zero backpressure, it did not reflect whether or not flashing would occur in an actual RCP seal, particularly in the lower stages or in the upper seal stages considering the backpressure supplied from the volume control tank.
This lack of established applicability of the AECL tests to the BJ RCP seal design 12 further cornpounded in the cost / benefit analysis of NUREG/CR-S167.
NUREGICR-5167 utilizes a time dependent Westinghouse seal failure model that does not recognize the reduced vulnerability to high temperature extrusion of BJ seal elastomers nor does it recognize the basic design differences (i.e.
hydrodynamic thin film design vs.
hydrostatic thick film design as.nentioned above) that results in orders of magnitude less seal face leakage from BJ RCP seals at all modes of operation / conditions.
C.
Industry Loss af RCP Seal Cooling Testing and Operational Experience NUREG/CR-4948, 4821. and NUREG-1401 evaluate industry ts. sting previously performed by - EDF on the Westinghouse seal, the 50 hour5.787037e-4 days <br />0.0139 hours <br />8.267196e-5 weeks <br />1.9025e-5 months <br /> SB0 test on the St. Lucie BJ seal, and the 30 minute San Onofre loss of seal cooling test on the BJ seat.
The NRC analyses discount the results of these testa for a variety of reasons-such as seal designs being ditferent, shaft motion not allowed, the pump was rotating during the test vice stationary as in the case of SBO, etc.
To the contrary, the fact that SB0 conditions and seal configurations are not exactly duplicated during the test, does not mean that useful information regarding seal reliability during a SB0 cvent could not be derived.
1-8 l
1.1 CENSE.088 Testing performed on va.-ious BJ seal cartridges and operational experience by another domestic CE plant which isolated cooling water to two RCp seals at approximately.30' F for a period of six hours and l
subsequently operated the pumps for three monshs until the next refueling outage, clearly demonstrate that the results of testing identified in the various NUREGs do not adequately reflect what would occur to BJ RCP seals during a loss of seal cooling water or 5B0 event.
Other operational experience includes a natural circulation cooldown at St. 1,ucie in 1980 during which CCW was not available for a period of 11 hours1.273148e-4 days <br />0.00306 hours <br />1.818783e-5 weeks <br />4.1855e-6 months <br />.
While the seals were replaced as a precaution, followup examination revealed that the seals were not damaged and the pumps could have been restarted.
In 1985 Waterford 3 experienced a loss of CCW to three RCPs.
After 45 minutes, one seal leaked externally at approximately three gpm.
During the event, the vapor seal failed as a result of controlled bleedof f (CBO) isolation but the lower and middle seal stages maintained their function.
The plant was cooled down, and the seals maintained their functional integrity throughout the event.
The newly designed BJ N-9000 RCP seat cart ridg. (utilized at ANO-1 and Waterford 3) has also had ex t. ens ive t<
performed to establish its capability to withstand loss of RCP seal aing conditions.
The N-9000 design is based on a systematic approach of over five years of RCP seal research which considered factors such as reactor coolant system steady-state and design trancients, compatibility with existing equipment, and seal response to extetual influences (such as pump shaft orbital characteristics experienced during pump operation).
To confinn the results of this extensive design effort the N4000 seal has successfully undergone over 400 individual component, single stage, and cartridge t es tr.
including over 6.600 hours0.00694 days <br />0.167 hours <br />9.920635e-4 weeks <br />2.283e-4 months <br /> of verification and qualification testing in a three stage cartridge configuration (note, this testing utilized a seal taken from actual plant service conditions).
Tests completed include steady state and transient operating conditions which meet or exceed all PVR reactor coolant sys,tems requirements.
In addition, the three stage cartridge has also successfully completed nearly eight hours of full scale testing without component cooling or seal injection flow.
This test was specifically designed to evaluate seal performance under postulated station blackout conditions.
The test results dynonstrated that the seal leakage rates expected for N-9000 RCP seal are orders -of magnitude less than that predicted by the time dependent Westinghouse seal failure model.
Item 3 of the proposed resolution to GI-23 and its resulting cost benefit analysis are based solely on the presumption that a loss of seal cooling during a SBO would result in excessive Icakage rates from the RCP seals and contribute significantly to the probability of core melt.
This presumption is based on conclusions reached by analytical modeling and testing referenced in the above NUREGs and the magnitude / quantity of past seal failures at operating plants.
This presumption is flawed for the following reasons:
1-9 a - ---- -.. --
a -.
~.
- ~
b1 CENSE.082 1)
Experimental testing reported in the foregoing NURECs has been inadequate since actual seal
- design, installation, and environmental conditions during a SBC were not duplicated, and 2)
No consideration has been made as to which seal "f ailures" (and the corresponding leakage magnitude) experlenced in operating plants have been attributable to loss of seal cooling events.
The entire spectrum of seal "fallures" are used to justify backfit of a cooling water system for station blackout events.
Most of the
" failures" that have been experienced in the industry did not involve a lack of cooling water to the RCP seals.
As such, the expectation that this backfit would reduce the frequency or magnitude of seal failures does not appear to be valid.
The decision in draf t NUREG-1401 to reject results of full-scale tests and industry experience to verify acceptable RCP seal performance during loss of seal cooling events (on the basis that " satisfactory performance without seal cooling cannot be demonstrated")
has resulted in over-sitnplification cf the analysis and overconservatism in estimating the probability and consequences of seal failures.
While the regulatory analysis has appropriately identified some minor limitations of the documented full-scale tests, it has relied on a significantly more I
limited experimental process using excessive simplifications, approximations, and extrapolations of research data to highlight potential safety concerns associated with RCP seal f ailures.
Thus, the regulatory analysis has based its concerns on non-prototypic sirnplified test data, and verified qualitatively that the resulting predictions provided conservative estimates of seal leakage rates.
This process has 2
led to the omission of the more reliable data obtained through full-scale tests, which, as concluded in NUREG/CR-4400, constitute the preferred approach to quantify the consequences of seal failure. This process also appears to be in direct conflict with the proposed resolution recommendation to. improve RCP seal performance through improved inspection and specification verification (as a result of a full 10CFR50 Appendix B QA progtam).
In one sense however, the NRC experimental tests have confirmed that an RCP seal fabricated in poor compliance with the vendors' specifications will perform poorly, but again, this doc:s not representative quantitative assessment of actual RCP seal equate to a performance.
