ML20151L059
| ML20151L059 | |
| Person / Time | |
|---|---|
| Issue date: | 07/16/1985 |
| From: | Jordan E NRC OFFICE OF INSPECTION & ENFORCEMENT (IE) |
| To: | Taylor J NRC OFFICE OF INSPECTION & ENFORCEMENT (IE) |
| Shared Package | |
| ML19284F128 | List: |
| References | |
| NUDOCS 8507230545 | |
| Download: ML20151L059 (12) | |
Text
__
lj
..,CG Cf 34'c,.
'f {. ;*'4.. c[i UNITED STATES i
NUCLSAR REGULATORY COMMISSION fc
. J* r.
I WA$r4tNGTON. D. C. 20$55
- . * ;5.
i '.
- *f.
.y u
- ewe *.
JUL 10190I r
i.
- 3,
- z
-- t :
_._r
~
-~
t l,,
J' 7 '
l'EM3RkNCbf FOR:
James M. Tay' lor, Director.
[;
~
(
j Office of Inspection and Enforcement;,
.~
I FROM:
-Edward L. Jordan, Director
~ Division of Emergency Preparedness and Engineering Response I
Office of Inspection and Enforcement '
i S*J5JE'CT:
STEAM BINDING IN AUXILI'ARY FEEDWATER SYSTEMS '
I L.
Te.porary Instruction 2515)67 directed'the regional offices to conduct surveys
~
of licensee responses to two identified safety issue's: Steam Binding in Auxil-h iary Feed ater Systems,and Mispositioned Control Rods.
These issues were Y
E::ressed more than a year ago by IE Information Notices (IN) and by Institute cf Nuclear Power (INDO) Significant Event Reports (SERs) and INPO Significant 0:erati,ng Experience Reports (50ERs).
The 50ERs contained specific recommended a:tions to alleviate the safety concern.
~ T*,e primary purpose of cur survey is to determine the actions that licensees a e tat:ing in response to the two selected safety issues.
The secondary cur;; css is to determine the actions that licensees are taking in response to
- s rt:er.er.dations in INPO's 50ERs.
a d'
I For.tne first issue, Steam Binding in Auxiliary Feedwater Systems, the respon ses rave been~ received and an initial review has been performed.
The results
~
are provided in the enclosure and can be summarized as follows:
g p
'1)
No immediate safety problems were found; that is, no hot pipes or disabled j!.
pumos were found by the inspectors.
I). The INDO recommendetion of primary safety importan e concerned monitoring auxiliary feedwar.er system temperature each shif t.
Of the 55 units surveyed, 39 had such monitoring to detect any back leakage of het water ir.to the system.
At most of the units, this is done by touching the pipe.
Setenteen ttnits had some degree of justification for no. monitoring, su:h as a normally closed gicbe valve in addition to check valves in the
+
system.
Two units did not have what we considered to be reas.pnable justification for the lack of monitoring.
These two units are THI-I and Trojan.
Based on follovup discussions with Regions I and V, we understand
.t that both facilities have now begun monitoring.
l
Contact:
M,.'S M h er, IE 492-4 l
/p l
Qcf46 y[
C'
James M. Taylor ;)
The INP0 recommendations next in safety importance concerned procedures and training for detecting and correcting back leakage.
Of the 58 units surveyed, 25 had both procedures and training, 20 lacked either procedures or training, and 13 had neit'her.
Procedures are cuite important to ensure that monitoring is consistently performed, both now and in the future..
For example, simply telling the auxiliary operators to check the pipe temperatures each shift, without explaining why or providing any guidance on recovery, will not provide a lasting solution of reasonable quality.
Therefore, we will propose an IE bulletin to request that all licensees
- develop and implement procedures.
(Training and awareness will follow as a matter of course from implementation of procedures.) We plan to meet with INPO to review results of their review of licensee actions to the 50ER and discuss our findings and planned actions.
4)
Other INPO recommendations of lesser safety significance had a lesser degree of implementation.
We do not plan any short-term action regarding these other recommendations.
E)
In tne longer run, all of the responses will be reviewed by NRR in tne precess of resolving Generic Issue 93, " Steam Binding of Auxiliary Feedwater Pumps."
In sum 5ary, for the steam binding issue, all but two units had alleviated the safety problem to some degree, but many licensees lacked the procedures er training needed for a lasting solution of reasonable quality....
Eescanses for the second issue, Mispositioned Control Rods, are expected later
- nis co-t5.
We will inform ycu of the results when a preliminary review has teen c;rpleted.
- 4kf l-5 1
Edward L. Jordan, Directer Division of Emergency Preparedness and Engineering Response Office of Inspection and Enforcerent Enciosure:
Summary of the Responses to TI 2515/67 Felated to Steam Einding of Auxiliary Feedwater CI S RIEi.' TIC *t l
- 5 DEPER r/f EGCB r/f Mleoner r/f RHVollmer SASchwartz EL5aer AWDromerick JGPartlow
~
Erdrices JWermill, NRR WLanning, AE03 CMolchan, NRR KSeyfrit, AEOD WMilstead, NRR I
- Zech DAllison Regional Prcjects Division Directors l
Mdegner VHodge l
'See previous concurr~ences c
. organ 7//J/E5
';EPER:IE "DEPER:IE
- PSB:IE
- DEPER:IE
- ASS:NRR *DEDER:IE M<.egner DAllison DGable RLBaer OParr SASchwartz 7/ /85 7/ /85 7/ /85 7/ /85 7/ /85 7! /85 L
g e
Summary of the Responses 1to TI 2515/67 Pelated to Steam Binding of Auxiliary Feedwater Tecaorary Instruction 2515/67 directed the rigional offices to conduct surveys of licensee responses to two icentified safety issues, steam binding in L
auxiliary feedwater rystems, and mispositioned control rods.
For the first issui, steam binding in auxiliary feedwater systems, the responses have been re:eived and an initi>l review has been performed.
Tabulation of those responses are given in the attachments.
No immed'iate safety problems were found:
that is, no hot' pipes ~or disabled' pumps were found by the inspectors.
Of the 58 units surveyed. 39 units monitor AFW piping temperature at least f
once every 12 hours1.388889e-4 days <br />0.00333 hours <br />1.984127e-5 weeks <br />4.566e-6 months <br /> (INPO recommendation 1).
The principal method used to conitor the AFW piping temperature-is touch.
Only four units have control room ' readout.
a w
r i
Of the remainder, 17 units had some degree of justification.
These' units are listed below and their justifications are summarized.
Calvert Cliffs 1 & 2 - mdnitoring is accomplished weekly.
No recent e
history of steam binding.
Appropriate check valve
~
design.
p In'::ian Point 3 - closed MOV in discharge line.
I I;illstone 2 - normally closed regulating (gicbe) valve.
Salem 1&2-closedisolationvalvh.
Ya.kee Rowe - complete separation of AFW trains makes multiple failures 24 unlikely.
Licenseealsoclaimscreditforwaten-stancers'i ability to detect but resident disagrees.
Crystal P.iver 3 - has ultrasonic flow detectors that also detect back-
."i leakage.
(Monitoring has begun since the survey.)
Oconee 1, 2, & 3 - a closed gate valve, r.o history of steam binding, and a long uninsulated discharge line.
I Turkey Point 3 & 4 - closed globe valves and self-venting pumps.
L Palisades - closed flow control (globe) valves.
f l
Davis-Eesse 1 - r.ormally closed MOVs.
Efron 1 pump casing vented daily.
Licensee also claims credit for water nammer prevention instrumentation'but resident disagrees.
Waterford 3 - closed isolation valves.
i h
i
Tac units that cid not monitor had little justification.
They are:
i Tl11 1 - no previous problems.
Check for leakage during plant startup.
The licensee planned to provide general awareness training.
(Since the survey, the licensee has ag ad to monitor once per shift.)
Tro'jan - considering monitoring.
(Monitoring has begun since the survey.)
Seventeen units monitor the temperature of AFW piping after each operation in addition to'the routine checks'(INPO recommendation 2).
Note that for f
plants already monitoring once a shift, this does not carry a great deal cf safety significance.
Taenty-five units had procedural guidance and training on identifying and.
correcting backleakage (INP0 recommendation 4), 13 had'neither, and the remaining 20 had less than full implementation of the recommendation.
Details are provided in Table,1.
F ecedural corrective actions include:
vent and flush, close isolation selves and slowly reopen, use AFW booster pump to cool, and stroke LEV to reseat check valve.
!;ine units leak test the valves or verify that they shut (INPO recommendation 5).
In-service testing (IST) required by ASME Section XI Part..IWV depencs uson the licensee's classification of the valve.
It usually entails oper-asility testing only, when any testing is required.
The answers to question l
7c left doubt about the respondent's definition of " inspection," but the
.3 'yes" answers probacly refer to the ASME in-service inspection.
4I.
1 he did not determine now many units had reviewed their check valve design l
fer suitability; that is, ability to seat with low AP (INPO recommendation 3).
ke did ncte that.13 units determined that procedural changes were needed to assure check valve seating and implemented them.
9 it did not determine whether or not unnecessary thermal insulation had been re.yeed (INPO reccmmendation 6).
Attach.Te nts:
1.
Tasle 1 - Tabular Summary 2.
Table 2 - Follow Up Commitments lientioned in Response to TI 2515/67 O
-_y,
-._,,_.--__,-._-...._,.-,,-..,._,,.,..-.e_._
_,,._~,..--.._,,.
%_.~.._..y_._.
L-
- :TAZLE 1 t
TABULAR
SUMMARY
y
. LAtJT _ _ _ _,
ACTION PROCEDURES FOR s TRAINING FOR ' *
~
~-
(JUSTIFY /
IDENTIFICATION /,
IDENTIFICATION /
E 2 ~ %n. _, _ _.... _. MONITOR) - CORRECTION _, ;,
CORRECTION JL*STIFY _
~*.
2,;. :. ;;-
~~
EALEM i
.f.
JUSTIFY BOTH 3
77 : ;
BOTH
/
5ALEM 2 1T ~_ : _
JUSTIFY BOTH -
p;, f BOTH
~
~:
~
JUSTIF'Y -
BOTH BOTH :
/
~ ~ ~
OCONEE 1 t
0 OrJEE 2 JUSTIFY BOTH BOTH ;
t
'ICONEE.3 JUSTIFY BOTH c-BOTH - :
CALVERT CLIFFS 1 JUSTIFY BOTH NEITHER.
.. L.
CALVERT CLIFFS 2 JUSTIFY
, BOTH NEITHER
~
YANi'EE-ROWE JUSTIFY CORRECT CORRECT EYRON 1 JUSTIFY CORRECT NEITHER.
k
' ~
CRYSTAL RIVER 3:.
JUSTIFY NEITHER BOTH
~
INDIAN POINT 3 JUSTIF NEITHER NEITHE't ~
b MILLETONE 2 JUSTIFY NEITHER NEITHER
~' '
TURIEY POINT 3 JUSTIFY NEITHER NEITHER TUR.MEY POINT 4 JUSTIFY NEITHER NEITHER e
~ :ALIEADES JUSTIFY.
NEITHER NEITHER
- AVIE-BEEEE.
JUSTIFY NEITHER NEITHER WATER OFi,D 3 l
^
JUSTIFY NEITHER NEITHER -
l
- ~ MONITCR
~
'EEAVER VALLEY 1 MONITOR BOTH BOTH It:OIAN POINT 2 MONITOR BOTH BOTH CATAWEA i MONITOR BOTH BOTH E. A:. LEY 1 MONITOR BOTH BOTH
.A:. LEY 2 MONITOR BOTH BOTH I
I M 33 IRE,1 MONITOR BOTH BOTH MbGUIRE 2 MONITOR BOTH BOTH RCEINSON 2 MONITOR BOTH BOTH H
EE UOYAH 1 MONITOR BOTH BOTH EEOUOYAH 2
. MONITOR BOTH BOTH EU:.RY 1 MONITOR BOTH EDTH EURRY. 2 MONITOR BOTH BOTH l
!!OtJ 1 MONITOR BOTH BOTH
[
IION 2 MONITOR BOTH BOTH
.7 CALHOUN MONITOR BOTH BOTH ANO 1 MONITOR BOTH BOTH ANO 2 MONITOR BOTH BOTH~
SAN ONOFRE 2 MONITOR BOTH BOTH SAN CNOFRE 3 MONITOR BOTH BOTH FALD VERDE 1 MONITOR BOTH BOTH i
h I
t
TABLE 1 TABULAR
SUMMARY
PLANT ACTION PROCEDURES FOR TRAININ3 FOR (JUSTIFY /
IDENTIFICATION /
IDENTIFICATIOtu MONITOR)
CORRECTION CORRECTION
-:ADOAM NECK MONITOR BOTH t4EITHER O C COOK 1 MONITOR BOTH
^
NEITHER O C COOK 2 MONITOR BOTH NEITHER T40RTH ANNA 1 MONITOst.
IDENTIFY IDENTIFY NORTH ANNA 2 MONITOR IDENTIFY IDENTIFY SAN ONOFRE 1 MONITOR IDENTIFY IDENTIFY ET LUCIE 1 MONITOR NEITHER BOTH ST LUCIE 2 MONITOR NEITHER BOTH EUMMER' MONITOR NEITHER BOTH OIABLO CANYON 1 MONITOR NEITHER BOTH
- RAIRIE ISLAND 1
- MONITOR, NEITHER IDENTIFY
~~
FRAIRIE ISLAND 2 MONITOR NEITHER IDENTIFY KEWAUNEE MONITOR NEITHER IDENTIFY
- OItJT EEACH 1 MONITOR NEITHER IDENTIFY FC. INT BEACH 2 MONITOR NEITHER IDENTIFY 3I!JNA MONITOR NEITHER NEITHER l;AINE YANi;EE MONITOR NEITHER NEITHER
?.'OLF CRsEK 1 MONITOR NEITHER NEITHER RANCHO SECO MOtJITOR eNEITHER NEITHER 14EITHER TP I 1 f4EITHER NEITHER NEITHER TF.00AN NEITHER NEITHER f4EITHER
=$
1 l
m O
w,-
.- w,-
~'
l olit.E **
FUI.LilN til' CtllIlll itlCNI G lil.N r lONI'D IN llLUI'llNGE TO TI 2515/67 i t. mn t.imt. larui slent Imu rarpor tent cusupl ute In110winsj starvey n
NERT CLIFFS 1 EU1 Alit.lSil LI:Al; CllCCI:G IN FilillitE.
. VERT CLIFFS 2 E!il AI)LISil I.Enl: I:llECl;S IN l'lJillitE.
IUN 1 ADDI T IONAL IRLVILN UF Al'l Isill'l(I A~l ENEGU UF NOTER IIAI1MLit ING'lltUtlENIS FUR S1H DINDING
'SI Al. ItIVER 3 ItEPLACE IIL'IRASUNll: Fl.11N !)E r. DEGIN MUNI1 Ult!NG.
- INSTALL ALARMS.
ti.EY l'U I N F 3 UEGIN Mlltil'IllilING DY 6/1/115. TRAINING AND PROCEDlJRAL (ilJ1 DANCE DY 6/1/05.
ll.EY POINT 4 DEGIN tlIJNIlllRING DY 6/1/115. train. 23 AND PROCEI) URAL GUIDANCE DY 6/1/115.
.IGADEU ACT IllN tlN LEnl< CIIECIC liY 9/ I /05.
bVER VALLEY I CUNSIDEltING INSTALLING 1EMI'ERAll1RE INDICATORS.
sulflE I CONIRUL Rilull COMPUlER PT FOR lEMP REAliUUT.
tilY 1 I)ESluN CilANGE cot 1CERNING 1EllP READOUT IN CONTROL ROON.
(ItY 2 UESIGN CllANGE CONCERNING TEMP READOUT IN CONTROL ROON.
CAlllOUN PERMANENT 1El4P MUNITORG TU DE INSTALLED FALL 05.
IMElt INSTALL CUN1RUL ItOUM ANNUNCIATUR.
l hDLO CANYON 1 DESIGN CllANGE TO INSTALL CONTROL ROOH ANNUNCIATOR.
kIRIE ISLAND 1 EVAllIATE LEAL; TEST, INSPECTION.
l blRIE ISLAND 2 EVALUAIE LEAIC TEUT, I NSPlIC TION.
t NT DEACil 1 REPLACE LtInt'ING CilECIC VALVE.
- l INA INSTALL. TEMP l>ETECTuitS IN OFNP DISCH LINE N/ REMOTE COMPUTER MONITORING AND ALARM 1
CIIECK FOR LEA 10 AGE AF'IEl1 EACll OPERATION. TRAIN NOTCit-STANDER. CONSIDER INSTRUtIENT IJAN tl0NIlUR, INSTALL PYit0MElElts,Pil0 VIDE PROCEDURAL. GUIDANCE AND TRAINING, INSPECT.