Therefore, the industry testing and operational experience that has been-performed for BJ - RCP seals serves to confirm conclusions that the time
' dependent Westinghouse seal failure model c:es not-recognize the reduced vulnerability of BJ RCP seals to high temperature elastomer extrusion, nor does it recognize the reduced leakage rates expected from BJ RCP seals at all modes of operation / conditions, i
1 - 10
.. _ _ - -., =
1.1 CENSE. 003 I
Section III:
Applicability of the Core Halt Estimates Presented in NUREG-1401 As discussed previously. ANO has made changes in the RCP seal rotating face materials utilized that results in improved performance compared to that assumed in the generic NRC cost benefit assessment.
Also discussed previously, the redesign efforts for the new BJ N-9000 seal have served to further reduce the vulnerability of this seal to loss of seal cooling related failures.
Discussed below is an additional design difference for ANO-2 and Vaterford that even the NUREGs recognize as reducing the risk due to RCP seal f ailure by close to an order of magnitude.
A.
Improved Performance Due to Fourt h__ Stage Vapor Seal on Combustion Engineering NSSS Plants NUREG/CR-4643 and NUREG/CR-4400 both noted that, if credited, the fourth stage vapor seal reduces the likelihood of RCP seal f ailure (compared to the failure of three stage RCP scals) by close to an order of magnitude.
NUREG/CR-4077 actually proposed as a
candidate design change for resolution of GI-23 the "addi' ion of a back-up ' safety' seal to each cartridge".
The August 1988 expe.f~:ce at ANO-2 in which a ruptured RCP seal sensing line hydraulically failed all four seal stages provides evidence that (even in a degraded state) with full RCS pressure applied to the vapor seal, leakage is considerably less than that assumed in the time dependent Vestinghouse seal f ailure model (40 gpm versus 400 gpm).
It is interesting to note that in resolution of the Post-TH1 Action Pls Item II.K.3.25 the NRC Safety Evaluation Report for ANO-2 quotes L t..
expected BJ RCP seal performance as, "should the vapor seal also fall, then leakage would not exceed 40 gpm".
The August 1988 ANO-2 RCP seal failure resulted in an approximate 40 gpm leakage, a significant portion of which could have been out the failed RCP seal sensing line.
Therefore, the presence of a fourth stage vapor seal that is designed to withstand full RCS pressure provides the same magnitude of risk reduction (compared to the generic baseline plant used in the NRC cost benefit assessment) as that estimated for addit.lon of the proposed independent seal cooling water system.
B.
Proposed GI-23 Resolution Effectively Negates Accepted Approaches to, ComplyinR With 10CFR50.63 In 1988, the NRC approved the Station Blackout Rule, 10CFR50.63, which l
requires that commercial nuclear power plants be evaluated to deterinine
- t. heir capability to withstand and recover from a loss of all ac power of a specified duration.
The station blackout rule in essence allowed t
plants to pursue an alternate. ac ( AAC) source or ac-independent (coping analyses) approach in demonstrating compliance with 10CFR50.63.
ANO is installing an independent AAC power source and Waterford has pursued a coping analysis approach in demonstrating compliance with 10CFR50.63.
l l
l l
r 1 - 11
- - -.....~.
LICENSE.082 A.N.O.
At ANO, Entergy Operations has committed to install.an independent AAC power source capable of providing electrical power under blackout conditions to the affected unit.
The AAC power source that will be utilized at ANO will meet the criteria specified in Appendix B of NtHARC 87-00.
The source will be available within 10 minutes of the onset of the station blackout event and will have sufficient capacity and capability to operate the necessary systems to bring the unit to a safe shutdown condition and maintain that condition for the duration of the event.
Although the design of the BJ RCP seals at ANO are such that they are much less vulnerable to loss of cooling water induced failures than the generic NRC assessment concludes, the installation of an AAC power source will reduce any residual risk due to postulable 5B0 induced RCP seal failures.
This is due primarily to the fact that the alternate power source w!!! restore the ability to provide safety injection in the unlikely event a RCP seal failure should occur (not due to the fact that it will provide the-_ capability to restore RCP seal cooling).
Consequently, the desired risk reduction for resolution of G1-23 is considered to be more than satisfied by the planned installation of an AAC power source at ANO.
However, a concern exists in that the draft Regulatory Guide DG-1008 reqvtrement is not consistent with Position 3.3.5 of Regulatory Guide 1.155 which discusses the regulatory requirements pertaining to AAC power sources.
The Staf f's position states that "AAC power source should be availabic in timely manner after the onset of station blackout," and that "the time required for making this equipment available should not be more than 1 hour1.157407e-5 days <br />2.777778e-4 hours <br />1.653439e-6 weeks <br />3.805e-7 months <br /> as demonstrated by test."
The 10 minute AAC availability provisions were established as follows:
"if the-AAC power source can be demonstrated by test to be available to power the shutdown buses within 10 minutes of the onset of the station blackout, no coping analysis is required."
NUMARC 87-00 Section -- 7.1. 2 and Q&A GS respectively clarify the AAC availability and the 10 minute criteria by stating that:
'Available within 10 minutes means that circuit breakers necessary to bring power to safe shutdown buses are capable of being actuated from the control room within that period."
1 - 12
~ -........ - -
-.u..
1.1 CENSE.082 and "The 10 minutes requireraent was meant to cover the period from when the operators realized that a station blackout has occurred until the AAC source is started from the Control Room.
Therefore, operators would perform the immediate steps in the E0Ps to verif y scram. primary system parametar, etc.,
and attempt to restore offsite power and start the EDGs f rom the Control Room per the E0Ps.
Station blackout begins when actions f rom the Control Room are unsuccessful in restoring offsite and onsite emergent AC power.
The ten minute criteria is met if the operators can start and be ready to load the AAC source within the following ten minutes from within the Control Room."
Appendix E to NUREG-1401 requires PWR licensees to " provide RCP seal cooling during of f-normal plant conditions such as station blackout" by initiating " cooling flow to the RCP seals in accordance with the specifications used for normal operation within the 10 minutes from the start of the off-normal plant conditions.
(Appendix E to NUREG-1401 states for jastification of the 10 minute requirement that, " analysis has shown that leakage through the RCP seals will remain subcooled during the first to minutes" following the onset of the loss of seal cooling event.)
This 10 minute requirement is evaluated in draft Regulatory Guide DG-1008 in view of SB0 guidance.
In this regard, draft Regulatory Guide DG-1008 states that "if, as part of the implementation of the station blackout rule, a plant is reestablishing seal cooling within 10 minutes (e.g., by an alte rnate ac power supply which powers the seal injection function), then seal cooling is not considered lost."
lloweve r, the proposed requirement that " seal injection function" or
" independent seal cooling" be re-initiated within 10 minutes f rom the onset of the SB0 (or loss of cooling event) may effectively disqualify the use of 10 minute AAC sources, if the 10 minute period f ollowing the onset of the initiating event does not cecommodate the performance of LOPS.
The GI-23 regulatory analysis does not provide a technical basis for selecting the 10 minute criterion pertain!ng to the initiation of the
" independent seal cooling" provisions duris.g off-normal conditions. This issue is briefly discussed in Appendix A to DG-1008, which includes functional design criteria of the independent seal cooling function.
These criteria require that
" sufficient seal cooling to maintain manufacturer's recommended temperature limits" be provided, and the seal
" cooling within 10 mirutes" of the onset of the loss of cooling event be re-established to " ensure that two-phase flow is avoided.
The discussion of the technical findings of NUREG/CR-4948 also fails to give e clear indication on the source of the 10 minute estimate.
In this regard, it appears that the proposed resolution attempts to apply the results of the Staff's res m ch data which indicate that potential unstable behavior is expected once saturated inlet fluid conditions at a 1 - 13
1.1 CENSE.082 seal stage are reached.
However, a clear justification of the selection of this design criteria has not been provided.
In particular, the need to impose the excessively stringent 10 minute timing requirement has not been established in view of the heat-up characteristics of the RCP seal assembly and the cnset of two-phase flow in the seal stages.
In discussions with Westinghouse regarding its observation that hydraulic instabilities were not recorded during 24 loss-of-all-seal-cooling events.
the staff pointed out that "none of the twenty-four loss-of-seal-cooling were of long enough duration to cause hydraulic instability failure" (even the events lasting 45 and 65 minutes may not be long enough).
In addition. NUREG/CR-5167 Appendix A estimates seal failure to occur approximately two hours after initiation of a SBO.
Therefore, justifiable technical basis does not appear to exist for the proposed requirement to restore RCP seal cooling within 10 minutes from the onset of a loss of seal cooling event.
Further, the proposed resolution is apparently inconsistent with requirements established as part of the resolution of the broader issue of SB0 (i.e.,
Waterford By assuming that all RCP seals will fall catastrophically, shortly af ter the onset of a loss-of-cooling event regardless of their design dependent susceptibility to failure and irrespective of the features of their auxiliary systems, the proposed resolution severely and adversely undermines all AC-independent SB0 compliance strategies.
The staff's regulatory problem statement does not attempt to evaluate the probability nor consequences of seal failure, and does not give any realistic estimates of the magnitude of the resulting leakage rates.
The justification for neglecting to account for the design dependent behavior of RCP seals following loss of seal cooling is discussed in NUREG-1401.
In this regard, the regulatory analysis states that providing an engineering solution to the loss of cooling seal failure problem would resolve the unknowns associated with " quantifying the relationship between loss of seal cooling, seal failure, the subsequent time dependent behaviors of the RCP seal assembly, and of the coolant leakage rate."
Using this justification, the regulatory analysis has not accounted for any differences in seal design features, and has not differentiated between various RCP seals according to their susceptibility to failure.
In doing so, all industry designs and experience were enveloped by overly conservative estimates of the consequences of RCP seal f ailures, and some of the more important conclusions of the research studies conducted to support the GI-23 resolution were neglected.
In particular, technical findings regarding differencec in seal design and in seal-component materials, which were demonstrated to play a critical role in determining the probability of seal failure, were not reflected in the proposed resolution (see e.g.
NUREG/CR-4821 NUREG/CR-4643, and NUREG/CR-4027).
Addh i nally, operating experience events and the results of testing wMd support the resilience of the seal to function under SB0 type cow 1stu ns or which support the validity of minimal leakage from the l
seals have not been reflected in the proposed resolution.
l 1 - 14 l
LICENSE.082 This limitation of the regulatory analysis effectively precludes the i
consideration of the AC independent coping approach for complying with 10CFR50.63, since the design of the <oping methods would be significantly compromised by an overly conservative upper-bound estimate of the expected leakage rate.
The proposed GI-23 resolution will ef fectively negate 10CFR50.63 by eliminating coping strategies to demonstrate compliance with 10CFR50.63.
1 - 15
LICENSE.082 i
ATTACHMENT 2 RESPONSES TO GENERIC ISSUE 23 FEDPRAL REGISTER QUESTIONS NN Federal Register Question 1.1 Has your operating experience with RCP seals changed since 19837 If it has, then information regariing the history of RCP seal failures, including occurrences of forced outages is of interest.
Information regarding all types of operation luding start-ups is desired.
Response
ANO-1 ANO-1 began commercici opetat ion in December 1974 RCP segl performance since 1983 has impro ed; since_ 1988 such improvements have been consistently documented.
P rior to 1988, ANO-1 utilized the Byron Jackson (BJ) 3 stage SU unmodified (1974-1982) and modified (1983-1987) seal.
In 1988 ANO-1 upgraded the RCP seals to the N-9000 seals by BJ.
5tCP seal degradation and failures have resulted in forced outages for seal replacement.
The RCP seal f ailures and catastrophic f ailures experienced at ANO-1 are summarized as follows:
1 - 08/76 RCP-D........ 25 gpm seal degradation 2 - 12/77 RCP-C........ 6 gpm seal degradation 3 - 05/80 RCP-C........ 400 gpm catastrophic seal f ailure 4 - 08/82 RCP-C........ 28 gpm seal degradation No forced outages have occt. red at ANO-1 since 1982 due to excessive leakage from RCP seals.
Implementation of the proposed improveaen u specified in the proposed resolution to GI-23 would not have prevented eivst of the types of seal failure or degradation that has been experienced at ANO-1.
More significantly, the proposed resolution of GI-23 would not have mitigated the 5/80 catastrophic failure because the root cause was determined to be (AND PROVIDE).
AN,,0_2 ANO-2 began commercial operation in March 1980.
As a result, the database for RCP seal perfotrance during the time fr ee from 1980 through 1983 is small and meaningful comparisons of pre-1983 and post-1983 RCP seal performance is quite limited. However, RCP seal performance since 1985 has improved.
2-1
.~
LICENSE.082 The incidence of RCP seal replacements during forced outages prior to 1985 was seven seals.
In all but one case, the forced outages experienced were initiated by a source other than RCP seal replacement.
ANO-2 has experienced only one RCP seal failure and no catastrophic RCP seal failures since 1985.
The failure was initiated by a RCP middle seal sensing line break resulting in the starvation of CB0 flow from the remaining.eal stages.