I i
t e
$$ $ e e
o e
L b
a ee D
+
4
/
e 4. '
J
=.,
4M.
-9 p
e Q
l
- k y-
[, J L
P I $
m a
g
.e s
<1 4
9. me p4 4
, w.
e +,
,. - g%
.g p 4 +
6-a s 3e
- e 4
4 9
e.
9 I
g w
g -
[
s.
=
74,
4 4 9
==
a A
I 9
4
+
N 6
t.i r
4 4
(P 4
9 s
k
.. f
+
E e
4 i
e A
9 5
e 4
I e
f i
t i
t
7 f
3.
The steps anticipated that licensees must carry out to complete the requirements are briefly described below:
a.
Are there any short-term and long-term requirements?
Is the proposed bulletin the definitive, comprehensive position on the subject or does it represent the first of a series of require-ments to be issue,d in the future?
The proposed bulletin discusses both these aspects.
Developing and implementing procedures for monitoring, detecting, and correcting AFW steam binding will satisfy the short-term re-quirements imposed by this proposed bulletin.
However, NRR also -
is reviewing this subject under Generic Issue 93 and may impose l
additicnal requirements in the future.
I b.
How does this requirement affect other requirements? Does this requirement mean that other items or systems or prior analyses need to be reassessed?
L We do not believe that the requirements of the proposed bulletin will affect any other requirements.
Neither will they change
+
p other assessments.
I c.
Is it only computation? Or does it require or may it entail engineering design of a new system or modification of any existing systems?
L The proposed bulletin states that the requirements can be satisfied by simple means, such as an operator touching the pipe to make sure it is not hot.
Some licensees may wish to make system modifications, such as adding a resistance temperature detector with alarm and readout in the control room, I
d.
What plant conditions are needed to install and conduct preoperational tests and to declare the plant operable?
The requirements of the proposed bulletin can be implemented independent of plant conditions,
..y i
e.
Is plant shutdown necessary?
I No.
1 l
f.
Does design need NRC approval?
i The requirements of the proposed bulletin do not impose design i
changes.
If designs are changed, however, 10 CFR 50.59 applies.
I The most likely design changes are the additions of-temperature i
monitors, which would not require NRC approval under 10 CFR 50.59.
6 g.
Does it require new equipment?
Is it available for purchase in a sufficient quantity by all affected licensees or must such equipment be designed? What is the lead time for availability?
No new equipment is required.
If some licensees choose to install temperature monitors, this can be done with readily t
available equipment.
2
h.,
?.
p.
h.
May equipment be used when it is installed or does it need staff p'
approval before use? Does it need Tech Spec changes before use?
r If temperature monitors are installed, they may be used once installed without staff approval.
No Tech Spec changes are required.
4.
The requirements of the proposed bulletin would apply to the addressees for action given in Attachment 1 of.the proposed bulletin.
This consists of 28 of the PWR units having an OL and all 26 PWR' units having a CP.
These PWRs have been determined to need procedures concerning this i
subject.
5.
For each applicable reactor category, additional information is supplied i
below.
/
L Note that the action addresse'es are 28 operating units and'26 units under construction.
This amounts to 36 sites.when we count multiple units, such as Byron Units 1 and 2, as a single site.
For the purpose,of;the analysis k
described below, there is no real difference between the operating units and the units under construction.
)
e a.
.A risk reduction assessment performed using a data base and Q
methodology commonly accepted within NRC.
The risk of core melt because of unavailable.AFW is estimated in the AE00 case study, Section 3.2, and in the NRR prioritization for Generic Issue 93.
y In the AEOD study, the results of the Reactor Safety Study
[:
Methodology Applications Program (NUREG/CR-1659) for Sequoyah were used.
There, the accident sequence (labeled TML) for the E.
loss of the steam conversion system after a transient event d
other than loss of offsite power was a dominant contributor to 1 risk of core melt.
The unavailability of the AFW system in-creased by a factor of 4 above previous estimates because of H.
steam binding.
This increase doubled the contribution of the L
TML sequence to the probability of a category PWR-3 release and
[
added at least 1x10 5/RY to the overall probability of core melt. <This increased the risk by about 60,000 man-rem for the remaining lifetime of all operating PWRs (47 units with an average remaining life of 27 years when the AE00 study was published in July 1984).
In the NRR prioritization, the results of the Reactor Safety Study (WASH-1400) for Surry were used.
There, the accident sequence (labeled TMLB') for the loss of the steam conversion system after a transient event with loss of offsite_ power was a dominant contributor to risk of core melt.
Because of uncer-tainty, NRR considered a range of values for AFW pump failure i
probability.
For the lower value (basic case), the steam binding would result in the doubling of AFW system unavailabili-ty; the contribution of the TMLB' sequence then also would be nearl.y doubled.
The difference from the Sequoyah case is represented by the higher starting unavailability for AFW resulting from loss of offsite power in the Surry case.
The weighted sum of increased public dose resulting from considera-tion only of the TMLB' sequence for the 90 PWRs expected to be 3
~
- j. 'P, operating having an average. life of 28.8 years is about 30,000 oK man-rem.
It is possible that the pump failure probabilities i
used in the basic case are overly optimistic. Accordingly, NRR-also made a more conservative estimate which yielded a dose increase of about 96,000 man-rem.
r
,i I
The requirements of the proposed bulletin are~ intended to ensure this risk is not realized.
Monitoring once per shift is expect-ed to decrease the likelihood of steam binding substantially.,
,~
As discussed previously, the monitoring already is being con-ducted at many operating PWRs and the nther operating PWRs have o
some justification for not monitoring once per shift.
- However, many of the operating PWRs as well as PWRs under construction do not have procedures. Without procedures, the gains in risk' reduction are expected to dissipate _over time.
For instance,
/
simply ~ telling operators to feel the pipes once per shift without any further' explanation or guidance will not provide a lasting solution of reasonable quality.
The bulletin seeks to-L require such procedures be prepared by those licensees that have not already done so.
The needed training and awareness should -
follow as a matter of course from implementation of the proce-dures and the solution should be lasting.
H i>
b.
An assessment of costs.
(1) To NRC:
We estimate a man-week is required for inspection and inspection reporting and review'for each site.
At $2000 per man week, this i
translates to a total cost of about $70,000.
After a year or two has passed, an additional $30,000 will be needed to review licensee responses, close the bulletin, and print the NUREG Thus, the total NRC costs are estimated to be about (2) To licensees:
b f
(a) Occupational dose.
No additional occupational radiation exposure is expected I
from the requirements" of the proposed bulletin.
The procedures deal with a secondary system that handles l
nonradioactive water.
1
~
(b) Complexity of plant and operations.
The requirements of the proposed bulletin do not add any foreseeable complexity to either plant or opera _tions.
The monitoring already is being performed, where needed.
If, as a result of the bulletin, additional licensees decide to perform monitoring, this can be readily accomplished in conjunction with routine tours already conducted by.
l -
operators.
~
(c) Total financial costs.
We estimate the one-time effort would require about 11.5 man-weeks per site or about 400 man-weeks in all.
This 4
i
- i. -., _ _ _ _., _
. ~
j L
t
) i:
~ ',
'~
includes writing, testing, and promulgating the procedures h
and writing the response to the bulletin.
The cost of
~
one-time effort for 36 sites is then about $800,000. We estimate continuing effort to require about 45 man-weeks per year for monitoring the system temperature and about l~
140 man-weeks per year for retraining and maintaining the procedures.
This. amounts to about $300,000 per year in e
all.
i a.
t These estimates assume that, at each of the 36 sites addressed for action, procedures are written and imple-mented.
With regard to the actual cost of monitoring, most plants that need monitoring are already conducting it.
It was assumed that 10 additional construction sites would i
perform monitoring as a result of the bulletin.
i 3
c.
The basis for requiring or permitting implementation by a given 8
date or on a particular schedule.
Requiring the response to the proposed bulletin within 120 days for operating reactors allows a reasonable time period for l[
licensees to develop and implement the procedures, which essen-l
.tially only document present practice.
The time period is not i
E so long that licensees would tend to neglect the work.
The 7
3 p'
limit of 1 year for plants under construction is to allow for
.j j
orderly closeout of the bulletin within a reasonable time.
d.
Other acceptable implementation schedules and.the basis f
therefor.
i We currently foresee no real need for requesting a different schedule, but would consider any licensee's request for addi-If tional time on its own mer.its.
Since the monitoring is already.
,M being conducted, where needed, different schedules for imple-I
!.~
menting the procedures would not present problems.
e.
Schedule for staff actions involved in completion of requirement (based on hypothesized effective date of approval).
F The staff will issue the proposed bulletin upon approval.
l' Depending on region and resident inspector schedules, the staff plans to close the bulletin in 2 years.
l I!'
f.
Prioritization of the proposed requirement con'sidered in light l
of all other safety-related activities under way at all affected
^
facilities.
The problem, which this proposed bulletin addresses.in the near-term, has been assigned high priority by NRR in its Generic Issue No. 93, which addresses the problem in the longer term.
L Thus, we consider this proposed bulletin to be of high priority.
ll g.
Does this proposed requirement involve recordkeeping'l
~
Not explicitly.
However, log entries will probably be made to 4
record the results of monitoring once per shift.
/
5 3
1 p - - - -
....----------------,.---m.
-o-.--
f
.o,
' IE position as to bether the requirement implements existing regulations i
6.
-or goes beyond them.
l Because plant technical specifications require the AFW system to be operable in modes 1, 2, and 3, the requirements of the proposed bulletin for detecting and correcting steam binding implement exist-ing regulations and do not go beyond them. The requirement for' procedures on monitoring temperature of the pump casing or the discharge pipe amounts to formalizing an existing good surveillance practice. This ensures the continuation of the practice and, thus.
the availability of the system. Accordingly, we conclude that all the requirements of the proposed bulletin implement regulations and do not exceed them.
i'
,7.
The proposed method of implementation along with the concurrence (and anyi coments) of OELD on the method proposed.
l The proposed bulletin has been concurred in by OELD with no comments.
,u D
8.
Regulatory analysis sufficient to address the IJ a.
Paperwork. Reduction Act b.
Regulatory Flexibility Act s
c.
Executive Order 12291 This request for information was approved by the Office of Management and I
Budget under blanket clearance number 3150-0011 as meeting requirements of the Paperwork Reduction Act and Executive Order 12291. Sufficient hours are included in the NRC budget for this request. Since this is not a rulemaking L
action, the Regulatory Flexibility Act does not apply.
l.: -4 i
i; i
F e
I-t i
,L
.,g
- V t1 L
- I 6
i
/ds,rb M & fl Q
)
p7
.3
~
.S No.:
6635
..(
- h....E @^ ; -
r T- 'm~
IN 84-06 UNITED STATES
-s p3 ns. E a.:
NUCLEAR REGULATORY COMMISSION' N
.~
1,~
T. _
0FFICE OF INSPECTION AND ENFORCEMENT' ~~~
4 l-WASHINGTON, D.C.
y..
.r; -
- 20555
..,.a.A e;
~
=
>.y q:
g:
.;; m :d t :
~7t January 25, 198,4_, ]' 7.E y {.F
~.--
- r..
~W;. j~~_,
, ~..
3
~
lE INFORMATION NOTICE NO. 84-06:
STEAM BINDING OF AUXIi.IARY FEEDWATER PUMPS s n; r~
3^
^
~
7--
, c. p.
e v. _,. -
. Addressees:
- n.
_ e o,.
, ~...
All pressurized water reactor (PWR) facilities holding'an operating license
(%.) or construction permit (CP).
v l
-c Furcose: n u m m
' =^:
', ^
' ' ~
p.
c, This information notice provides notification of a problem pertaining to steam l'
binding in the auxiliary feedwater (AFW) pumps due to. leakage from the main,
L fetawater system.
It is expected that addressees will review the information prov' cec for applicability to their facil'ities.
No specific action or respense N...
is required.
[
?
Cescrio-Jicn cf Circumstances:
.On' April 19, 1993, Carolina Power and Oght reported that the two motor-driven AFW pumps started automatically on low steam generator level following a ir,anual
-scra at tne H. B. Robinson nuclear plant.. After two minutes, the B train AFW g'g The trip was attributed to a signal from lov discharge pressure, pu p tripced.
Tne cisc5arge piping frcm the motor-driven AFW train is connected to the rain 5 t
feeceater pi;ing near the steam generator.
(See Figure 1.) Hot water, ateut 225'F, fecm tne cain'feedwater system leaked back through the first check s alve, the,otor-c;e-ated valve, and the second check valve to the pump and 5
l-flashed to steam because of the lower pressure in the AFW system.
(A signi-y~
ficant amcunt of steam was vented from the pump casing during the testing to cetermine the cause of the trip.) When the motor-driven pumps started, the -
instrumsr.tation sensed a low discharge pressure. The steam binding reduced flow anc prevented discharge pressure from increasing above tne low pressure set-point in the 30 seconcs before the instrumentation tripped the pump.
Con-censation could have further lowered the pressure to the sensors.
dobinson had experienced leakage through valves in the discharge piping and consecuent trips of the A train AFW pump on June 11 and 16, 1981.
On July 21,
!?E3 the steam-driven pump was declared ineperable because of potential steam binding caused by leakage from the feedwater system.
Crystal River 3 repor ed 4
two steam-voiding events which caused the emergency feedwater system train B te
- e declared inoperable.
Two similar events were repcrted at D.Cs Cook Unit 2 i
8312 _20153 I
.._ h..,
,, C ~f q q -
d --:Es - a -,
/
~
'~
~
IN 84-L
' ] 'Q,. !.
January 25, 1984
- J'
"~
Page 2 of 2 r
in 1981.
(Reference LERs 50-261/83-044,83-016, and 81-016; 50-301/82-076, and 83-045; and 50-316/81-032 and 81-063.)
A special interim procedure at Robinson calls for the venting of all three pumps once each shift, monitoring of the casing temperatures, and operating the pumps as required to prevent saturation' conditions in the system.
Cook also monitors the AFW system temperature.
Robinson is exploring a design change or..
replacement of the check valves as a long-term solution.
The safety implication of the'se events is that leakage into the AFW from the feedwater system constitutes a common mode failure that can lead to the loss *of all AFW capability.
Further, there is the potential for water hammer damage if an AFW pump discharges relatively cold water into a region of the piping systtm l
that contains steam.
Since the design of the AFW at Robinson is typical of other PWRs, the potential for backleakage exists in other operating plants.
Routine monitoring of the AFW system temperature would detect backleakage so i
that the system could be periodically vented to prev'ent steam binding until an
[
appropriate long-term solution is developed.
No written response to this notice is required.
If you have any questions regarding this matter, please contact the Regional Administrator pf the appro-g.iate NRC Regional Office, or this office.
i Mar L ordan, Director Divisio f Emergency Preparedness and En ineering Response Office of Inspectinn and Enforcement d
i Technical Contacts:
M. S. Wegner, IE 301-492-4511 J. J. Zudans, IE 301-492-4255 Attachments:
~
t
~ _. Figure 1, " Simplified Schematic of Feedwater.
I and Auxiliary Feedwater Systems" 2.
List cf Recently Issued IE Information Notices
'}
i 9
I I
\\
k i
e schm2nt 1
. 84-06 J nusry 25, 1984 Page 1 of 1 r
i r
r l'1
~~
- 1 )
/
l
' Main Feedwater A/
\\
6 Vs g
s s
l Feedwater Bypass
/>>>,,,ii,,i/>>,,ir i,,
h Cen:rator i
Motor Driven A
Pump j
b MOV i
Auxiliary l
y fmf Feedwater i
(Suction)
Auxiliary l
' h <i (Dis:harge)
Motor l
h
/
Driven
]
2 r
B Pur.ip j
i 7
- d j
i
-? '
l t
(
! 4-Condensate l
Storage g
i Q
t sank y
l Steam Driven Pump i
Flow Designators:
l G No. mat Plant Czerating, AFW Shatdown l
h Post Trip Plant Shutcown, AFW i!
0:erating l1 1
Figure 1. Simplified Schematic of Feedwater and Auxiliary Feedwater Systems i
e
t I
E' g g 1,
1_'r,;J f
t AEOD/C404 STEAM BINDING OF-AUXILIARY FE5DWATEli PUMPS c
Reactor Operations Analysis Branch
~ '
/
Office for. Analysis and Evaluation r
p.
of Operational Data p
J' Y
p-JULY 1984 h-l t
i r
i
+
Prepared by: Wayna D. Lanning I
g t
l
~
NOTE:
Tnis report documents the results of study completed to date by the Office for Analysis and Evaluation of Operational Data with regard to a particular operational situation.