This eventually resulted in a hydraulically induced seal failure, since CB0 flow provides cooling and lubrication for seal faces.
Implementation of the improvements specified in the proposed resolution to Generic Issue 23 would not have prevented this type of seal failure.
Waterford 3 Waterford 3 began commercial operation in September 1985.
The Wat erford 3 operat ing performance history for seals is summarized in Table 1.
Since October 1982, there have been ten occurrences of inadequate seal performance.
One of the occurrences necessitated a
forced outage.
Two important observations about the operating experience with the RCP scals are highlighted:
1)
The seals held pressure and prevented excessive leakage for all occurrences 2)
The seal performance history has significantly improved since December 1995. The October 1988 occurrence was as a result of a new type of seal installation-with seal degradation attributed to seal manufacturing tolerances and design deficienciese The October 1987 c currence was the result of the rotating baffle in RCP 2B coming lose and hitting the lower stage of the seal, causing the lower seal stage to fail.
During the plant startup, RCP 2B failed to stage properly.
Since 1985, Waterford 3 has not had any RCP sesi performance problems which necessitated a forced outage.
l l
l I
2~2
LICENSE.082 TABLE I WATERFORD 3 RCP SEAL PERFORMANCE HISTORY FORCED MAX DATE TYPE OUTAGE LEAKAGE COMMENTS RCP1A; LOVER STAGE HAD 10/82 FU NO ZERO DP DURING COLD HYDRO.
CHANGED AFTER HYDRO.
04/83 SU NO RCP1A&IB; PRESSURE OSCILLATIONS DURING HFT.
CHANGED SEALS AFTER HFT.
RCP2A; HIGH CB0 FLOW 01/85 SU NO AFTER FILLING SYSTEM.
02/85 SU N0 3.0 GPM OPERATOR ERROR ACTUATED CONTAINMENT ISOLATION, SECURING CCW TO SEALS.
ALL SEALS CHANGED OUT.
RCP1A&2B; AFTER PLANT 05/85 SU NO TRIP, LOWER SEAL STAGES FAILED ON RESTART.
REPLACED DURING TG OUTAGE.
RCP2A; CB0 FLOW AND 07/85 SU NO PRESSURE OSCILLATIONS INCREASED OVER 2 WEEK PERIOD.
REPLACED DURING TG OUTAGE.
10/85 SU YES RCP2B; CB0 FLOW AND STAGING PRESSURE OSCIL-LATIONS-INCREASED OVER l
~
SEVERAL WEEKS FORCING SHUTDOWN.
DURING RECOVERY FROM TRIP, 12/85 SU NO 2B FAILED TO STAGE, 2A STAGE DP DECREASED. ALL FOUR SEALS REPLACED.
DURING PLANT STARTUP RCP 2B 10/87 SU NO SEAL FAILED TO STAGE t
PROPERLY.
SEAL WAS REPLACED WITH N-9000 SEAL.
PLANT SHUTDOWN FOR NON-SEAL 11/88 N-9000 NO RELATED CAUSE; DURING PLANT STARTUP SEALS WOULD NOT STAGE AT NOP/NOT.
ALL SEALS REPLACED WITH SU TYPE SEALS.
2-3 4
L2 CENSE.088 NRC Tederal Register Question 1.2 If your operating experience has changed, to what do you attribute the change (e.g.,
improved quality assurance and quality control, improved maintenance, i
better procedures, improved instrumentation design changes)?
Response
ANO-1&2, Waterford 3 RCP seals are an important part of the RCS which have the potential, due to forced outages, to contribute significantly to lost generating capacity, increased cost, and exposure.
Consequently, considerable attention and effort are given to issues that arfect RCP seal performance.
Enhanced seal performance is the result of the integration and implementation of many factors which affect seal performance including quality assurance and quality control, maintenance, maintenance training, design improvements, operating procedures, and increased engineering involvement and root cause analysis.
A.
Improved Quality Assurance / Quality Control As a result of the B&W Seal Performance Improvement program, BJ initiated an improved QA program on their seal parts.
A detailed seal part inspection plan has been implemented which identifies critical part dimensions and provides verification of those dimensions for each seal component.
Deviations from the specified dimensions are evaluated and dispositioned as appropriate.
The seal part inspection results are provided to the purchaser so that independent verification may be accomplished if deemed necessary.
This was previously performed as a
100% inspection; however, due to improved vandor performance, the inspection frequency has been reduced.
Entergy Operations has previously provided QC personnel to vitness the BJ inspection prior to shipping in lieu of on-site verification.
Sustained satisfactory vendor per formance in the quality of parts provided has been verified by Entergy Operations QC review and confirmed, in part, by improved seal performance.
l l
B.
Improved Maintenance ANO-1&2 Training of maintenance personnel with regard to RCP seal handling and i
maintenance has also improved.
Training on the RCP seal maintenance procedures is now accomplished using an actual seal cartridge in a classroom training environment providing craft personnel hands-on training.
I In addition to the various industry /awners group efforts.
ANO had independent contractors (MPR Associates, Inc.
and B&W) review seal maintenance arrivities/ procedures and spare parts inventory inspections.
These reviews esulted in the following improvements which have improved seal performance at ANO:
l.
1)
Maintenance procedures were changed in terms of clearer procedures with improved seal handling techniques.
2-4
. - _ - ~
~ - ~ -.
LICENSE.083
.',/
2).
. Improved seal lif ting hoists were purchased which - provided - finer controls for lif ting seals and corrected a clearance problem with the pump shaft during-seal installation.
Additionally _the use-of a
a - seal. lif t. plate was implemented.
These improvements minimized
~
possibility of damage to _ the seal-cartridge during transport and installation and have effectively eliminated seal- " infant mortality" problems.
3)
Improved operator sensitivity to Intermediate Cooling Water (ICW) for ANO-1 and Component Cooling Water (CCW) for ANO-2 temperature control has resulted in fewer thermal -transients to the eenis and has minimized pressure oscillations in the seal stages.
Waterford-3 A major-. = f actor - in the improved performance of the RCP seal is
_ aterford 3 has worked improved _ maintenance techniques - and _ training.
W
.with AECL and BJ to enhance maintenance techniques and training.
Improved courses and lesson = plans have been ' developed which - cover RCP
-seal disassembly, reassembly, and testing _ including RCP seal shaft removal and replacement.
. The training involves both classroom
-instruction as well as? intensive hands _on training.:
C.
Improved Design Changes-ANO-1&2-ANO-1&2' have -participated in. various industry and owners groups activities aimed at improving seal reliability and performance.
These
' included participation; in the B&W Seal Performance Improvement Program.-
CEOG RCP; Seal Meetings.. and ' input into _the BJ development of ' the - N-9000 seal.
- Activities Especific to; one unit's owners group' activities were applied' as ~ appropriate to the other unit at-ANO. Design and material.
changes.that have._been made as a result of these efforts' include:
-1)
Seal = operating experience has - improved at ANO by the installation of flex: hoses on RCP seal instrument sensing lines.-
This installation minimizes the possibility of_ seal sensing.line failures due to low stress. high cycle fatigue..The possibility of
. pressure transients which occuriduring a seal sensing line break
' ANO has not experienced a sensing-line' t
are - therefore minimized.
break since the installation 'of the flex hoses.
'2)
The improved RCP SU Modified seal performance -(1985-1988). has been
~
attributed to a material change of the.otating. face.
Kennametal KZ-801 L uas/is used: versus the original Carmet R-099.
(ANO p utilized this material when the SU Modified was is service;- ANO-2 still utilizes this material in the SU design.)- 'The new material i
~is.much less susceptible to fracture.
L l_
2-5 t
l __
ll
4 L2 CENSE.083 3)
In 1988 the N-9000 seal was installed in ANO-1 with a complete design change in the stationary and rotating face materials.
This material change has resulted in improved performance, as evidenced by the absence of heat checking.
The N-9000 seal has been
- designed, tested, and manufactured in accordance "with the requirements of a detailed equipment specification and is expected to operate for a minimum of two to three fuel cycles between refurbishments.
Waterford 3 Four N-9000 seal cartridges were installed at Waterford 3 on March 1988.
After about six months, excessive CB0 leakage occurred and replacement of the eattridges was required when mis-staging of the seal cavity pressures occurred.
In three of the cartridges, accumulation of solid particulate matter caused damage to the carbon sealing surfaces.
Radial slots in the carbon face, which had allowed abrasive solids to enter the gap and cause excessive wear, were eliminated from the carbon face with additional revisions made to maintain the required lubrication in the sealing gap.
The secondary seal configuration was changed to reduce the friction between the secondary seal and the balance sleeve.
D.
Increased Engineering Involvement and Root Cause Analysis Waterford 3 Since the November 1988 occurrence, increased engineering involvement and root cause analyses have provided technical basis for identifying areas beyond the seal cartridges for enhancing seal performance.
For example, during installation of the present seal cartridges, considerable effort was expended to address elements beyond the seal cartridges that may affect seal performance, such as RCP mis-alignment, axial play, and shaft eccentricities.
2-6
LICENSE.083 1
NRC Federal Register Question 1.3 liow of ten are seals being routinely replaced (e.g., every refueling)7
Response
ANO-1 & 2. Waterford 3, on average. RCP sea).s at the Entergy Operations' PVR units have been routinely replaced every refueling outage.
There are, however, several instances where seals at Atio-1&2 have been in place longer than one fuel cycle.
The improved performance history of the RCP seals at ANO and Waterford suggests longer replacement cycle times.
The current goal for ANO and Waterford is to replace the RCP seals every other refueling outage.
l 1
l l
l l
l l
2-7
..=.
LICENSE,082 NRC Federal Register Question 2 The NRC staff is interested in obtaining any available data regarding degraded cooling or loss of cooling to the seals to support assertions that seals can survive long periods of time (i.e., hours) without cooling.
Response
ANO-1 ANO-1 bc:: not experienced a prolonged loss of cooling to the RCP seals. Normal plant operation at ANO-1 does expose RCP seals to higher than normal temperatures during certain plant operations.
During heatups and cooldowns, operation of all four RCPs is prohibited below 500'F RCS temperature due to core uplif t concerns.
This limits the effectiveness of seal cooling for the idle RCP to only that cooling provided by the thermal barrier heat exchanger.
Lower seal temperatures of approximately 250 F and CB0 temperatures of approximately 165 F for pt:riods of hours are not uncommon.
Normal seal perfermance (staging pressures and CB0 flow) is observed during pump idle periods and subsequent pump starts.
These parameters and performance have been observed on both new seals and seals which have been in service.
ANO-2 ANO-2 has not experienced a prolonged loss of cooling to the RCP seals.
Waterford 3 In February 1985, Waterford 3 experienced a loss of CCW to 3 RCPs.
After 45 minutes, one pump leaked at approximately 3 gpm.
During the event, the vapor seal failed as a result of CB0 isolation, but the lower - and middle seals maintained their function. The plant was cooled down, and the seals maintained their functional integrity throughout the event.
2-8
LICENSE.082 NRC Federal Register Question 3.1 i
Are there procedures currently in place that are intended to prevent seal leaks from becoming small-break LOCAs during both normal plant operation and loss-of-seal-cooling events such as station blackout?
Are the required operator actions (e.g.,
isolating leakoff lines) the same for normal plant operation and loss-of-seal-cooling events?
Response
ANO-1 Yes - The ANO-1 Reac or Coolant Pump Emergency Procedure provides direction to operators to manually trip the reactor and stop all RCPs if ICW flow is lost and cannot be restored within a short amount of time.
The CB0 is isolated to prevent further seal degradation only if ICW flow is lost a_nd CB0 temperature n
exceeds 200 F and the RCP is secured.
The station blackout procedure provides direction to tho operators to isolate CB0 flow from the RCPs if RCS inventory control problems are experienced. This action is accomplished to reduce RCS inventory losses and not to protect the RCP seal, CB0 is not isolated for any events other than those discussed above.
ANO-2 Yes - The ANO-2 Reactor Coolant Pump Emergency Procedure ptavides direction to operators to manually trip the reactor and stop all RCPs if CCW flow is lost and cannot be restored within a short amount of time.
The controlled bleedoff (CBO) is isolated to prevent further seal degradation if CCW flow is lost and the RCP is secured.
Waterford 3 Yes - Procedures are in place which are Intended to prevent seal leaks frcm becoming small break LOCAs during both normal plant operation and loss of seal cooling events.
O f f-No'rmal Procedures provide direction to operators to manually trip the reactor and begin an RCS cooldown if more than one seal stage fails on an RCP.
Of f-Normal Procedures provide direction to onerators to manually trip the reactor and stop all RCPs if CCW flow is lost and cannot be restored within a short period of tima. Off-Normal Procedures also preclude the restart of any RCP if CCW has been lost for 10 minutes or longer.
s 2-9
LICENSE.083 1
f 5
- t}jf Federal Register Question 3.2 Has the RCP instrumentation been evaluated to determine whether operators have sufficient information to implement the procedures?