The findings and re:ommendations do not necessarily represent the position or require-
}
ments of the responsible program office nor the Nuclear Regulatory l
Co imi ssion.
<t, 0
4 >()
- 6 t
6 N'f8 t
i I
c f
n.
.?
.o TABLE OF CONTENTS L?
PAGE EXECUTIVE
SUMMARY
1
1.0 INTRODUCTION
.......'...............'.b'.
4 oa 2.0 AUXILIARY FEE 0 WATER SYSTEM DESCRIPTIONS.;. c. :. r!.Mi..
5 2.1 We stinghouse Pl ants......................
6-2.2 Babcock and Wil cox Pl ants................... 10 2.3 Combustion Engineering Plants.... s. s..'.......
13 p
2.4 Summa ry o f AFW De si g n s............ ;.....
.. 14/
30 ANALYSIS OF BACKLEAKAGE EVENTS................... 16 l
Operational Experience... :.......,.,........
=v-1 3.1 17 3.2 Sa fe ty Si g ni fi c anc e......
............... 28 E,
4.0 CAUSES FOR VALVE LEAKAGE...................... 34 5.0 LEAK DETECTION........................... 39 l-6.0 FINDINGS AND CONCLUSIONS...................... 43
,7.0 RECOMMENDATION............................ 46
8.0 REFERENCES
....................... ~...... 49 l
+
FIGURES z$
Figure 1 Schematic of H. B. Robinson Auxiliary Feedwater System...
8 Figure 2 Schematic of Crystal River Auxiliary Feedwater System... 12 J
Figure 3 Schematic of Calvert Cliffs Auxiliary Feedwater System... 15 TABLE Tabl.e 1 Summary of Backleakage Events (Since 1981)..........
21 t
i 1
APPENDIX r
Appendix A Vapor Binding of Auxiliary Feedwater Pumps........ A-1
.A 4
1 4
1 l
i
r t
E ~
EXECUTIVE SUMARY A case study was completed to evaluate the generic safety.impiteations of backleakage to the auxiliary feedwater ( AFW) system.
Backleakage is defined as 1
the leakage of hot main feedwater or steam from the steam conversion system to theAFVsystem. The AFW system is a safety system ort a pressurized water '
reactor (PWR) whose safety ' function is to provide a source of water for'the l
t steam generators when the main feedwater system is not available and to mitigate design basis accidents.
y g
Operational experience has shown that on numerous' occasions an AFW pump was L
rendered inoperable due to steam binding resulting from the leakage o'f hot 7
feedwater to the AFW system. Multiple valves in series between the steam conversion system and the AFW system leaked and failed to provide isolation between the interfacing systems. The safety implication of these operating j
events was that backleakage repres'ents a potential common cause' failure for the AFW system that can cause the loss of its safety function.
- 5 g
Coerzting experience involving backleakage to the AFW system since 1981 f
- [
included twenty-two events at six operating PWRs in the United States and one foreign plant.
These events involved the misoperation or failure of t:
about 60 check valves and five motor-operated valves installed to prevent
- i reverse leakage. Other plants were known to have experienced backleakage, ii but tr.e events were not considered as reportable occurrences. Tne events at 3
3arry power Station Unit 2', H. B. Robinson Unit 2, and Joseph M. Farley Units 1 and 2, provided evidence that more than one AFW pump can be simultane-
[
cusly adversely affected by backleakage.
The recent Surry event is the f
most significant event analyzed and is considered a precursor to a potentially
{
i e
3-
~
t L
v.,.,
- 2 serious accident sce:iario involving tiie loss of all feedwater. At
~
~Surry, the simultaneous steam binding of a pump in each train of the AFJ systein rendered the system Incapable of performing its design function.
The ciajor findings of the, study are:
.t
.. s:
1.
The trend of the operating events involving backleakage to the AFW,
j system increased sharply in 1983 when 13 of the 22 events occurred at
[
five Westinghouse-designed p1 ants.
I-F 2.
AE00's assessment of the safety significance of the events showed that t.
l (a)- the loss of a : ingle train due to steam binding is significant L
because it is presently an undetectable failure that jeopordizes the b
capability of the AFW system to meet single failure criterion, f.e.,
potential common mode failure, and (b) the unavailability of.tn'e AFM systen due to steam binding contributes significantly to risk of core melt in PWRs.
2g t
3.
The potential for backleakage into the AFW system is generic to all operating PMRs. The review of the AFW designs for the tnree P.'C vendors fou'nd that check valves and remotely-operated valves in some 1
designs isolate tne AFU system frau tne staan conversion systaa.
The t'
AFJ designs at Westingnouse-designed plants apoeared.. lore suscaptiale
- I i
to oackleakage and steam binding of the pumps because the reaotely-l operated valve is of ten nomally open.
Operating experi nce shewed f
that backleakage occurred primarily at Westingnouse-designed plants.
l t
t t
I f
r t
3-l L
4.
The potential for common mode failure of the AFW system is present
~
whenever one pap is steam bound because the pups are connected by common piping (discharge header and/or rec rculation piping) with only a single check valve to prevent backleakage of hot water to a second or third pup.
q 5.
While a potential exists for backleakage to other safety systems in both PWRs and boiling water reactors (BWRs), there is no known report p
of steam binding of a pap in other safety systems. The standby s,afety systems are isolated from operating systems at higher pressures and p
temperatures by check valves and motor-operated valves similar to the AFW systems.
The potential for steam binding is minimized because the
[
remotely operated valve is nonna11y closed and is leak tested (the AFW U
' valves are not).
However, leakage through an upstream, check valve ~ has t
caused the remotely-operated valve to fail to open due to thermal binding g
ano cther reasons--a separate concern from steam binding. A previous
[d AEOD study recommended measures to ensure the function of the valves I which should address this concern.
Since some BWR systems ceploy a i
smaller ninber of valves than were available in the AFW systems that experienced backleakage and steam binding of the pumps; a separate AE00 effort will further evaluate the safety significance of backleakage in
~
EWR systems.
6.
Tne analyses of the causes for check valve leakage did not identify any pattern or single major cause of the failures of the check valves.
The causes differed between plants and involved different valve designs.
4 8
6 m
O
+
p__
.. +.. -,... - - -
-~*--*.ee~
- ~*.-e-w e
s-~~
- ~=e~~
t
-E-
-. y w
- n:v 7.
The study did not identify any regulatory requirsaents or. unifom plant practices to reduce the likelihood of steam binding of tne AFJ punps and common mode failure of,the, AFil system.,,.
AEOD recommends that the'dffice of Nuclear Reactor Regulation require P'in-( '
licensees to monitor the AFW system 'to detect backleakage and ensure thatitne i
, fluid conditions within the AFW system are well below saturation conditions to 'revent steam binding of the AFW pumps.
p
>c.
n AE00 suggested a possible nethod for further consideration that could recuce tne likelihood of steam binding the AFil pumps and common mode, failure of the AFW systea. The method'contains two basic elements: first, preventive nessures to ' ensure that the isolation valves can perfona their intended function; and second, surveillances to ensure that the isolation function does not experi-i ence undetected degradation during operations.
F 1.0 INTR 03C TI0ll 5
Recurrent operational events at the H. B. Robinson Nuclear Power Plant involving auto:natic trips of the auxiliary feedwater ( AF;J) pu.aps prompted AE00 to perfom an engineering evaluation of the events.
The pumps tripped due to a los pressure sensed in the discharge piping after the pgaps were startea automatically. The cause of the low pressure was attributed to stoa.. oinding i
of the pumps from leakage of main feedwater (;4FW) into the AFW systeil (back-leikage).
The hot ;4F'./ (about 425'F) leaked past two check valves and a closed
.sotor-operated valve and flashed to stea,a in the lower pressure AF.i syste:3.
Althougn the events inv'olved only a single pump, tne sar.te phenomena hao I
t
.o.
P
.c l
occurred in different AFW pumps at different times. Thus, a safety concern
.t.
" as raised that both ' trains could become steam bound simultaneously.
The Engineering Evaluation (Ref.1) concluded that'an Infomation notice.
should be issued promptly 'to inform other licensees of the potential for tne loss of AFW capability due to backleakage and steam femation in the AFW
/
system. The Office of Inspection and Enforcement issued the Infomation i
_ Notice on January 25, 1934.
In the meantime, AE00 proceeded with a case 4
~
study to evaluate the generic safety implications and to identify potential I
. changes to technical specifications and inservice testing programs to detect L
leakage into the AFW system and prevent steam binding.
f.
Dis case study report documents the results of AE00's activities with
-egard to steam binding of the AFW pumps.
Representative designs of AFW f
systes at operating PWRs are evaluated in Section 2.
An evaluation of the f
- perational experience dealing with reported sacxleakage events is contained i
1 r
in Section 3 followed by a discussion of tne causes for valve leakage in section 4 Requirments for leakage detection are, sodressed in Section 3.
3ec-f on 5 presents the findings and conclusions develooed fraa tne analysis sad evaluation of AFW steam oinding phenomena which form the bases for t::e
-ecoinendations contained in Section 7..
h
]
2.0 AUXILIARY FEEDWATER SYSTE:1 DESCRIPTIONS
'i Thi.s section su::tnarizes the AFW system designs for operating.Jestingnousei l
Sasco:k and Wilcox, and Combustion Engineering plants based on :ne AF',.' systou t
i-
- es:riptions compiled by the 3RC Bulletins and Orders Task Force in..U.ES-OH3, 1
1
{~q h..
NUREG-0611, and NUREG-0635, respectively.
t (Note that as a result of this and other post-TMI activities the designs of some AFW systems were changed or are l
_-i.
undergoing changes.
The descriptions contained in this section reflect the configuration of the AFW system at the time of this study.) The designs are reviewed to detemine (1) whether the potential for backleakage exists during h
nomal plant operation when the AFW system is not operating and (2) whether f
~
multiple pumps could be simultaneously affected. Thus, the focus of the f
review is to first highlight the number and kinds of valyes that are used to t
isolate the AFW system from the steam conversion system (main feedwater and f
steam generators) and then identify common piping between the AFW trains which L
could provide a flow path for MFW or steam that could lead to simultaneous f
. ste'am binding of.the AFW pumps.
2.1 Westinghouse Plants'
~
The review of the AFW designs at Westinghouse operating plants found that the i
~
AFW and MFW systems are isolated either by multiple check valves in series if t
beca'use the remotely operated valve is nonna11y open, or by multiple check valves and a nomally closed remotely-operated valve.
Thirty of the operating ti plants use only check valves while only three plants (Robinson, Sequoyah, and
. Turkey Point) employ the latter configuration for isolating the interfacing sys tems.
Since Robinson experienced more backleakage event's than other f
plants, its AFW system is described in this section.
l ff The primary differences among the plants that use only check valves to ikj; isolate the interfacing systems is that in two plants the AFW system is 49
~
lt o
- -.. - -~..
i
.c mconnected directly to the steam generators via the ;tFW bypass piping ratner tilan to tne !!TW piping as for the other 28 designs. These two plants n
(::c3uire and Summer) employ the tiodel D preneat steam generator design.
A saall feedwater flow rate is maintained in the bypass pt. ping to prevent back1'eakage~
i from tne steam generator.* However, even in these two designs the AFW and.;FJ systems also interface at another location that provides the same potential for backleakage as for other West.inghouse steam generator designs.
A schematic of the AF;l system for the H. B. Robinson. Unit 2 plant is snown in Figure 1.
The two motor-and one turbine-driven pumps share a coanon suction from the concensate storage tank. The two motor-driven pumps, discharge to a
[
c:imon header before tne flow is piped to each of the three steam generators.
A sir.gle eneck valve exists between each pump and the header. A check valve and a nomally closed motor-operated valve exist in the piping betwen esca stoa.1 generator and the connon header. Thus, if the check valve and motor-operated 4
valve leak in tne pipe to either steam generator, only a single checx is availaole
- o crotect tne pump from oackleakage.
(
The taroine-oriven AFJ pump is connected to the ;tF'.! pipin; oy a separate u
This piping contains only one enecx flow patn via tne. *tFW bypass piping.
valve anc a nor ally closed cotor-operated valve for isolation.
7ne tisenarge of all :nree punps is connected by a coxton recirculation pipe I
to cre condensate storage tank (see Figure 1).
.l hen the 't~!! Icaks i.ito tae AF'a systri, the water in tne..ornally filled pipes is transferred to tne concensate storage tank througn the recirculation oiping.
i t
i i
1
f t
L E
/
h Main v
~
Feedwater Bypass f
Steam Generators 2 and 3 Steam (Same Valve Arrangement)
Generator s'u's s
Dnven
/
A Pump Motor Recirculation b
Auxiliary Piping Feedwater L
(Discharge)
Mt Motor i
B Pump e
Steam Generators 2 and 3
[
ISame Valve
?.
Arrangement)
Condensate Storage i
Tank Steam
\\8 l-Driven
-E Pump 34 s
Flow Designators:
k Normal Plant Ocerating, AFW Shutdown h P:st Trip Plant Shutcown, AFW Operating i
Figure 1 Schematic of H. B. Robinson Auxiliary Feedwater System Suk D
O I
l O
..; 9 l
A single check valve in the recirculation piping separates each of the pumps 1"n the AFW system. Hence, in the event that a pump becomes steam bound, the recirculation piping provides a flow path for the hot water to the ottier pu:nps if the single check valve in the piping to each of the othier puinps leaks. The seating force for the check valve is provided by the column of water i
from the valves to the condensate storage tank, which may not be effective in properly seating the valve to prevent backleakage because both sides of the check valve comunicate with the condensate storage tank. The valve in the recirculation piping near the pump discharge opens each time the AFW pump cperates. Operating the pump would augment proper seating of tte check valves l
c in the recirculation piping to the other pumps. But after the pump is shutdown, t
the differential pressure across the check valves may equalize, possibly unseating them.
}
1.
The I:Guire and Sumer plants have Westinghouse Model D preheat steam generators
~- [
t and separate AFW nozzles to the steam generator. For these plants, the AFW system is connected to the steam generator via the MFW bypass piping rather than to the MFW no:zle as is the case for older Westinghouse stean generator designs. A small feedwater flow rate is maintained in the bypass piping to prevent backleakage from the steam generator.
However, the AFW system lj is still connected to the MFW piping at an upstream location.
At McGu1re,
~
- f one motor-driven pump and the turbine-driven pump share a common discharge header.
There are **-) check valves between each AFW pump and thE connection g
to the MFW piping and the remotely-operated valve is nomally open. Both plants have temperature indicators on the MFW bypass piping near the auxiliary
.ii!
l W
j7
--n.-
~[
feedwater nozzle and downstream of the intersection of the AFW and MFW piping.
The purpose of this instrumentation is to detect steam in the-
. feedwater bypass piping to prevent water hammers in. this piping, rather than to detect backleakage to the. AFW peps from the MFW system. The instrmen-tation is not capable of monitoring for backleakage to the pnps because of l
its location. ;
L
_ y
~
In general, all AFW systems in Westinghouse operating plants have at least one che'ck valve and a remotely-operated valve in series which can isolate an AFW L
h train from the main feedwater system.
However, the remotely-operated valve is
~
nomally open in most systems. As a result, only the check valve (s) provide the isolation functio'n between the AFW and MFW systems. The flow control valve is nomally closed in some plants, but this valve is not intended to be an isolation valve.
In about two-thirds of the Westinghouse designs, at least two'of the AFW pumps share a common discharge header with a single check valve
- f
-between the header and a pump. For some plants, only the motor-driven pumps share a conmon discharge header; in other designs, all three pumps share a common header.
A few designs have separate flow paths to the steam generators from each pump.
Backleakage has occurred in each of the designs.
For most AFW systems, all pumps share a common suction header from
}
the condensate storage tank and are connected by the recirculation piping.:
2.2 Babcock and Wilcox plants
~
The review of the AFW designs at Babcock and Wilcox (B&W) operating plano found that the AFW system is connected only to the steam generator in all designs except l
for Crystal Rive',which is connected to both the steam generators and the MFW system.
?
i
I j.
Only Davis Besse, Oconee and Arkansas Unit 1 employ nIrmally closed isola-tion valves in addition to che,ck valves to isolate the interfacing systems.
The other B&W plants use only check ' valves to isolate _the AFW system because
~
the isolation valves are normally open.. The flow control valves are nomally closed in most plants, but these are not intended for isolation purposes.
All designs have a common pump discharge header except for Davis Besse,
/
Oconee and Crystal River.