Response
ANO-1&2 The ANO seal inst rumentation was evaluated for adequacy during the design change which added RCP seal parameter recorders for both units, The ANO-1&2 Reactor Coolant Pump Emergency Procedures are written based on these RCP seal parameters. The available parameters are summarized below:
Controlled bleedoff flow e
e Controlled bleedoff temperature Middle cavity pressure e
Upper cavity pressure e
Vapor seal cavity pressure (ANO-2 only) e Lower seal temperature e
RCP pump / motor ICW/CCW (ANO-1/ANO-2) flow alarm RCP pump vibration o
RCP pump shaft axial position e
Each RCP seal is provided with a two speed strip chart recorder monitoring seal temperatures, flow, and cavity pressures.
This feature provides improved operator capability in responding to and analyzing RCP seal conditions as well as providing the capability for assisting in analysis of root cause should seal problems arise.
These points are also monitored by the plant computer and printed daily on an hourly log.
Pump vibration and axial shaft position is monitored and trended on a dynamic data manager.
Control Room and computer alarms are provided for CB0 Flow High/ Low, CB0 Temperature High. RCP High Vibration, ICW (CCW for ANO-2 only) Discharge from RCP Low Flow, and ICW (CCW for ANO-2) Loop II Low Flow.
,Waterford The Waterford 3 instrumentation has been evaluated.
This evaluation has concluded that operators have sufficient information to implement the off -
Normal Procedures.
l l
The following parameters are available to operators for providing indication of RCP seal failure or loss of CCW:
RCP Seal Failure RCP Controlled Bleedof f temperature (rising)
RCP Seal cavities pressure (rising)
Seal Water Cooler CCW outlet temperature (rising)
RCP Controlled Bleedoff flow (rising) 2 - 10
LICENS2.GG8 Loss of CCW Seal Water Cooler CCW outlet temperature (rising)
Seal Water Cooler CCW outlet temperature (dropping)
RCP Controlled Bleedoff temperature (rising)
RCP stat-or. winding temperature (rising)
RCP bearing temperature (rising)
RCP Seal Cooler CCW valve closed The following alarms are available to operators to warn of RCP failure or loss of CCW:
RCP Failure RCP Controlled Bleedoff Temperature Hi RCP Controlled Bleedoff Header Pressure Hi RCP Controlled Bleedoff Header Pressure H1-Hi RCP Seal Cooler CCW Tc.nperature Hi RCP CCW Flow Lost RCP Seal Cooler CCW Valve Closed RCP Controlled Bleedoff Flow High on PMC RCP Seal Cavity Pressure High on PMC
~
Loss of CCW RCP CCW Flow Lo 3CP CCW Pressure Lo RCP Seal Cooler CCW Temperature Hi RCP CCW Flow Lost RCP Controlled Bleodof f Temperature Hi RCP Seal Cooler CCW Pressure Hi RCP 1A, 1B, 2A, 2B Seal Cooler CCW Valve Closed 2 - 11
LICENSE.082 NRC Federal Register Question 3.3 l
How is RCP seal vendor information used in establishing operation and maintenance practices for the RCP seals?
Response
ANO-1 & 2, Waterford 3 RCP seal vendor technical manuals are used as the basis for technical content of operation and maintenance procedures and training lesson plans for RCPs.
Additionally, seal vendor and independent contractor reviews of RCP seal related procedures have been performed.
The instrumentation installed and listed in the response to Question 3.2 was based on seal vendor input.
2 - 12
LICEESE.083 NRC Federal Register Question 3.4 In some cases, industry practice allows continued plant operation with the RCP seal when first or second stages have failed.
Do you limit this practice?
If so, what are the limiting conditions?
Response
ANO-1 & 2, ANO permits continued plant operation with one seal s ts.ge degraded.
This practice is limited as follows:
A.
The reactor and af fected RCP are tripped if any of the folicwing exist; (1) Three stages have degraded on any RCP, (2) vapor seal pressure is greater than or equal to 1500 psia, (3) Indications of no controlled bicedoff, or (4) controlled bleedoff temperature exceeds 200'F (not attributable to a. ICV (ANO-1) or CCW (ANO-2) failure).
B.
A controlled plant shutdown is commenced if any of the following exist; (1) two stages have degraded on any RCP, or (2) Controlled bleedoff temperature exceeds 165 F (not attributable to a ICV (ANO-1) or - CCW (ANO-2) failure).
Each of the three (ANO-1) or four (ANO-2) individual scaling s, ages in designed to withstand full _RCS pressure indefinitely with the RCP idle and for a limited period with the pump running at a nominal 1200 RPM (ANO-1) or 900 RPM (ANO-2).
Waterford 3 Procedures at Waterford 3 allow continued plant operation if the first stage has failed.- Off-Normal Procedures provide direction to operators to manually trip the reactor and begin an RCS cooldown if more than one seal stage falls on an RCP.
As a practical matter, Waterford 3 has not operated for any extended period of time with a failed seal stage and it is expected that Waterford 3 type which could lead to
[
would. shut down well before any degradation of a l
failure of the second stage, since CCW flow to the seal heat exchanger is isolated at 155 F.
l
\\
\\
2 - 13
c 1.ICENSE.082 NRC Federal Register Question 3.5 What additional quality assurance and procedural measures can be taken regarding RCP seals to improve S-fety?
Response
ANO-1&2, Waterford 3 As discussed in the response to Question 1.2, significant quality improvement ef forts have been made and corresponding improvements have been observed in the seal performance at ANO and Waterford.
Since a RCP seal is a complex high-precision assembly, the procurement, fabrication, assembly, transport, and installation of RCP seal assemblies are carefully controlled to assure high reliability and performance.
Experience at ANO and Waterford with Borg Warner Industrial Products (BW/IP)
Pump Division, of which BJ is a subsidiary, has shown that they have a comprehensive quality assurance p rog re.m.
They have a formal and controlled quality manual which specifies and defines the requirements of the program.
Included in the Quality Manual is a statement of policy which provides a list of standards to which the Quality Manual complies.
Other BW/IP documents which support the Quality Program are described as follows:
IP-1500 General Inspection Requirements for Nuclear Spare Parts. This document contains detailed requirements for the inspection of nuclear seal parts. Under these requirements, in addition to 100% dimensional inspection, specified dimensions are recorded on certain parts considered critical to the safe operation of the seals.