The recirculation piping and the piping from the i
condensate storage tank are common to all pumps in' all designs. For discus-sion purposes, a diagram of the AFW system for Crystal River is shown in Figure 2.
P At. Crystal River the pumps consist of a full-capacity turbine-driven pump L
and a full-capacity motor-driven pump (some B&W plants have 3 AFW pumps).
The piping arrcnoument is that either pump can deliver emergency feedwater to botn steam generators.
The piping is separated so that the pumps do not
- g
, share a ccamon discharge header, e.g., isolation valves exist between the pumps and the piping connection to the steam generators. The recirculation piping, however, is common to both pumps with two check valves available to i
crevent cross-flow between pumps, f
The AFW flow f r D&W plants is through separate nozzles in.the upper region ento the steam generator tubes.
As a result, the termination of the AFW discnarge piping is in a slightly superheated steam environment.
This design subsects the AFW system to higher pressures and te.nperatuies from the
{
steam generator than other AFW designs.
The Crystal River design also I
g,,y _-
i 9
i.
3;,
,.. IOU I!
- nip '
"V
^: 0 '
r 4
Stam Y
l'
.. l t*
(
Generseer r s
~
,, t ci;,
P' y"*0
'in' 47' h.
~
~
+
j
.?
... r..
- l l- '
u'-
. b,a v-~:n +
.. 5 's
.;ninu. r...,
C':
~'
u r-
.,:, : t'c : n v t v u
..u.
,..J
.r.
~
a u
- s
' r s
r.
c p
L
.-s f
f A.
y
- 1 t,
9 i
t.
.s..
-m
-r O
5 '
T' ya v-t Steem Generecer No. 2
-l i-(Same vow Anvengemene t-
' P s'5 j'g 9
I
,f,
..i,
+
2 4
4 T
- lA e
4I
\\
E Aecwcuoten Pe ng
[,
- Moto, Turbee Cusven D'*'*n Coo.eerwe r mo Pu'"#
stoesge u
Teae i
Flow Designatete:
teormes.Piens Coe'steg. APW Shutoown h A8W Operaten MFW Shuteewn
. - - +
Figure 2 Schematic of Crystal River Ausiuary Feedwater System O
s f.
,,,,n-
...n..,-.ws
~
~--~~*~~e**~~e-~-~w
'**'**'m****'*
' ' ' ~ * * * ' * '
' 53 -
6
'~
~
connects the AFW and MFW piping as shown in Figure 2.
During nomal operation, four check valves isolate the, interfacing systems to prevent hot MFW from leaking into the AFW system, since the remotely-operated valves are nomally e
open. Three of these check., valves and.an additional check valve near,.,<. :
t the steam generator separate the steam enviroment from an AFW pump.
1 T'he other B&W designs do not interface with the MFW system. However, steam 1
j binding of the AFW pumps could result from the steam fomed in the system l
if the AFW is heated to saturation conditions by the leakage of steam from the steam generators.
Backleakage to'an AFW pump is prevented in these designs by four check valves and a normally closed remotely-operated valve L
in the Oconee plants and two check valves in the Rancho Seco plant (the l
motor cperated valves are noma 11y open) or a combination of at least two
+
~
check valves and a nomally closed motor-operated valv'e at the Davis Besse and Arkansas Unit 1 plants'.
e 4
i The pumps in most designs share a common discharge header and recirculation pi pi.,g. At least one check valve is available to prevent backfles to the other pump if one pump becomes steam bound in all designs except for Crystal River (discussed above) and Davis Besse.
The latter plant does not have a common t
discharge header, but the recirculation piping connects both punos with a Neck valve to protect each pump as is the case for other B&W plants.
I 2.3 fenbustion Engineering plants I
The designs of the AFW systems for Combustion Engineering (CE) plants differ between plants and there does not appear to be an AFW design that is typical a
i f
i
I
_ 14 _-
for operating CE plants. A diagram of the Calvert Cliffs AFW system is shown in Figure 3 for reference.
The AFW pumps usually consist of either a 100% capacity motor-and a 100t capacity steam-driven pump 'o'r one full-capacity turbine-driven and two half-capacity motor-drive' planps (Calvert Cliffs has two 100% capacity n
1 turbine-driven pumps and a 100% capacity motor-driven pump).
Except for l
Arkansas Nuclear One, Unit 2, the other CE AFW designs employ a common discharge header for at least two pumps. The recirculation piping is common to all pumps with a single check valve to prevent backleakage from another L
pump.
Typically, the AFW designs employ tw check valves and one nomally closed remotely-operated valve to isolate the AFW from the steam conversion system, except for Calvert Cliffs, where the remotely ' operated valves are
~
partially open.
The discharge of the AFW system is connected either directly to the steam generator downcomer by separate nozzles or to the MFW piping 4
upstream of the main feedwater nozzle.
For some of the plants that have the; separate AFW nozzles, the AFW piping is also connected to the main feedwater piping.
Ii 2.4 Summary of AFW Designs To struarize the review of the various AFW designs, all systems contain j
cultiple check valves and,at least one remotely-operated valve in series that can isolate the steam conversion and AFW systems.
The operation and designs of AFW systems vary considerably among operating PWRs.
ene primary difference betaeen the AFW designs for the three PWR vendors is that the remotely-operated I
j valve (s) is nomally open in Westinghouse plants (except Robinson, Sequoyah, I
t I
j
m j,
e,
/
3*
(
~
Main Feedweter g -
_ y,
,4._
ts t
,.L,. $tgem -,.
s
.Gener8tof
,,,_y.
,,7'",,,,..
i::
W " *, F
- - No.1
- 9
6 " '
~
- g. a u
I
\\'
m, 7_
.. : sW.i..
- n.
i
' ~ ' '
j_
4 1
I
['
~
O Steam Generator No. 2 D
Steam c,nerator No. 2 (Same Valve Arrangement) 1 P-s's
.fSame Velve Arrangemeno
,as 4
<>+
1
- k.,'.
t[
L 6
'N*
Plec,lrculation Motor Pioing
- J' '-
, Driven j
,j; L
1 qume F
i c:o Turbine Turbine r
Driven Driven Pumo Pump j
Condensste Storage s
Tank
,4 f
Q h!
i' f.,
Flow Designators:
Normal Plant f
Oce st'ng. AFW Snutcown h AFW Ocerating
'~-
MFW Shutcown
^
~
1 f
f Figure 3 Schematic of.Calvert Cliffs Auxiliary Feedwater System f
l
. I
t 7 _
and Turkey Point) and B&W plants (except for Davis Besse, Oconee and Arkansas, Onit 1) and is nonna11y closed in CE plants (except Calvert Citffs). Hence, a majority of operating PWRs have only check valves to isolate the AFW and -
e steam conversion system whi,ch could make them more susceptible to backleakage.
I and steam binding of the AFW pumps than the plants which have a normally closed remotely-operated valve. The AFW pumps in most designs have common #
piping which increases the potential for steam binding of a second or third pump if any one pump becomes steam bound.
p The review of the AFW designs did not identify any available instrumentation to
[
monitor or detect backleakage from the steam conversion system. '
3. 0' ANALYSIS OF BACKLEAKAGE EVENTS Backleakage is defined as the leakage of feedwater or steam from the steam conversion system to the AFW system. This backleakage occurs while the AFW E
system is idle and the main feedwater system is operational. As discussed t
in the AFW system descriptions, check valves and, in some designs, remotely-operated valves isolate the AFW system from the steam conversion system.
Because the MFW and, steam generators are at a higher temperature and pressure than tne AFW system, the effect of backleakage on the AFW system is to 1
increase the temperature of the AFW fluid to saturation conditions that can s
result in steam in the system.
Steam in the pump ising can prevent full I
flow or cause pump cavitation and possible damage to the pump due t'o overspeed and vibration, t
The potential exists for steam binding of the redundant AFW pumps because the trains are crass-connected in most designs by a common discharge header i
I I
i I
p.
17 -
~
e.
l
~
I:
and/cr by common recirculation piping.,In order for a single pump to ex;erience l
steam binding, it is relevant to note that leakage aust occur through multiple
~
- valves in series. 'Howev'er, after one pump is exposed to steam in most plants',
~
7 only a single check valve in connecting piping is usually available to prevent
~
leakage and potential steam binding of another pump.
/
.w
~
~',
, 3.1 Doerational Experience -
r s
Since 1981 and more frequently in 1983, events involv,ing backleakage were reported at H. B. Robinson (Refs. 2-5), D. C. Cock Unit 2 (Refs. 6-3),.tillisa y
B..icGuire Unit 1 (Ref. 9), Crystal afver Unit 3 (Refs.10 and 11), Surry r-Power Station Unit 2 (Ref.12), and KRSK0 !1uclear Project (Yugoslavia, Ref. 13).
Taole 1 proyides a tabulation of these events.
The events at al. 3.
[
Rooinson and D. C. Cook are described in Reference 1, ahich is enclosed in
~
Appendix A.
The reader is referred to the Appendix for details of these ever.ts, particularly the events at Rooinson.
n-If 5
~
The event at the Surry Power Station, Unit 2, is the most significant operating L
event oecause it is considered a precursor to a potentially serious accident scenario involving the loss of all f,eedwater (:1F.1 and AF~1).
'hile at power i
.<itn MF.! available, one of the AF'l motor-driven pu.1ps and the taroine-driven pro are sinultaneously stean bound leaving only a one 'nalf ca:acity i
.aotor-driven pump available. Thus, the AF'.1 systen was not capaole 'of I.
perfoming its design function, although one punp nay oe sufficient to I
The systen as inoperable oursuant to
(
recove decay heat (see Section 3.2).
I the technical specifications, and this event highlignts the co:.uon cause i
1
.-~-.-....-.e~.~..eweem-
~> --.
+
-.-..-.+,-_n
18 -
t failure potential of the AFU system.due to backleakage and steam binding of the pumps.
Backleakage had, occurred previously, but it had affected only a single AFU train.
Fortunately in the event of loss of all feedwater, at e
Surry Unit 2, there is the capability for the AFW system at Unit 1 to supply i
emergency feedwater to Unit 2.
n
/
The coincident failures of two AFW trains was the identified concern resulting from the previous analyses of the Rcbinson events (see the Appendix).
Separate trains had failed at differ Ht times at Robinson, but elevated AFil t
temperatures provided evidence that multiple pumps could oe simultaneously i
a f fected. This led to the conclusion that the failures of single AFW trains
~
should not only be considered as random failures, but also as contributing events leading to the potential common code failure of the AFW system.
The Surry event provides additional evidence to support this h6nclusion'.
)
b It is noteworthy that backleakage in these events was detected indirectly and E
t r.eported only because the AFW train was declared inoperaole.
For exanpl e, three events at Rooinson involved an automatic trip of aither a motor-or stean-driven AFU pump due to low discharge pressure after an automatic start which caused the train to be declared inoperable.
The events at Crystal River involved a single train of AFW systen seing declared inoper-i able cecause a flow sensor failed. The backleakage at 3. C. Cook was i
i detected during a routine operator tour by feeling the piping before the I
pumps were required to operate, but wb reported only after tn'e pump was isolated to work on the check valves.
Backleakage from the stemi generators o
6 4
.w
=.
l
,0 g O.
19 -
at XRSKO was identified after experiencing a severe waterhammer. The events at Surry were reported because. the pumps were steam bound.
At Crystal River, recurring failures of the ultrason{c AFW flow instrument were attributed to backleak'a'ge which increased the temperature of the AFW piping and fluid and caused steam formation in the system which resulted in erratic indications and subsequent failure of the instrumentation.
The train was declared inoperable. The pump was not thought to be affected by the 1eakage, although the pump casing was not checked after the event for high I
temperatures indicative of backleakage.
For this case, three additional check L
valves were availabie between the leaking check valve and the pum;i. Thus, ins.trument readings and eventual failure provided an indirect indication of check valve leakage.
Tne events reported at the William B. McGuire (Ref. 9) and H. B. Robinson L~,
(Ref.14) plants involved backleakage which caused the suction piping of the
- 4.
p
),FW. system to be overpres:urized.
These events were caused by either the slow c1cstrg (McGuire) or improper seating of the discharge check valve (Robinson)
L' whien permitted the,MFW to pressurize the piping. Although these events l
,invo ved gross oackleakage, they represent another mode where the AFW pumps can j
f become stea:n bound.*
s i
Gross backleakage frcm the steam generators to the AFW pumps occurred at the l
KRSK3 plant in July 1981 during hot functional testing. KRSXO is a two-loop L
Westinghouse plant with preheat (Type D) steam generators (separate AFW nozzle to the steam generator).. The significance of the event was that a
- AiOD nad previously analyzed these events as they related to overpressuriza-tion in an Engineering Evaluation Report (Ref.15).
i
.r L
, i e,
severe waterhammer caused damage to the AFW piping and hangers associated sith both steam generators. The damage was not discovered until'several
weeks after the incident was believed to have occurred.
The main feedwater r.
system was probably not in. operation at the time (Ref.14).
Based on the l
infomation available, the AFW pumps were started and stopped during the [
testing.
It could not be ascertained whether the AFW pumps tripped or the #
intemittent starting and stopping of the pumps was perfomed by the operator.
L Two check valves in the piping to each motor-drive'n pump leaked while the pumps were idle between restarts, and was indicated by evidence of high temperatures (blistered or discolored paint) on the AFW piping back to the motor-driven AFW pumps. The turbine-driven pump is believed not to have been
~
affected, although this train was not checked for leakage at the time of the event. The AFW pumps were not required to be operable because the event occurred during preoperational testing.
h dd As tne result of the KRSKO event, temperature instrumentation was insta11edi on the MFW bypass piping near the steam generators at McGuire and Summer to detect and prevent waterhammer events similar to the one that occurred at KRSK0 because the steam generator design (Type D) and AFW piping layout are similar. This instrumentation is not intended to detect leakage to the AFW pumps because the connection of the AFW and MFW piping is upstream of the
}
instrumentation. ~ A small constant flow rate is also maintained in.the bypass piping to prevent steam fomation or backleakage in this small section of the E
AFW piping.
The AFW and MFW systems are connected at an upstream location, providing a potential leakage path to the AFW system.
O
(
77'
~~
1.~
~ '
)
4 Ib ik
P
[
- -: 21 -
TABLE 1 SultMRY OF BACKLEAXAGE EVE!!TS (Since 1981) flo. of Valves, Concents
~~
Plant Date Leaking e
Cook-2., -
_7/12/81
. 2 check Turbine-driven AFil pump (TDAF'.JP)
' valves (CV) casing was hot. Pump isolated ind the train declared inoperable.
Cook-2 10/29/91 ~
2CV TDAF'.iP casing was hot. Pump isolated and the train declared inoperaole.
Cook-2 1/16/83 2CV TDAFilP casing was not. Plant in
(~
operational node not requiring pumps
[
~
to be operaole.
I Crystal ?.tver-3 12/20/82 1CV Train declared inoperable. Backleakage F
caused flow sensor to fail.
E' F
Crystal River-3 10/03/33 1CV Train declared inoperable. Backleakage b
caused flow sensor to fail.
~ '
Robinson-2 6/11/81 2CV and 1 itotor-driven AFJ pump (. DAFWP) tripped -
motor-operated during plant startup.
valve (:10V) f L
P.coinson-2 6/16/81 2CV and
- CAF.lP tripoed after reactor trip.
r 1 MOV 4
L
?.coi nso n-2 6/19/81 Unknown
!!DAF;1P tripped on low oisenar"ge pressure after reactor trio. T)AF'.I out of service. Puap trip celieved to be caused by improper discnarge
'i valve tnrottle setting. Saae pun; tripped on 6/16/S1 due to stean binding.
.lo ci n so n-2 4/19/33 2CV and
- 10AFJP tripped after reactor trip.
1 i10V Stean vented frca p::ap casing.
Roainson-2 4/20/33 4CV and Both.iDAFilP casings were ilot. The i
2:10V or leakage pat'.i for the not sater to I
3CV and the second pump was not identified.
I fl0V Leakage to the second Dunp nay nave j
been tnrougn eitner the co:non dis-cnarge header or ::le recirculation piping through a single cneck valve.
(See Figure 1).
P.obinson-2 7/21/83 1CV and T]AF;!P casing was ilot and sti:r1 ventes
.l 110V from tne casing. Train inoperaole.
I
f,
- i i TABLE 1 (contd) l SuittARY OF BACKLEAKAGE_ EVENTS (Since 1981)
L Plant Date No. of Yalves Co::nents Leaking e
Surry-2
.11/18/83 4CV
!!DAFWP steam bound and failed to develop flow.
e -
Surry-2
'11/20/83 8CF ltDAFWP and TDAFWP steam bound, AFU systen was inoperable.
Surry-2 12/06/83 4CY ltDAFWP steam bound. Train declared inoperable.
i Farl ey-1/2 Ongoing 4-12CV MDAFWP arid TDAFUP casings were i
since per unit hot, sometimes at the same time.
L mid-83*
Pumps were run to reduce tenperature.
[-
No pump declared inoperaole by L
licensee.
[
KRSKO 7/81 4CV Waterham:sers occurred when AF'.lP started. Event occurred during pre-operational testing. (Pumps not required to be operable.)
OcGui re-1 3/25/31 2CV Slos closing of CVs caused tne AFW pump suction piping to be overpressurized.
$d i
e I
I i
)
1 l.
A nininua of six events are assumed to have occurred at oath Farley units althougn each train has been affr.cted more than one time since 19d3.
3
l
- 2 3
. ;q.
L 1
A search of the operating experience data. bases did not identify bac. leakage I
~
pr:olens affecting AF;l pumps a,t other operating plants..It codla no be ascertained wnether other p1_ ants had'not experienced,this' problem or unether the problea existed, but was not identified as the root cause for the reported events involving dnoperable AFW trains. Foi exampl e, unen <i. ' 3.'
7.obinson experienced failures (LERs 79-32, 33 and 34) of the AF~.! pump
/
cis:harge notor-operated valve to open, the initial causes were attributed
~.
to ettner the Limitorque operator or the inadvertent operation of the power l,
supply creaker. The final evaluation of the valvre failures concluded that tnemal binding caused the valve to stick closeu, which ultiaately affected e
- ne in eraction between the torque switch and valve internals.
Backleai: age g
was':ne reason icentified for tne themal binoing.
Crystal River nas also reported failures of the.~ motor-operated valve to open, but the cause was attributed to other reasons, although oackleakaSe is known to nave occurred in ti.eir AF'.! systen.
H. B. Rooinson hao also reported AFil pump trios due.tg 7
i1:r.::er tnrottle valve settings.
It is possiole th'at backleakage iay i: ave t'
- suse: the trio because the trip was initiated by the same low cis:har;e k
- ressure instrumentation that caused the pump trip wtien stea1 binding of the
?
pro.eas p:sitively identified.
I
.g scoulu :e n ted ::iat unless the backlaa.: age results in an eva.:.nica is t
5
- ..eNise re;:ortaa'le sy the tecnnical' specifications, the fact t.ut ca:k-
~
t I
lea': age occurred is not reportable. We nave been told inforully t:iat
,{
- ner cperating plants, in addition to 3. C. Cook (Ref. 9), have experien:20
~
leacJe of the AF'.! discharge valves. This leakage has occurred in botii AF'.!
trains several tices and was never judged to be a reportable event.
i
[
k
, D' On several occasions since very early in the plant' life of Farley Nuclear Plant, usually after stopping. flow through the check valves, leakage past
.the stop check and the check valve downstream of the flow control valve (FCV) has been observed..In. each case the AFW pumps were started to flush water through the check valves and, when the pumps were secured, the check valves would reseat.
l In the summer of 1983, the symptoms of check valve leakage changed. The Valves started leaking without an initiating event, i.e., without flow L
through the valve.
When this occurred, surveillance of the feedwater ifne r
p.
temperature was increased to once every 4 hours4.62963e-5 days <br />0.00111 hours <br />6.613757e-6 weeks <br />1.522e-6 months <br />.
In late July, 'backleakage was detected past the motor-driven and turbine-driven AFW pumps discharge r
. check valves. The existing on hand spare parts were used to repair the motor-driven AFW pump check valve. Sufficient parts to repair all the valves were not in stock at that time.
Some parts were already on order and addi-d tional parts were placed on order to repair all the check valves.
The i
I leakage past the check valves was slowly getting worse and the surveillance of the auxiliary feedwater piping temperature was increased.
In order to keep the temperature' below 180*F, the AFW Pumps were typically being run once i
a day.
[
4 In November 1983, some of the parts were obtained and the worst leaking f
check valves were repaired. However, the leakage continued to require the periodic running of the auxilicry feedwater pump to cool down the lines I
I.
cownstream of the motor-driven and the turbine-driven pumps' discharge check valves.
s r
j
- 2s -
On December 4,1983, the motor-driven AFW. pump was run about-20 minutes to' cool down the auxiliary feedwater lines.; During the check of the feedwater a
line temperature after running.the pump, the pump discharge piping was found to be hat and the suction relief valve was relieving. The pump was started to, I
cool down the lines but was immediately secured when pump cavitation occurrid.
,t s
The other pump was started and the system cooled down and the motor-driven pump was declared inoperable and isolated.
It was determined that the back l-leakage through the check valve nearest.the pump was cassed by one missing ~
p and one worn hinge pin bushing in the valve. These parts were replaced by -
a b bushings designed by the valve manufacturer to prevent recurrence of the problem.
It should be noted that the valve failure which resulted in steam
- binoing of the motor-driven AFW pump was a sudden type failure with substan-r t.
tially greater back leakage than had previously occurred.
6-s The only event reported by the Farley plant (Ref.17) was the backleakage through the check valve closest to the motor-driven AFW pump at Unit 1.
The AFW train was declared inoperable to repair the valve.. No mention aas made j
.in tne report that the three upstream valves had also leaked in order for tnis valve to leak, and that the backleakage caused the relief valve on the AFW pumo suction to lift. An operator during his routine rounds ncticed g
that the relief valve was opening, and measured piping temperatures in,n.
4 h[
excess of 200*F.
For some time, both units of the Farley plant have experi-en:ed recurring events involving backleakage through check valves that were t-1-
not reported. Presently,.the AFW pumps at both units are run periodically
- j to reduce the AFW fluid temperature.
I c -es e-e~-
N ew=4mm ++ -
= " "
.-,,e.,e=
r t-1
- 26 ;
There have been numerous AFW check valve failures reported in Licensee Event keports. The descriptions of the events address single valves and do not identify multiple check valve failures that could lead to steam binding off the AFW pumps.. As a result,.these events have not been included in this study.
These events, particularly the failures of the check valves to close and separation of the disc from the disc am, contribute to the potential for I
backleakage and steam binding of the AFW pumps.
In addition, there have been reported failures of the single check valve in the recirculation piping that could provide a path for steam or hot water to reach the other AFW pumps.
u Operational experience shows that check valves, in general, have' a history of I
leakage problems in all systems.
Most plants consider check valve leakage as routine and expected.
Operating experience shows that the check valves in the AFW system also fail open or leak.
1 SteambindingoftheAFWpumpscanalsoresultwhentheAFWpumpsareexposegto 24 hot water besides leakage from the MFW systems.
On April 7,1980, both AFW pumps 7'
lost suction due to steam binding at Arkansas Nuclear One, Unit 2 (Ref.16).
The y
suction of the AFW pumps was aligned to both the ' condensate storage tank and
.to tne startup and blowdown demineralizer effluent.
The hot water from the startup and blowdown demineralizers f1 ashed to steam at tne' pump suction and I
caused cavitation of both pumps.
The operators isolated the flow from the demineralizers and vented the pumps.
The procedures were revised to isolate the effluent from the demineralizers when MFW is available.
(
i t
I
I t
- 27 ine-e are other systems in Pilas where the interface between opercting systems a
nigh temperatures and pressures are separcted frca stenoay systems bj :ne:k
'vtives and ' remotely-operated.valvebin series, e.g., the emergency core
- ling system (ECO3).
Thus, the potential also esists for occkleakage to e
L these systcas.
Although there are reports of these valves leaking, no ever.t is Knoan to involve stea.a binding of the pumps.
The remotely-cperate: valve is n:rmally closed wnich should minimize the potential for cackleakage to the Additionally, tnese va5ves are periodically leak tested -(the A7,i pu p.
vaives are not) to ensure their. leak integrity (" ee S'ection 5).
n:wev'er, s
c
...e reverse leakage through the check valve 'can adversely affect the cperaoility y
~
of ne motor-operated valve as evident by the Robinsen events-(see Se::ica P-C.
An AE00 study (Ref. IS) of valve cperator-related events also fcur.d
- nat checx valve leakage can cause failure of the motor-operated valve to cpen nnen required.
Thus, check valve leakage has other sa'fety implica:fors in,adcition : ::e:m oinding of cumps.
-T
=
ing ;0tential also exists for backleakage frca.the :L' sys em t: nfi y-rela ed L'
sys e?.s in boiling water reactors (BWRs).
For example, a cr.eck valve and a i
nc - Illy close:! ceter-operated valve isolate tqe high pressure 001:nt i-je:-i,n
}
M cT.: ) system fro tne '!F1.' system.
(' lote inat this represen s a s :all e r cer cf valves in series than in the AFn' systens na; experien:ed bachleakage.)
3 tI T-i: valve arrangement is also true for the P.ea:00r Core !s:lati:r, C::,iir.g i
(20:0) sys tem.
Events have been reported at 3',lRs involving tne :::Lieckage fron :ne GF..' and reactor coolant systems to tne ri?CI and P.C systens.
inis Fl
- dy did net attempt to evaluate the safety iaplications of b :kleakage to 1
l am.
~ - - - - +. -
r l
F However,' a separata AE00 effort will review the operating these systems.
experience and safety implications of backleakage in BWR systems.
y i-3.2 Safety Significance 1- -
~
[
system is that it represents a potential common cause failure that could -
render both trains of the AFW system inoperable.
Some plants are more vulnerable than others depending on the piping configuration and layout, the t.
b number of pumps, the number and type of isolation valves, the normal operating L
position of the valves, and the maintenance and surveillance practices in f'
effect.
The events involving single AFW trains, particularly th.e recurring
{
events at H. B. Robinson, should not only be considered random failures of
.' sir.';'is AFW trains, but as contributing events which portend the potential loss of AFW capability due to a common cause failure..The loss of a single train by itself is significant because its failure may not be detected
+
iI[
ur.:11 it is required to operate.
This jeopardizes the capability of the AFW-i t
syste, to meet single failure criterion.
That is, the margin inherent in the design of the system to meet single failure criterion is reduced due to the N
h potential degradation of the remaining single check valve to isolate the two
- rtir.s cf the AFW system, i.e., common cause failure.
I g
Since 1931, the 22 reported events involving backleakage to the AFW system f
represent about 60 check valve failures to prevent leakage.
In 1983, there f
were *4 ever.ts that rendered an AFW train inoperatie (only six events were
}
ccenteo at Farley Units 1 and 2 although every pump was affected more than or.e time).
Thirteen of these events occurred at operating Westinghouse
[
pl ants.
i I
t 6
r q.
AE C assessed the safety significance of the less of the AFW system due to.'
.i.-'
stea?. binding of the pump using a risk-based approach.
The accident I
- :::r ;
t sequence considered is the loss:of the steam conversion system lafter a '
trar.sient event other -han loss of offsite power.
This sequence (T:*Qis-e
-.s
.e a dominate contributor to risk based on~a probabilistic risk assessment for-i.,
s-~
7._
n the Segeoyah plan: and results. in a category PWR-3 release (Ref.19). The 'f aul t i
trees for the AFW system do not 1nclude steam binding as a separate failure'
~
~
F s:
- ~
l c:de for the AFW pumps.
- c,.-
s p,.
-... -. 4,-
Tne loss of both the MFW and the AFW systems is a severe accident sequenceL F
tha terminates all feedwater flow to the steam generators.
Without feedwater, the steam generator secondary side ' boils dry, resulting in the loss of 'the
. ber.: sir.k to remove energy from the reactor coolant system (ROS). The R 5
- ressure will then increase, causing the safety and relief valves to cpen.
The R 5 inventory will be lost through the valves, which require tne opera:icn r
f
- ,f ESS systems for makeup in order to avoid core uncovery and eventual ::re
'4 t
cel t.
1 Tne feed anc bised mode of _ decay heat removal is an alternate method cf pr:vidir.g adequate : ore cooling when the AFW system is unavailable o provide e.cr;tr.cf f acs.e ter to the steam generators.
Altncugh scme other srecabi-
- istl: risk assessmer.ts incorporate this mode of decay neat re.n: val, tne l
- e:as;1n analyses (P.ef.19) did not take credit for it.
Due to the design of equi-.ent and human factors considerations, credit for this raode of heat g
t re.:: val is,;iant specific, and must be evaluated on a case by case basis.
y
(
~
t
~~.---I-
I or -.
e.
For ',lestinghouse-designed plants, the steam generator dryout times range from ap.prcximately 13 to 40 minutes.
In the event the AFW pumps are steam bound, i
n the c;erator must identify that this is the failure mode of the AFW system, stop the pumps before pemanent damage occurs, and restore,their function in
- f......
order to interrupt this sequence.
Unless the operator immediately recognizes that the pumps are steam bound and recovery actions (which must be perfomed, locally at the pumps and coordinated with the control room) are timely and successful, the likelihood of preventing steam generator dryout is small, L
Boiling the steam generators dry does not always result in core melt scenarios, f
e.g., feed and bleed may be an alternate method of removing core decay heat at some plants.
~
i
.The prc:abilistic risk assessments for some plants sho'w that successful AFW system operation (sufficient to remo e decay heat after shutdown) requires the flow equivalent to one pump to one steam generator.
Hence, the flow from a ene-nalf capacity p.:mp may be sufficient to prevent steam generator dryou based on best estimate analyses.
However, the expected increase in the r.
relia:ility of the AFW system, assuming successful operation with only one R
pump, may be reduced by the potential common mode failure contribution in tete --ining the overall reliability of the system.
1 Using a risk-based approach for detemining safety importance, the unavailatility of an AFW system containing three pumps is calculated based on the operating experience for PWP.s for 1983 ~(a conservative approach since most events were re;cr se in 1983). First, the unavailability of one or more AFW pumps due to l
_3 steam cinding is about 7x10 / demand (13 ever.cs at 47 operating plants each i
i
i g.,
- a, r
- U *
.:n
.~
wi-h 3 A W pumps suoject to 15 demands per reactor year (RY) based en
~
kh sirr.eill'ence tests and three AFW challenges after reactor trips). Secondly, the conditicnal failure probabilitf for a second pump to fail due to steam binding is
^
0.23 (3.cf the 13 svents involved two peps). 'For this calculation, a pump 'is
- conservatively considered to'be steam bound when hot water is detected at tne peo, i.e., the hot water flashes to steam when the pump starts and binds the
- c p.
Two of the events involved this condition; the third ever.t involved actual steam bincing of two pumps. 'The probability.of a third pir:p bec ming L
'stec: bcund is assumed to be 0.1 based on the common cause dependency for the l
hard,;are Set.:een trains having the same design and subject to the same I
e nvi rc.:.e r.:.
Comoining tne f ailure probabilities for the three pumps, tne i
r 4
una"ti;tofif ty cf the AFW system is about 1.5x10 / demand.
Fcr designs with i
two ASW pu:ps, the unavailability is increased by 50%.
Thc' ntvafiability of the AFW system for the Sequoyah plar.t withcut loss F
-5 g
- f 1: ;: ter (1c.: ur. availability for the onsite power) is tocut 4x10 per ;
(
d e. t '.d.
Tneref:re, the core-melt probability censicering the s eam binding 6
- f :.c ;%* p.; ps fcr tr.e TML sequence is increased frca about 2.5x10 /KY.::
]
-5 a b:v: ". *./.10
/F.Y (an increase by a factor of four).
Tnis is cbtainec by e r:'.; ne..av:ii aoili y of the AFW syste.-. due to s ea. bindin; :: the
-Igu:yan value and using the probabilities contained in th's Secuoyah. analysis -
4
-2 ic-trtnsients (7/RY) and loss of the power conversicn system (10 /denano).
T515 heresse doubles the contribution of the TML sequence to the prcbability
- f 3 ca
- epory pWP.-3 release, which is already the most probable release ca ege y it Sequeyan.
For a PWR-3 release, this would represen a rish h'I increase of about 45 man-rems /RY based on a dose calculation for a PWR and i
{
typical mid aestern.':'eteorology (NUREG/CR-2S03).
Using this technique, the i
A
.-~_ - -.
r h
I estimcted risk increase is about 60 thousand man-rems for the remaining lifetime of all operating Pbs (47 units with an average remaining life of 27
~
years)7 7.-
~
.s.
3 e t
I' These estimates are based on known operating. experience involving backleakage I
h
- to t e AFW system and assume's.that the events are unrelated and independent.
Although the events may not be. clearly distinct (this was one reason the i
- number of events was limited at Farley), this conservatism is believed to be i
tempered by using only the number of reported events in gaining a risk L
perspective associated with steam binding. As disqus' sed previously, reported operating experience may not accurately reflect the number or frequency of steam binding events, or the number of pumps that are affected, because I
backleakage is not by itself a reportable event.