IP-1500 also includes guidelines for inspection of surface irregularities on seal parts and inspection criteria for the measurement of flat lapped surfaces.
GS-1518 Procedure for Material Procurement and Manufacture of Class 1 Nuclear Pumps and Seal Parts.
This document provides a compilation of BW/IP Pump Divisibn procedures commonly used in the procurement and manuf acture for Class 1 nuclear pumps and mechanical seals.
All procedures meet the requirements of the appropriate portions of the ASME Code and ASTM Stantards. GW-1518 includes typical manufacturing and inspection procedures, doccaentation furnished with the completed order.-and a typical listing of materials of const m2ction, and alternate materials from which parts can be manufactured.
4 518 also m3terials stipulates that vendors used by BW/IP Pump Division to supply
- e the requirements of NCA-3800 of the ASME Nuclear Code, isection III, must meet as indicated in the applicable manufacturing procedure.
The BJ Quality Assurance program is passed on to BJ's sub-suppliers and these sub-suppliers are audited to meet the requirements of Appendix B.
The PW/IP QA requirements discussed above, in combination with the site procedures dealing with RCP installation and maintenance, make new, formal NRC requirements unnecessary to achieve the desired risk reductions associated with RCP seal failures.
2 - 14
LICENSE.083 7
o NRC Federal Register Question 4.1 l
Is the staff's model, or other models, adequate to predict RCP beal leakage i
(i.e.,
modes of seal failure, time-dependent f ailure probabiltty, and leakage estimates) and handle the uncertainties in the models? Do the models correlate to actual plant or test datat
Response
ANO-1&2, Waterford 3 The NRC staff's model is for the Westinghouse hydrostatic thick-film RCP seal and is therefore not applicable to hydrodynamic thin-film RCP seals like the SU caal and N-9000 seal.
In addition, the AECL model of a hydrodynamic seal is not considered to be adequate to accurately quantify expected RCP seal behavior.
The AECL model is substantially different in both materials and physical configuration compared to an actual hydrodynamic RCP seal.
The research report utilir.ing this model did not consider the importance of many of the details that, f rom our experience with BJ, must be included in the model.
Instead, more general assumptions were employed.
Therefore, the conclusion reached that low balance ratio hydrodynamic seals (implicating the N-9000 scal) would become " unstable" when subjected to two-phase flow across the faces is not justified.
The prediction that carbon face erosion would occur is also considered to be incorrect, since the face profile and face leakage on an actual seal are much dif ferent than the AEG, model predicted.
As previously mentioned, the equipment specification for the N-9000 seal cartridge included many analytical requirements and reflected input from participants in the Pump Seal Improvement Program.
These analytical requirements included stress analysis and clearance requirements over the entire range of specified opers. ting conditions (including high temperature station blackout conditions) for each piece part of the seal, To comply with the analytical requirements, BJ established a three-dimensional model of the N-9000 hydrodynamic seal faces that accurately predicted seal de flec,tions, leakage rates, fluid flim profile, and the effects of temperature and pressure.
The correlation between actual and predicted behavior at both steady-state and transient conditions has been excellent.
This correlation has produced a high level of confidence in the capabilities of the model and the seal.
The BJ N-9000 RCP seal model did not include high-temperature two phase flow across - the seal faces.
However, BJ has determined that the seal face model does indicate that the ring moments acting on the carbon and carbide face rings will tend to keep the faces divergent, while the wavint.ss in the stationary face (produced by slow effects) creates a hydrodynamic lifting effect to produce a fluid film and some lubricating leakage.
In the static condition, such as station blackout, these ring moments produce a profile that tends to pinch of f the' f aces at the outer edge.
This has been confirmed by the near zero leakage experienced through the N-9000 seal faces during actual static conditions.
This interpretation of the seal model analysis and ictual seal behavior lead BJ to the conclusion that two-phase instability was very unlikely.
This conclusion was confirmed by the results of an actual loss-of-cooling test performed by BJ, daring which no indication of instability was observed and no two-phase flow erosion of the f aces was observed.
2 - 15
1,1 CENSE,082 Aside from the seal faces, the effects on seal cavity pressures due to choked flow in the top seal staging coil were estimated by BWNS prior to the loss-of-cooling test.
Other leakage paths (aside from seal face leakage or instability) were also considered.
One particular o-ring was predicted to be the most likely to fail, owing to the clearance at elevated temperature and the relatively small o-ring cross-section.
The leakage rate associated with failure of this o-ring was computed by BJ prior to the test.
At the end of the loss of cooling test, this small o-ring failed in the top seal stage, resulting in a maximum leakage rate of 1.7 gpm.
Although the failure was later determined to have been precipitated by a manufacturing process defect (since I
corrected), the resulting leakage was as predicted.
With respect to time-related degradation and failure modes, it has been obvious in the N-9000 program that the elastormrs will tend to be the first components to degrade at elevated temperature.
However, the clearances available in the N-9000 seal, even if elastomers fail, are such that the leak rates would be well within the capabilities of the makeup system (and within the capabilities for SB0 coping demonstrated by ANO-1 and Waterford 3).
The Icakage rates are also orders of magnitude less than that predicted in the time dependent Westinghouse seal failure model.
1 2 - 16 l
~_.. _.
1.ICENSE.082
, o.
NRC Federal Register Question 4.2 of particular interest to the staff are siternatives to the probabilistic RCP seal leakage modes developed for Vestinghouse seals and alternative models for other seal designs (i.e.,
for seals by Byron-Jackson, Bia.gham Int.,
or Combustion Engineering /KSB) to predict seal leakage during loss-of-all-seal-cooling events.
Can you provide information regstding any alternative models?
Response
ANO-1, Waterford 3, As mentioned in response to question 4.1 and 5, the actual performance of the N-9000 RCP seal under various conditions was predicted analytically and confirned by test.
The primary purpose of the analytic predictions was to ptovide assurance thnt stringent design requirements would be met by the proposed design.
The design requirements used by BJ for the N-9000 seal include consideration of expected performance during extended loss of cooling events.
Detailed, conservative thermal-hydraulic analysis was performed for the N-9000 under loss of all cooling conditions, the results of which were used as input to a finite element structural analysis.
Detailed analysis was performed on all critical clearances to meet specific requirements throughout the range of conditions from normal operating temperatures to SB0 conditions at the worst case tolerance stack-up.
Both thermal and mechanical effects (including pressure-induced mechanical effects) were incorporated in the analysis of each piece part la the entire N-9000 seal assembly.