In addition, the reasons are not clear for the absence of steam binding events at CE and B&W operating plants since the AFW designs are very similar to the Westingh'ouse designs.
Ccnsei;uently, the s.all population of steam binding events are not sufficient I
to predict future occurrences with certainty, and the risk could be higher 5 g
than indicated by the point estimates based on the reported operating experience.
Cn the other hand, there may be plant specific features that make some plants s':
less susceptible to steam binding than 6ther plants.
Thus, there is some l
uncertainty associated with the estimates, but tney still p,rovide some perspective on the safety implications associated with steam binding.
l 4,
.s The le,ssons learned from the. evaluation of the operating experience for reactor trip breakers after the Salem anticipated transient without scram j
( A3:5) events should not be forgotten in assessing the significance of available Il operating experience for steam binding events.
One of the important lessons i
1 was that routine statistical analyses of single failures and failure rate data i
'h
---.. -.. E __
I D_
anno: :y itself predict potential comacn code failures, even.enen a reldtively i
1ar;e :: uia:icn exis:s (as in.tne. case of trip breakers), as cor.4ared to :ne
. ~
- aucity cf availaole steam binding data.
.,,.: gr e
i Stailar to ne observed pattern for reactor trip breaker fail'ures,,the ccerating ex;ierience for steam binding events shoves that asmall nizaber of
' ants are appdently experiencing difficulties With the che:r. va5ves failing to
- event backleakage.
Thus, the random nature and low frequency Of steaa g
L singing events should ce regarded as potentially important Edety 'pe blens.
~
Eut like reactor trip breaker events, the licensee reports fati.t: conne::
[
re: causes (wnen identified) with common code failure potential.
Thus, a
.ajcr c:= en mode failure may exist that may not be fully re:ognized by 6
91:i.:se: and eviden:ed by their c;erating er.;crien:e.
Furthemore,
'a eperational ca: ability of the che:k valves to perfom their isolation fun: tion is a;;arently re:siving less licensea attention than did tne rea::ce trio
- retaers cefore ne Sale 1 A7.l5 events, e.g.,
esting and nain:ense:e 'ig e
5e::icn 5).
. : sever, one important difference oetween the ooera:1.; u:erien:e f:-.Me
. :. evens is tr.at a precursor even; er.ists for sten.:in:f r.; e.en.:
.o s.::c;r: -r.e icentified potential conmon icde failu e of :ne A..-l sys em.
(
_.. s.:.Stry, stet sincing of :ne AF',l pumps represents a potenticily si.:nifi:an:
- }
. s c.fi:/ issue.
5 ten oinding of a pump (s) is :resently an unde e: a le failure f
I
,i
- .a Oc.1d resul; in tne coamon node failure of tne AFW system.
ur-iier, tne T
.na: :ne unavalia ility of the AFW systea due to tnis failure a:Je :entrisutes f,
signifi:antly to risc of core melt in P JP.s.
h I
- 1
!h.*
--ew,..w-
r I-
~
.. L
- '3 CA'JSES FOR VALVE LEAXAGE
~
~
- n order for the hot main feedwater to reach an FW pump, the water must leak cas multiple valves in series. Operational experience showed that multiple t
caeck valves in series or in one case even two cneck valves and a closed actor-operated valve, have. leaked in a single AFU train and leakage has occurred in two trains, sanetimes simultaneously.
Hence, an unexpected nuader of valves are leaking concurrently. The purpose of this section is to evaluate the causes for the valve leakages..,,
f H. 3. Robinson experienced recurring leakage through the check valve (s) and ne closed' motor-operated valve in both the motor-and turbine-driven AFW t
One event involved both motor-driven pumps.
Tne trains at different times.
n icentified causes of check valve leakage have included a burr on the hinge, a l
oin hole in the seal weld, leakage by the seat assemoly, a.3d sics closing.
The eneck valves (4-inch Crane, Model 973, drawing NY 434112-5379-3C5),cere a
14k
- res'. aced with uni s of the same design, and leakage nas continued to cc:us.
_c Tr.'e licensee has now decided to replace the check valves witn a different
- esign in the near future.
The..stor-c;erated valves at Robinson were replaced in 1979 witn a different valve design.
Backleakage tnrough the check valve acparently caused tneraal t
i :ing of, tne originally installed motor-operated valve, wnicn tne licensee 3elieves caused the valve to leak and fail to open on several occasions.
I
.-:.ccver, leakage has also occurred thrcugh the replace.nent val _ves.
After the I
j
.r:ine-driven pump was steam-bound on July 21, 1933, the licensee re. corr.ec so of tne three motor-operated valves in the steam powered train.
A pin nele i
l
[____._
I
. n I v
l was geld repaired on one valve and the seats in the other ~ valves stare la: ped to
.ensere a g:od. seal.-
y, q, c.,: ;;
L i- :
Jr.tti tne check valves are replaced, H. B. Robinson.nas caoe changes arid additions to the procedures to minimize and detect backleakage in the AF4 2
e system. First, the AF'd pumps are now vented each shift (the initial time intervali Gas four hours before corrective action on the valves) to prevent a' temperature p
increase sufficient to cause steam binding of the pumps.
55condly, the
": 1
- i:;.-
(
prucsure for shutting down the APA pumps has been changed to delay : losing
-l tne notor-operated valves until the che:k valves have had time to seat properly.
F
~
to 174 pump trips or ba:kleakage have been reported since these pro:edural c
a.
- har.ges were.3ade in July 1983.
r w'
Jhe preventive action.taken by Robinson suggests that tne : heck valves were g
L not seating preperly due to an inadequate pressure differential acr:ss ne F-v 6.l ve s.
For check valves in serir.s, it is'not clear how all tne :ne:s valves j
5 g
, can seat properly unless.all the valves.close at tne excet saae time, s.nich.
2:: cars unlikely.
As a result, the availaole numoer ci che:h valves in li series to isolate,the AF'A system may be misleading ce:ause only one caeck j
valve sy te effe::ively preventing the ca:hleahage due to the differeatial i
oressure available to seat valves in series.
This nype:nesis is sup:ortes my
+
tr.e Lery evaluation of check valve leakages wnere only :nrae of tne :..eive s
[
1eaaing check valves showed any damage and the reasons for the o:ner valves l
' ~
t icehing could not se deter oned. The danaged valves ware ones located ci:sest
- , ne 77.' pi ping.
t I
i I
e t.
At Surry power Station Unit 2 in December 1983, four check valve's in each train leaked and each of the two pumps became steam bound.
The motor-operated valves are normally open.
A single check valve near the discharge of the remaining pump prevented it...from also becoming steam b'ound.
All pumps share t
a common discharge header.
Because of previous external leakage problems, t
the check valves were replaced with units of the same type (3 and 6-inch Cra6e, Model 175.5X, Dwg. B-363-534) during the December 1981 refueling outage.
An evaluation of the valves' internals revealed that tihe check valves nearest the MFW piping had steam cuts on their s,!ats caused by the flashing of the hot water as it leaked by the valve.
The other check valves did not t
show any visible damage.
All valves were refurbished.
Hence, the reasons fd the failures of these latter check valves to seat properly are not kr.o cn.
i Even after the check valves were repaired in December 1983, i.eakage was again identified in January 1984.
The 3-inch valves were removed from the system and sent to tne vendor for refurbishment.
The 6-inch valves were reworked by ed i
h the licensee. The causes for some of the check valves to leak were excessive
.cVement in the valve disk and small holes in the valve seat.
The plant L
h
- rocedures now require frequent checking of the AFW system for elevated l
- s9;eratures by using a hand held pyrometer during operator rounds.
In addition, the valves will be tested for leak integrity during future refueling I
cutages.
1 The cnly check valve failure reported by the Joseph M. Farley Plant was caused b
sy one missing and one worn hinge pin bushing.
Tne check valve failed to
~
t:
)
cicse after surveillance testing. However, the three upstream valves were t
- r a
'I
.k
I 37 -
- kncwn to be leaking before the test.
As a result, gross cEckleakage caused the relief valve at th5 ~AFW pump suction to'open.
The causes for the other check valves to leak could not be determined.~ One possible reescn being,
, evtluated by the licensee and valve manufacturer is the valves ( Anchor-Darling, 4-in:n, Model 900) are no't suitable for preventing backleakage to the AFW -
t i
system, e.g., large differential pressures are required to close the valve./
In January 1984, Farley initiated design changes to replace the auxiliary fee 6: ster check valves ( Anchor-Darling) with a different valve design.
I
~
However, after conversations with Anchor-Darling and with Sechtel Power p
g
- rp: ration (the A/E for this system), Farley management de:ided not to E
re;: lace the velves but rather to modify the existing valves.
The li:enses has m:dified the AFW check valves in both Farley units. This modification consists of adding additional weight to the backside of tne che:k vaive discs to ensure proper seating of :ne dis:s against tne ba:.: res-sure in the system.
In addition, design changes were initiated in Januarf
";21 to inst:ll tem;erature monitoring with annunciators lo: ally ar.d in.he c:n:rol roca on the, auxiliary feedwater systems.
The local annunciation m diff:stien is currently in planning for implerer.tation.
The con r:1 cm annunciation modification is currently scheduled for the next refueling i
f cu sg2 :n each unit.
Farley continues to monitor the auxiliary fee 6 tater systems to detect backleakage which may occur.
I.c:n Surry and Farley p1 ants indicated that they were considering replacing L
- ne check valves in tne AFW system (a second time for Surry).
Ironically, ne i
h I
n
.-+w, e.w..e-
-e ne o-
-238 -
replacement valves under consideration by each plant were the valves with which L
the other plant was experiencing problems, i.e., Surry was evaluating the_..
i replacement of the Crane valve 5 with Anchor-Darling valves, while Farley was evaluating just the opposite._ As a result of our discussions and suggestions, '
q
'the plants coordinated wiWeach other.to resolve the backleakage prob 1c<n.-
The cause for the backleakage through the two check valves at D. C. Cook, '/
Unit 2, was identified as incorrect assembly of the check valve internals.
The corrective action was to assemble the check vaives properly and to
~
i hand-check the temperature of the'AFW system during routine shift rounds by the operators.
The motor-operated valves at Cook are nonna11y open.
^
u:
Etekleakage through a check valve at Crystal River Unit 3, was identified
[
indirectly because the water heated by the steam increased the pipe temperature which adversely affected the AFW flow indicator.
Although it' was certain that g
at le'ast one check valve leaked, the licensee did not check for leakage of k
other check valves at the time of the events.
The plant had experienced i
numercus failures of this flow instrumentation, but only two of the reported i
q events identified backleakt.ge as the root cause.. The latest event (Ref.12).
h identified steam in tiie piping which caused the indicator to fail due to high te.oerature.
The causes for the check valve leakage were not identified.
The L
che:k valve has been reworked and an additional engineering evaluation by the licensee will be perfonned to determine if additional corrective actions are-necessarf. Upon receiving erratic instrument indications, an AFX pump is run to put cool water into the AFW piping.
Backleakage to a pump has not been
- experienced.
i
r D
Tns William B. McGuire event-(Ref.10) did not involve leakage of tr,e check
. valve, but rather, slow closing of the check valve which peraitted dFX to flow s
7 t in:c the AF;f system, overpressurizing it.
To mitigate futuhe events, r.elief o
valves.were installed in the AFW pump suction piping.. liowever,.tNis action
. -~
aces nct address the concer,n for steam binding... Slow check valve response
~....
l.
cr failure to close represents another means which could cause failure off tne
~
AFW system due to steam binding of the AFil pu:sps.
~
The reason for the check valves leaking at the KRSK0' plant was not reported.
Reference 14 only indicated that the check valves were refurbished.-
F p
Tne plants that have experienced backleakage were not alsays successful F-ir. crecisely i::entifying the root cause for check v.aive lealiage.
- n gc.ner:1,
~
~
saluati'ons are still. underway by the affected plants to identify and correct l-L ct.eck valve leckage proolems.
4
.g
..T.cre a;;pects to be no pattern or single :aajor cause f:,r eneck valva leagage.
Tne causes differ between the events discussed at the six plants waere leakage rr In.,est c.as cecarred, and involve different valve designs or.1anufacturers.
T cases, the check valves have experiitnced recurring leakage, even aftar rt; air an: repl ac e1cnt.
The causes for check valve lea;: ages.ci;11 continue :c se
~
~
~
evaluated.
-~
[
i 5.0 1.EAK DETECTION Existing ie;ulatory requirements were revie.ted to de:er.1tne iT there are
, any requirements for the check valves or re.aotely-operated valves to ac leak T
i
~,
- tested or unether conitoring the AFU system for valve leakage is part of
~
' existing surveil'1ance requirements contained in the technical specifications.
~
These issues'were discussed with' members of the Containment Systems Branch, e
the ;lechanical Engineering Branch, the Aux'iliary Systems Branch, and the p
~
~
Standardization and Special Projects Branch from the' Office of fluclear Reictor 2 '
Regulation. The discussions indicated that neither leak testing nor tempe#rature monitoring of the AFW system are required for the reasons discussed below.
It
~
Regulatory requirements to leak test valves are co'ntained in 10 CFR 50, Appendix J for containment leakage testing and in 10 CFR 50.55a, paragrapa g for the Inservice Tes-ing (IST) program. Leak testing is pri:harily required only for containment isolation valves.
For the valves tnat receive an automatic centai=ent. isolation signal, thit technical specifications require that the valves can be closed within a specified time interval.
~
~
k,Yg Al n:ugn a remotely-operated val've in the AP.! system piping is identified ps a J
con:airment isolation valve pursuant to G3C 57, the valve is not included in tne
\\ ;'
~
containment ledage testing pursuant to Appendix J because.the AF'.! piping is assumed to se filled with wa.ter, precluding air leakage.
As a result, the valve is not required to close automatically on a centain.1ent isolation j
signal.
The Appendix J 1eakage limits apply to the in ograted containment f
leakage rate and.not to specific valve leakage.
Thus, even if the valves were
[
incl.uded in Appendix J testing, they could be leaking but the total leakage of l
all valves could be below the allowable leakage rate for the cor.cainment, and e
..nus corrective action would not be required for any particular valve or valves.
i I
T e
t u
. 41 -
T' e IST pr: gram for valves includes thoseNalves designated as Class 1, 2, or *.
~
~
un:er Se:ition III of the 'ASME' Code andilhose function is required for safety, t
N a
7 ar.d also includes those valves not categorized as ASME Class 1, 2, er 3. but
~
The valve test procedures 'are prescrioec
~ - -
L
,thien are considered safety-related.
.c by Section XI of the 'ASAE Code and the type of testing depends on the category
- f ne valve as defir.ed by Regulatory Guide 1.25.
The AFU valves are ice.4tifiec p
.as Category C valves and the IST pfogram requires the safety function, of the p
valves to be verified.
U F:r the AF.i valves, the identified safety function of the AFW valves is to pen
-: ;r: vide a sterger.:y feedwater flow path to the steam generater.
- ien:e,
.e
- ST requirments ensure that the' valve disc opens freely. The AF. valves are,
[
therefore, net required to be leak tested as part of the IST program.
Ex:ancing the definition of the safety function ~of the AF'.! valves to in:iu:e a
p w isola-ion of tne AFW system from tne steam conversion sys s. to :reve{.-
k.
leu. age could result in defining them as Category A valves whicn wouls p.
~
r!o,ever, tTe - Me re:u're leak testing cf the valves in tne IST progra.i.
inkerval cetaeen tests (e.g., during refueling outages) wauld nct a:cear
!F 1
t
- :r:.ide an effe:tive method by itself to preven steam fe:-'ati:n i;
-.e :.F.1 system, es:ecially when small leakages are a con':ern.
Inis is n:-
i to say that inservice testing would not be effective as par of an :lerall
~
i
- re ra i.
For example, the combination of the IST progran and periodic
.l
- 3.;..tillance fer leakage during the interval between IST tests coul d.tinhi
- e f
tr.e li%elihs00 for steam fomation in the AF.l systee..
e i
e t
,,e
=
w w,m.
e e.-.3--
--~--m--
f 4
r Tne IST program could identify reverse leakage through individual valves and when corrected, minimize the potential for gross leakage to occur simultanecusly through all the valves in series.
By including these valves in the IST, program, tne leak testing could additionally ensure that each valve perfoms its intended function of preventing re' verse rotation of the pump impeller.
t Tne existing technical specifi_ cation requirements for the AFW system verify i
the capability of the pumps.and valves in the system to deliver energency i
feedwater to the steam generator. The surveillance requirements do not include monitoring the AFil fluid for elevated temperature to detect back-leakage frcn the steam conversion system.