Special attention was focused on maintaining clearances t.ha t would allow proper tracking of axial shaft motions at elevated temperatures, since RCS pressure changes affect the thrust acting on the pump shaft, resulting in spring flexure of the motor bearing housing, causing axial shaft movement.
As previously mentioned, elastomer extrusion was another factor that was consid,ered in the N-9000 design.
Some of the critical clearances in the N-9000 increase in size at elevated temperature due te dif ferential thermal growth.
These areas were carefully evaluated with respect to possible extrusion of the ethyler.e propylene elastomers.
Ethylene propylene was chosen as an elastomer material because it offers higher temperature capabilities than other materials available for this application. Where critical interfaces were involved, seal materialr wexe chosen that provide appropriate mechanical properties while minimizing differential thermal expansion coefficients between various seal parts.
Therefore, as discussed above and in response to questions 4.1 and 5,
the rigorous modeling, design, and subsequent ccnfirmatory testing indicate that the N-9000 RCP seal will experience minimal leakage during extended loss of seal cooling conditions.
Since gross seal leakage is not expected under these conditions, we consider further resources put forth on additional analytical modeling to be of questionable benefit.
l i
2 - 17
1,1 CENSE,082 t
I - o.
NRC Federal Register Question S In exploring alternatives to providing additional seal cooling, one approach might be to test the existing seah, to demonstrate conclusively that they will not leak excessively if not cooled for extended,ceticis of time, even though such conditions exceed the seal design basis and possibly th's conditic.ns of the warranty.
If testing was an option to demonstrate acceptable s'eal perfarmance under loss of cooling conditions, what conservative conditions shculd bc Imposed on the RCP seal for the test program (e.g.,
length of time, maximum wear on seal, number of tests)?
Response
ANO-1, Wat e rford 3 Testing has already been performed on a BJ N-9000 RCP seal cartridge.
This tenting was performed subsequent to detailed assessments of the ability of the design to meet specific of f-normal performance requirements and was performed in order to verify the capability of the N-0000 seal cartridge to withstand a prolonged, approximately 8 hour9.259259e-5 days <br />0.00222 hours <br />1.322751e-5 weeks <br />3.044e-6 months <br /> loss of cooling event.
BVNS performed thermal-hydraulic analysis of the RCS response during station blackout conditions to conservatively predict the specified transient to which the-RCP seal wnuld be subjected.
In addition, axial shaft movementt were included in the test to follow predicted RCS pressure changes.
RCS pressure changes af fect the thrust acting on the pump shaf t, resulting in spring flexure of the motor bearing housing causing axial shaf t movement.
The test was quite rigorous and the specification of test conditions carefully considered the worst case transient as it relates to imposing the most stress on the seals.
In addition, it was decided very early in the planning that the test would have to be on a full scale actual seal cartridge and that the cartridge would have to be " aged" by subjecting it to a 5000 hour0.0579 days <br />1.389 hours <br />0.00827 weeks <br />0.0019 months <br /> seal qualification test prior to the loss of cooling test.
As far as seal wear is concerned, from the experience of BJ, the only suitable method, identified to
- age"- the seal cartridge has been to actually accumulate run-time hours.
In order to facilitate a loss of cooling test on a non-contaminated piece of equipment, the run-time must be accumulated in a test.
facility.
In addition, simulated plant transients such as startup, shutdowns, and reactor trips are needed to properly exercise the seal f aces and elastomers to achieve normal wear.
It has been Entergy Operations' experience that the expense of accumulating appropriate run-time wear is very high.
In response to the question of what number of tests should be performed, experience indicates that this should be determined by a review of individual ceal design characteristics and the rigor of design analysis or modeling performed at tne appilcable conditions.
l 2 - 18
LICENSE.083 a.
For the N-9000 seal. the extensiveness and accuracy of analysis and the fact that the seal was designed from the beginning to satisfy specific loss of cooling requirements makes a single test of a full size actual production seal cartridge appropriate and ruperior to the multiple " scoping" tests performed by NRC contractors to date.
Additional testing of the N-9000 seal would be very costly and would not, provide any significant additional confidence.
In short, it is believed that the specific combination of analysis and empirical verificaticn test results that were conducted for the N-9000 seal (with respect to the loss of cooling issue) was the right balance to achieve the maximum cost-beneficial confidence of performance for a real loss of cooling event.
d 2 - 19
/
LICENSE.082 i
e NRC I*ederal Register Question 6 If, after consideration of public comments, the NRC decides that additional RCP seal requirements are necessary, what method of irnpos it ion should be used (e.g., by rulemaking, orders, or generic letter)?
Response
ANO-1 g Waterford 3 The record supporting the NRC Staf f's proposed resolution reflects that GI-23 is highly plant-specific and that not all PWRs are equally vulnerable to RCP seal failure.
In its evaluatica of RCP seal failure, the Staff acknowledged fundamental ditferences in RCP seal designs and instrumentation, and various modes of seal operation. The resulting analysis reflected the realization that
" generic evaluations cannot be performed because of the vast differences in system designs for va;lous plant vendors or even among plants representing the same vendor but different vintages." Hawever, the proposed resolution is based largely on consideration of Ve tinghouse seals installed in Westinghouse PVR units, comprising $7 of the affected PWRs.
As concluded by the ACRS and others, it is not clear that other seal designs have similar characteristics.
The GI-23 concerns are highly dependent on the design details of the ICW or CCW and RCP seal auxiliary systems, the sensitivity of varying RCP seal designs to loss of CCW and RCP seal auxiliary systems, and on the individual plant's vulnerability to SBO.
Thus, the analysis supporting the proposed resolution should include adequate consideration of these matters, as well ac of plant-specific capabilities such as those implemented by Entergy Operations and otner utilities in response to the SB0 rule.
Due to these large dif ferences and in view of the recent and rapid improvement in seal design and performance, it would be more appropriate for the NRC to focusing only on those plants at which proceed on a case-by-case basis some action may be 6ppropriate to plant-specific evaluations demonstrate that reduce the seal failure-induced risk.
The Commission should exercise its discrelion to use means other than generic rulemaking to accomplish this objective.
Recent judicial interpretations highlight the discretion of federal agencies to choose case-by-case solutions where more generic actions, such as rulemaking, are not appropriate.
In any case, the NRC must satisfy the standards of the backfitting rule.
Entergy Operations believes that new requirements related to this issue should not be imposed on licensees unless they are demonstrated to be necessary and reasonable and t na t. the issuance of orders to individual plants is a viable means of addressing the GI-23 concerns at those plants which are shown to be vulnerable, i
t 2 - 20