The review of the yarious AFW r
I designs dio not identify any existing instrumentation that could ce used for Inis purpose.
At a small number of operating plants surveyed, the AFW piping and pump casings 1
are ::uched by an operator during his rcutine rounds of the plant to determine 24 t
if the piping temperature is hot.
This practice was limited to those plants that had previously experienced backleakage.
Typically, tne operater checks
?
the piping and pum,p casings each shift and checks more frequently when ele-vated temperatures are detected. Although tnis procedure has usually been effective at the affected plants, a pump oecame steam bound at accinson f
altnough tne pump was checked every four hours.
~
i i
The most effective method of reducing the potential for steam. binding of the i
AF'.! pumps is to centinuously monitor the AFil piping for elevated temperatures t
bet..een the pump discharge and the interface with the steam conversion system.
t t
e 7:r example, temperature instrumentation with an alarn in tne ::ntroi roca
.:Ould alert the operators that inleakage to the AF!! system nas c: curred such r.at ::rre:tive actions could be taken before the hot water reaches saturation r
t e
conditions and flashes to.. steam before or af ter the AFN pgap is started.
'f r
=
5.3 FI!!DI!!35 At:0 C0!!CLUS10!ls 1
The evaluation of the operating experience for leakage of hot dF'! into tne AF.! system found that 20 of the 22 events occurred at 'lestingneuse-designed
?
pl ants: thirteen events cccurred in 1983 at five plar[ts.
So:..e of the events, i
~
- articularly at Surry, Farley, and Robinson, indicated that backleakage
- ar. De a potential c::. con cause failure for tne AFil systaa. Altncugh the i.
- tner ever.ts affected single AFW trains, AE00 concludes that these events snould not only be considered randon failures of single AF'! trains, cut tiso as :entributing events that can lead to potential loss of EF.! capaoili y due
- s. a ::.nncn c a us e.
r
'I
' AI;; :elieves :nat the number of identified events is not a rue indica-fon e
- f isata;c r:: lens at operating plants because leakege into the AF'.! sys c.?
i is no, oy itself, a resortable event.
Thus,.cackleakage may be a acre
're:.en c::urrence than indicated by the oserating exrerience.
Tnis ::ck-i l
'.eakage is causing an unwarranted challenge to a safety.systen.
Tne eeneri:
[
safe y significance of tais leakage in tae AF.i system has asparently not seen I
fully recognized.
t I
AETs's assessmen of the safety significance of tne identiff eo events found r.a- (1) less of a sir.gle tr.ain due to steam ainding is significant ce:ause it is resently an undetectable failure that jeepordi:es the capasility of the AF.I I
r system to meet single failure critert'on, i.e., common mode failure and (2)
[
the unavailability of the AFW system due to steam binding contributes signifi-cantly to risk of core melt in PWRs.
The potential for backleakage,may be generic to other safety systems in both -
BWRs and PRRs because the stan.dby safety systems are isolated from the
/
. operating systems, which are at higher pressures and temperatures, by che:k '
i valves and a normally closed motor-operated valve.
However, there are no i
known reports of steam binding of the pumps in other' safety systems.
Operating F
experience shows, however, that check valve leakage can cause the motor-operated r
r valve to fail to open due to thermal binding (Robinson) or other reasons r
(Ref.18)--a safety concern different from steam binding.
n In Reference 18, r'-
' AEOD recommended measures to ensure the function of the motor-operated valves, which when implemented, should address this concern." In addition, the safety implications of check valve failures to open or leak in other zij safety systems will be further evaluated.
i F
The review of the AFW system designs for the three types of PWRs found that Ii the potential for bagkleakage is generic to all AFW designs because check valves isolate the AFW system from the steam conversion system in most I
operating pir.nts.
Some designs also enploy a normally closed renotely-operated I
valve in addition to check valves to isolate the interfacing systems..There may be plant specific features that make the AFW systems at some plants less su:estible to backleakage and steam binding than c'.her plants.
i~he AFW designs i
t k
i t
t i
i y
.45 -
at bestinghouse plants appeared more susceptible to backleakage tnan the o her cesigns because the remotely-operated. valve is nomally open in most Westinghouse plants.
Operating experience supports this conclusion although multiple events c curred at Robinson, which employs a.1omally closed motor-operated valve to '
is:1 ate the interfacing systems.
- . 7 ;, _7 - - ;
The study con:ludes that the potential for common mode hailure of the AFW /
systems due to steam binding of the pumps is present whenever one pump is I
stea.i bound because the pumps are connected by common piping (dis:harge header and/or recirculation piping) with only a single check valve to prevent backleakage of n:t vater to a second or third pump.
In addition, the capability of tnese h
check valves to prevent cross flow between pumps is uncertain because of a Ic.e g
l pressere differential a:rcss tne valves to ensure they are pre;erly stated.
l The 22 eve'its represent approximately 60 che:k valve failures to prevent i
-eve?se leakage. The analyses of the causes for check valve leakage dic.:t icentify any pattern or single major cause for eneck valve leakage.
Tne ca*uses i
differed between plants and involved different valve designs and manufacturers, id Tnis study did not identify any regulatory requirements or uniform plant
, ::: ice : red.::e the likelibcod of steam binding of the AFW pumps.
- r e s e nt'.y,
tr.ere are no regulatory requirements to leak test any of t'he valves isolating I
}
ne AFW system frca the MFW system as part of the containment leac rate testing or inservice testing programs to ensure the isolation function of the 3
l val ve s.
Existing technical specifications presently do not contain surveillar.:e 1
requirements to monitor or detect leakage into the AFW system.
A smail n coer of plants presently have g hoc procedures for the operator to touch the AFW w
o, e---
m.
r
-+e s-~*~--
~
4g _
piping to detect elevated temperatures during routine dif ft r5undsi ' This'
~
practice exists primarily at those plants that have experienced backleakage.
~
l
- -One of:the plants that experienced steam' binding of the AFW pumps has in-...
L
- stalled temperature instrumentation on the AFW piping"with indication in the control room. This is a co'minendable self-initiated monitoring method that we endorse for opefating plants, especially those which have experienced backle9kage.
The loss of the AFW system due to steam binding of the pumps is a potentia 11y' f
!sig'nificant safety issue requiring attention.
The 1pss.of the AFW systeni is 4
f a major contributor to dominant core melt accident sequences.
Although r
an Informatioit Notice was issued to a.lert licensees to the potential for L.
backleakage and steam. binding of the pumps, adequate measures to detect and moriitor backleakage.do not now exist in all plants to minimize the likelihood of the common mode failure of the AFW system.
L -
7.0 F'.E 07.ENDAT:0N
".J i"
- AE03 recommends that the Office of Nuclear Reactor Regulation either (1) re:uire the regular monitoring of the AFW system to detect leakage and ensure that the fluid conditions are well below saturaticn conditions; (2) confirm
-hat such a oractice is already being implemented; or (3) determine that backleakace is not a safety problem and no additional actions are necessary.
tr
{
The purpose of this recommendation is to minimize the potential for steam binding of the AFW pumps due to backleakage to the AFW system from the steam conversion system. The method shous o include two basic elements:
i first, preventive measures to ensure that the valves can perform their il 5
F
.em.
- =-,--c.-
e.s.
.-E.
.-_e
.....~
w...
t 47 -
~
in:enced func:fon; and second, surveillances to ensure that tne valves' isolating function has not degraded with time...
or example.-the first elemhnt could include leak testing of tne isolation
.c i
valves in the A:V system as. part of the operability requirement: for the syste:.
This testing could ensure that the val'ves can perfom their t
inte~ed function and be maintained in an operational condition required for safety equipment.
In addition, this testing could identify individual I
les or.g valves and minimize the potential for gross reverse 1eakage thro, ugh all vtives.
Leak testing could be required prior to a startup from an outage if testing had not been completed in the previous six months.
.40 ever, this element by itself is not considered to be fully acceptable, f
te::Use of tr.e long time interval between leckage tests.
~
g The se:ond element suggests a technical specification survei11ence require-L ter.:.n ranitor and detect backleakage during the leak tes interval as r
.; g For example, a part of :ne cperability require ents for the AFW systam.
3 L
emperature limit on the AFW fluid could be required as a Limiting Condition f:r C:eration.
In order to meet the Limiting Conditions for Operation, the
.e:perature of the fluid must be known.
The fluid temperature could be to tt. i r.E: either by (') installing instrumentation to continuously aonitor i
ne :c :erature near the discharge of the AFW pump wi n an alar. in ne con:roi room, or (2) measuring the temperature periccically using a hand-i held :y ometer.
The plant procedures should adequately address corrective i
i i.
actions to r,e taken in response to a high temperature condition.
The frequency I-h e
_,%,e eww->
- so we--wee
- + ~ ~ ~
~-
m
,w.w
-u.
- 3 in the latter case might be adjusted based on the history of measurement 7
results, i.e., increase the frequency if the temperature is frequently found
.high, or decrease the frequency'if the temperature is routinelyIfound acceptable.
~
Depending on the design and operation of the AFW system, there may be site specific provisions that pr'eclude backleakage and no additional licensee actions may be necessary. ~ - :e n e
's s.-
a In the interim, until an approveh method is implemented at operating PWRs,
~
~
k plant administrative procedures should require an dperator to measure the temperature of the AFW piping and pump casings with a pyrometer and record the reading in the check-off lists that are used during plant tours. Monitor-ing of the AFW piping. temperature should be completed after AFW system surveillance testing or whenever the AFW pumps are operated to ensure 'that the isolation valves are properly se'ated.
These' actions should ensure that backleakage is minimized and detected before a pump becomes steam bound, and reduce the likelihood for the 5
5 co. on mode failure of the AFW system.
~
y 9
l
(
L 4
a h
I y
I E
49 -
e-5.S REFERE:40E3 U... "e?. randun fro a J. Nel temes, J r., 'to R. DeYoun'g. and in 'Denton,.:.C, 5;:jec : Vacer Binding of Auxiliary Feedwater Pumps,' dated
.:vemoer 21, 1933.*
,, s
\\
~2. 2Carclina Power' and Li ht Company, Licensee Event Report 33-4, 3
D::ks-50-251, dated, lay 18, 1983.*
r-
~3.~
' Carolina P:wer~ and Cfght Company, Licensee Event Report 31-015,
.a r 7.
Jo:ket ilo. 50-261, dated July 10, 1981.*
eu
~
-> e w r
u
. - Carolina Power and Light Company,' Licenses Event Report 53-015, i
2cche t No. 50-251, dated, August 19, 1983.*
L 5.
Carolina Power and Light Company, Licensee Event Report 7.S-34, 7,ev.11, Cocket No. 50-251, dated March 25, 1982.*
e --
e
~.
Indiana and Michigan Electric Company,' Licensee Event Report SI-663, b
Occket No. 50-316, dated November 30, 1981.*
P b
7.
- r. diana and Mi
- higan Electric Company, Licensee Event Rep;rt 31-532,
- chet !. 5'J-315, dated August 11, 1931.*
F-
.3.
Fersonal ::mmunicati:n with Eric S.:anson, Senior.?.esident Inspb: :r, O. C. Co:k ;iu: lear Power Plant.
Event similar to Referer.ces 4 anu 5 i
c::urred on January 16, 1983 but the plant was,in an operational co:e whica did not recuire a report froa sne licensee.
r
?;%e P:wsr
- any, Li:ensee Event Report 01-135, Oc ke- :0. 5 0-3 d ?*,
a ed September 6, 1981.*
, [4[
t 7
- u.
Florida Power Corporation, Licensee Event Report 33-43, soc %e. umoer 50-302, dated riovember 2,1983.*
l
- 1.
~1erida P:wer Corporation, Licensee Event Report J2-J75, 0 cket,;u.nber b.
50-3C2, da ed, January,19, 1983.*
- ~.
lf r;ir.is Electric & Power Co:apany, Licensee Ivent.P.epert L3-55,
- c
- net :.uncer 53-231, dated Decer. der 16, 1933.-
Le-ter from T. 2. Trama, Cocoonwealth Edison 0:n:any, :: !!. Centon, ~;20, Su je:t: by~ren Station Units 1 and 2, :later kr.iner ?r: 3:-icn for
[
Oc:4e ts 50-454, 50-455, 50-456, and 50-547, d:ted Septe :er 9,1902.*
k
- Carolina Power and Light Company, Licensee Event Report 17-13, I
J::r.et i!o. 50-251, cated Septenber 6,1977.'
l,I T
Availaole in ::RC PDR for inspection and copying for a fae.
I I
l 1
-z.
L,,
15.
Memorandum from D. Zukor to C. Michelson, NRC,
Subject:
Engineering Evaluation Report on McGuire Overpressurization Event 'of August 25, 1981, AE0D/E248, dated November 2,1982.*
~15.
Arkansas Power and Light Company, Licensee Event Report 80-18, Docket
~
Kucer 50-368, dated.May 5,1980.*
i 17.
AlabaT,a Power Company, Licensee Event Report 83-84 Docket Number 50-348, j
dated December 27, 1983.*
.....~
i 18.
Memorandum from C. Michelson to Multiple Addressees. NRC,
Subject:
Survey of Valve Operator-Related Events Occurring During 1978, 1979,
/
and 1980, AEOD/C203, dated May 28, 1982.*
19.
i U.S. Nuclear Regulatory Commission, " Reactor Safety Study Methodolo y Applications Program: Sequoyah #1 PWR Power Plant,." NUP,EG/CR-1659, ol. 1, L
dated April 1981.*
F:
t t
n.
i
( ^r ed 5
n
.]
Avaiiaole in NRC PDR for inspection and copying for a fee.
=
~[
' APPE!iDIX A I
APOR sit!DI:lG OF A'JXILI ARY FEE 3k'ATER PU 'PS
- ;te e
m g
MD wD e
gir
% W 9
~
s.
t J
i e
g 9
e
(.
4 w,
I um P
b ?'.
.m e
M B
l g.
F O
C
+
D I
L hN r.
- 4 9
9 0
l I
i 5
0 1:
~
e h
e esse e+vano
L A !
g.o %s.e,*
e e
L UNITED ST ATEs
! (y g 7 ;,
NUCL EAR REGUL ATORY COMMISSION 7-e3 ic wasa me1ow. o. c. :m
,, s. "..., }#
c NDY 211333 IM3RAx3UM FOR:
Richard C.' DeYoung, Director Office of Inspection and Enforcement-Harold. R. Denton, Director
_ Office of Nuclear Reactor Regulation FRC#.:
C. J. He1tmes, Jr., Director Office for Analysis and Evaluation of OperatTonal Data L
SUFJECT:
YAPOR 51,WDING 0,7 AUXILIARY FEEDVATEP, PW.P5 r
Fo Enciosed is an engineering evaluation report on the vapor binding of :ne L
au:t'.ia y feedsater ( AFV) pumps at H. E. Robinson Wu: lear Power Flan:,
Ur.i: 2.
The safety ir. plication of the events at Robinson is that the lettage of main feed <ater to the AFV systs constitutes a etc.cr cause f aiiure that can render both trains of tha AFV syste inoperable, ai-trwough only single trains have' been adversely affe:ted to date.
Si.ilar events have also o :urred at D. C. Cook, Unit 2.
Tne p tential for the loss of AFV syste due *,: backieakage appears generi:
\\
- e:ause the designs of tne systes a Retinsta '.nd Cott are typi:t1 ef 4
c.ner ?k'Rs, i.e., i solation between the stea, :snversion syste a,: ge ATV syste is accom.plished by check valves and rctor-operated valves.
1.E03 bas initiated a case study to better define One generic ir.;iications and esublish the bases for revising the. technical specifications :: ensure na-
..e AFh' tapera,ure is monitored an:/or na
- ne inservi:e ins ec:1:n r
1 prog ans test th; isciation capability of the eneck ialves.
r I
g
':r :.e interi:n ur,til the case study is ccepleted, the Of fice of ;-spe: ien a.: Enfor: pent is re:; vested to issue an :nic aticn betite to ;r:c::1y
)
.:,ify PiR licensees of these events and aler: :nr t: :ne :s:te-- t al f or i
- sekage fr:rn the fetesater systs to the ATV sy!!ec and s tem Oincts; cf
, ee AFi ptmps.
I f
Tne. Cffice of Ku: lear Reactor Regulation is provided a cepy of the rep:r:
a this tir,e to highlich: the significan:e cf the even s a?i:' pr: vide an ir.:.u: f r.:o on..in; WRR a: tivi ties.
We nave only e:ently ber re ava t cf I
Ti' E2-55 entitled, *Rebinson/C ystal River I - A V Cne:c ialve '.eauge,'
anc enderse the actio.n to evaluate generic te:hnicai spe:ifica ic cnanges.
i 1
is icportan to note that R:binson events analy:ec in the en: lese: reper have cccurred since the TIA was initiated in 1952.
These events r.ay pec-ice additional inforr.ation pertinent to the resolu icn of tne TI A.
[
^
g3 o i
f A.-3
..-r--
If y^ujet ire anv a:ditional inforr.ation, ;: lease conta :
- 3y e tanrin: 3.,
ne is cva11able to assist you in re;;1ving this it.p: rte.r.: i ssue.
R 7Ag mLg r., Dire::cr -
C'b.Heiteries, C,
e for Analysis and E,aluhtion of Operatio.a1 Data
.... n. s.. a e.s a..,... s.
/e,c1:sure:
I usisn w. ski, r.ecion II G'-:el ahan, NRR
.W g
O p-p
- ss..,
??
sl^
p.
e e
l c
t b
?
6 I
i t,
b e
f l
4
~~_.
B r,
A-4
.l AEDD ENGINEECING EVALUATION REPORT' Uk:T:
H. B. Robinson, Unit No. 2 0;;r.ET:
50-251 EE REPORT NO.:AE03/E22E OATE:
NovemLer 21 l'E3
';;ENIEE:
Carolina Power & Light Company Ei A' UATOR/CONTA T:
- k..' annin; NJSI/AE:
Vestin; house /Ebasco e
525.;E;T:
YA?DR BINDING OF AUXILIARY FEEDWATER PU"?S AT ROEINSON, UN:.T 2 t
EVENT DATE:
April 19, 1953-I
REFERENCE:
Carolina Power & Light Company, Licensee
{
Event Report E3,0?4, Docket 50-251, cated p
May 1E, 1932.
IU "..i.2. Y L
R:b'nson has experienced 4 failures of AFW pumps due to low dischar;e pressure
- -i
- s caused by s, team formation in the AFW piping and pump casings.
Tne steta was f:rmed when hot water from the feedwater system leaked through tw:
ne:t vain ard a motor-operated valve in the piping to either the motor,- cr stear-riven AFV pumpr.
Althouch the backleakage has' caused only a single train of :ne AFV sjstac to fail, the potential exists for both trains to fail simul anece:iy sin:e backleakege has occurred repetitively in be:h trains.
inree e.en : have
(..
~
AFW piping and pump casing.
a '. 5 : s::urrec a: Cook-2 involving backleaka;e and elevatec tem: era ure of :ne Tne evaluation concludes that Robinson has implemen ed acceptable c:ere:-ive a:-itns to orevent steam formation in the AFV system.
Sin:e the cer';
- ' -he
- '.::n AFV system is typical of other operatin; ?'-7.s, an E lr.ferma-i;- N:: ice pak r:ui: be issued to inform other licensees of.the p::ential for staam fn:ing
[
Of tr,e AFV system.
t 1.n 'E 3 case study 's recomended to further evaluate -he generi,: ir:'. i:1-i on s f:r c:ner A. V systems and develop appropriate re:ommen:ations to cir.imi:e the
- e--tal for steag binding of the system.
Generi: te:nnical spe:ifica-ion L'
- .a. ;es sn:uld be evaluated to re:vire ina: appropria e su vei'.lan:t e o:e:eres l
+:e # :' emen ed, if net alreacy available, o dete:: leata;e an: ; e.en s eam
- rr.a-ion in :ne AFV system.
t t
ti ty()I i
r $(a f"* : a : co:ur.en: supper:s ong ing AEOD and WRC ac;ivities and coes n:: re; esent b
ne p:sition or requirements of the respensitie NF.: prograc effice.
l
I Q-o s
1-I
.........n..,.._.
. =....-,.
=..
s t
vrin; the review of operating experience,.the. referenced ' ER ses icen-ifie:
...as.a_:i;..ificant event ind-warran ed AEOL evaluation.be:ause cf the a: tr.-ial F
f:r ::mr.. r.:ce failure of tne auxiliary feed.'a:er'( AFW) system.
ine purs:se
- f' nis en;ineerin5 evaluatio0 is to summarize the event, evaluate the saft v i
impli:ations. and ceterm.ine' whether adcitional licensee er NR; acti:ns are
. u. s..,. t...v..
- a. ~
.n
.. Foil esi ng a manus'1 reactor:: trip on Apri1 ^19,.1983, the two motor _ driven.
L auxiliary feedsater_ pumps started automatically on low steam genera :r.
. 1: vel.. Af ter acout 2 minutesi the "B" AFW pump tripped.
During esting.
~
I(
~
- f :ne :vmo, a si;nifi: ant amount of steam was vented from the sum; casty.;.
The p=:p trip was attributed q a prote:: ion trip signal generated by ne pressure instrumentation in response to a low discharge pres:ure.
Tne dis-harge piping from the motor-driven AFW train.is conne::ed to.the.
mr'- feedsa er piping near the steam generat:r.
Hot wa ter ( abou; a25'F)
- f. am -he feed ster systen leaked through :s0 ca.eck va;ver end a motor-c;erated I
valve 'n ne :iping to the AJW pumps.
This water flashed in tne cis:ne ;e F
.. :';i ; and puma casing because the AFW system was at a lower. pressure : nan
- 'nen the AFW pumps s arted, the i..strumenta-ion ne f-e:sa ter sys tem.
a
..,.'e ::s:harge piping sensed a icw pressure and signa..
ir -
isto a pu:; ris.
3...= wa- -a'sa.d.
A..v e..=..>..
N.'.."..a.-..
a.'.
... a.
4......
s....
-e:::e: :ne fi c.
and preven'.e: the dis:narge pressure frst, in:reastn; a::ve l
re pressure setpsi.n in tne 30 seten:s re:u're: f:r :ne dis:narge c
p e::;re sens;r to. time out.
.ondensat;en e:Te: s woui,. :On:r,.rute to tre i
1:s.:ressure concition.
p.
.- i L: e.-ial exi sts for both co or-crive r AFW ;um:s
- tri; :se
- : a:k*. s ei-
'.-ir.-
.".-.o.-*. s.=.-
- 5.o-*
.*..-4.-*..,=..r*
c,. c.
.g. w".s.-ha.r-a. g e
-g
,.y g
.2 o
- r, zum;s snare a cunmon cischarge header.
This ~ is eviden:e: :
- ne r
e.evate:
erperature measursc.ents obtainec f:r bc n pump ss'n;s 0;rin; 5e L
e.
.. 2 3... e....
e
... s... s. s j,. y. -. e..,. e. s, c.
- g..v....
. s...... s.
...s...
df >:.-t ge pipin; and wa s n:: affected cire:-ly.
H:veve,
- e m: o -c n en 1
a: 1 e ar.-: riven cum:s share a car.non sv:-icn he ader f GT. the c:n=ensa e 5 :rt;e ank and be'ccleakage coulc affe:
211 pum:
5.
.n. =.. "... s h a v a.
a c.rm,. a n s ".. '.. r. ". a. a - a..,
.58. e '..'. '.
- a. '.....'
.- n..=...sa.3
......... g s.,... w~ i d.S. ud.o. i x si.n..a.
h...*.
s.i.5..
.i ~
m a. t. 2..=.
I
-:t.e 1 : s re:u:ing :ne po "n-ial for sa er a. :ne sa:-4 Of Se A7 : r::
- :.e ene saturation :enditicns and flest
.en :u de:.
in's soui.c ce':er:,
e f
- ' ::u-se, on :ne ieuttge r.ste an: the time availatie ::
ai se int te?.rern t -e
- f - e sa:-icn wa ter.
Eased on this event, ;be :y:Sint tce of r+se :-:
e......
- e.. rs-. a. v a.
.c a. l.v a.' '. e c s *i..*.r
.55 ss-o-- 7.-.....'-'.a A.....
..w
....s e.,vj....
- g..hs.
e..wq.w...,
g mo-.
- e. 4,..
.4 g J. 4..-...s. s..
... s. - e., e. r,..
w a.
r w.
re s' e: =coled.
t F.::'rs n had experien:ed prior leakage through ins ci>:narpe ;ip;r; cr.:
c:. se:sen-trips of :ne me :r-driven AFW : :p
- A" en June :: and 15, *;51 l
i. E. E *. - 0 *. i ).
ine ur.it wa s at 921 po-er cur'n; :ne se:en: tven # :h :n*.y a j
- t.;ie AFV pe:p rei.aining to prodde emerge n:y fee:sa tr oe:a.re -.e : ea -
c-ver pumo was incperat.ie.
The valves -ere e:a; red an: tr.e ta:t,etra;e i
s.p ni fic ar.-l y r e:u: ed.
Tne pump ri;pe: a;ain en June 15, ;'i; :.E 3;
- 7',
e.,. a. '. a c *."w "...'."g,
u. *. n. c. ba".se at. a... '.ve.'."..- i" p-
.--.*.a**
..--..._y'.-
e...
t s
j
f valve setting cf :ne discharge valve although steam bincing couic nave caused :ne low pressure trip.
On July 21',19EI, a similar event (LER 83-016) occurred resulting in the s eam-criven AFW pump being ce:1ered inoperable due to p ential stern bir. ding, d:
water from -he feedwater system leaked through tne dis:ha.ga che:k vaive and the meter-eperated valve producing steam at the suction vent antcis:harge c ain cf the pum;. Ine discharge piping from the steam-driven pump is conne::ed,
ne ee, water bypass piping.
.he.poten:1a.i existed ter pump cavi at,.cn anc i
trip on low dis:harge pressure following an automatic start.
The seeam was dis:evered during a routine.. check of the AFW train and the pump'had not been re uired to operate.
On August 17, 1977, tne steam-driven train experienced a f ailure of a eneck valve to c. lose which caused :ne relief vai es on the su:: ion header to lift.
i
~
C:her failures of the motor-operated valves occurred on September 5, 5 and 15, 1979 when they failed to cpen (LERs 79-32, 7c-23 and 79-34).
Tne cause for ne failures was due to a thermal overcurrent,rel.ay trip resulting from i
the failure cf :ne torque ssit:h o de-energize the motor af ter the valve was fully closed.
Excessive wear of the worm gear prevented pr:per operati:n
.L
- f :ne torque swit:h.
The excessive wear is believed to have occurrec during
- evious events when the valve stuck closed due to thermal bincing caused by
- ne leaking upstream check valve.
Thermal binding can lead to deforma-ion
- ? :ne valve internais and leakage.
The three eneck valves and -he three m::: -:: era et valves we e re;iated with :ne same tyoe (a-in:h ;rane,m::ei 572, crawing N:431*12-5279-305) in 1950 to cor e::
ne backleakage.
~
r The cesign of :ne AFW system at c her opera-ing piants also generally include
- ne:x valves a9: me:or-operated valves in series to prevent b!:kleakage d am :ne fee:sa t-system :c :ne AFW system.
Inis sugges s a po entiti I,
peneri: con:ern.
%:-ever, a revite cf 0:eratin; ex:erience ic- :ne pas 24L 1 years icen-ifiec Only three similar events.
These events o: urre: at i cck-2 (LEP.s 5*.-22, and 51-53) where the valves leake: in :ne sten.m-crive-r rain and an a:n=rmally high temperature was observec for the cump casing, sa::i:7 and di s:nar;e piping. The pump had n:- been re:uired ::
- e a:< in L
any even:.
Tne Resicent inspector identified :ne :niro event
.ica oc:urrec cn J anuary 6,1952 This event was not reported in an LER be:ause :ne m::e l
Of. Opera-icn cic n : require the AFW system
- be Operable.
t
- 1. n:::n the cesi:n of :ne
.. systee a Coo.t is simila.r to r.::inson, ne vai ve 'n :ne pump ci scharge pi:'n: is io:t.e -cren curinc
- -coera e:
5
- .e t-i on.
Tre is:*.ation of :ne ATW system ' rom ne main fee: a:e-sys em
.s a:nieve: sy o : neck valves (4-in:n A
- c ". rriii, c ra.i n; 7 2;:*.iF';.
The l
-eason for -ie :nect valve leakage is attributec to imprr,*er salve assee:1y
-at er than cesign oeficiencies reported f or ne Robinson valves.
Never:ne-
- + s, Ine :::e.-ial for ba:lltakege may be grea er a: Cook : nan Ko:inson,
- e:3.se ne m: :r-crerated salve coes ne: Orovice isolation :a atiit:y.
f
.c-e.er, :ne cor.s e:uen:es of ackieakage at Coot is significandy less,
- e:a.se :ne m: :r-criven pu :s ce not snare a :omm:n c15:r.ar;e nea:er, l
e.g., be-h pum;: cannet become steam bounc cue :: leakage in a singie 1
ci s:harge line.
All :ne pur.ps do share a cunm:n suction from :ne con'-
- entate st: rage ank. Like F.obinson, there ere no temperatu e inci:aters f:r :ne auxiiit y feecsater.
The events a-P.ctinson and Coek sugges:
na backleakage is a p::en-ial generic con:ern since cif feren ene:c vaive i
i p
.... -... - - -. ~
I
~
g
..3...
s.
< c.
4 cest;ns 'are employed in.the AFW-system and both units have experiencee.:
L sa:xiestage resulting in incperable trains of the AFW systemr 22
?? :.-
n :~
..y, r.~.
L A specia1{ interim pro:edure has been implemented at Retinson ::. vent b:tn
-he c::or-and s eam-criven pu=ps once-ea:n shif t.. In a:diti:n, :ne h
L reperature of the pump casings are monitored lo: ally and the pumps are
.e;arated.as ne:essary.to thsure that the water-in the ATV systs: remains ce:1 anc well below saturation _ conditions.
Cook also monit:rs the ee:trature curing' routine checks by.the auxfliary operator curing sp.if-
.-: inspe:: ions..i
..s.
ng= ;u -
~
- -.. ~.
In :ne longer term, Robinson is evaluating a design ena'nge' er repla:emen of ne check valves located in each of the AFW pump discharge piping.
Depending
.cn the results cf -he evaluation, the check v11ve leakage shs.ic be c: nee: ed.
during tne refueling outage btginning in Decemoer 1953 or during the i:eam generator replacement outage beginning in June 1984.
A pr:; ram is a*.se under-o
.ay :: improve the performance cf limitorcus valves by developing valve per-4 formance histories to monitor and iden:!iy valve. degradation in the future.
~F:WC*NGS AN3 CON LUSIONS w
Robinson has experienced four events in the pas: two years inv 1ving
.f ailures of AFW pumps oue to steam binding resuiting from ietkage of. feed-E water ::.the AFV system.
It apoears that :ne f ailure of ne cneck valve
- pre,en; sacifio- :auses the m: or-c;erated valve to leak and is :ne primiry cause for :ne events.
Eased en operating exoerien:e : e leakage in ene A?W train has not af fected the c:her train altnou;h tne p en-f a*.
exi sts f:r cann:n moce f ailure.
The primary concern is, n:--ever, nat Lack'.eaka;e vill occur simultaneously in ea:h of ne AFW trains causin; failu-e of -he AFW system o perform i s safety func:fon.
m
.W(
ine safety implica:icns cf the four events a: Robinsen is na: :ne ieaka:e cf fte:,a er to :ne ATW system constitutes a co ecn cause fatiure na;*c'n a
rer.:e-s::n rains of tne AFW system incoeratie.
Ai:neu;n :ne events
.s ta:e save irv;1<ed :ne f ailure f a sic;;e :-ain, ai*. Of :ne e.er s -ave 1
teen :ause of :ne simultaneous leakin; f
vo er th-ee is:11:!:n va;ses in i
series.
inese even s snould nc-be consicered ran:om f a",ures c# sin;ie Ar4 rair.s, but as c:ntributing ever.:s leacin; i: poter-tal 1:ss :f AFv ca: :iiity
- t
- a : - :n :euse f ailure.
ine ren: c f :ne s e ei e n s ::.: : re s '-i *. a-if
- :ne trend of the ren:::r trip creaker f ai*ures a Saie an: ::ner 2;an s ef:re tr.e Salem ATW5 events.
a
- '.:e :ne cesign cf :ne AFW sys em a R:: ins:n is :yoi:ai cf ::.er ru.s.
ne ;;;en-tal f:r backleakage exists in ; ner c:eratir; ;1aa s ev'dence: by
-.e events at Cook-2.
M: nit: ring of the ten:erature of ne AFi ;/:: : s s i r.;,
se:-ton and dis:narge pipin; sh uid be pe-f:rmed c a ro 13e s:ne: *.e ::
t
- t*.e:- leakage int: the AFW system an: preven: 1 tam din:'n; :' - e syste.
- c:insen nas icolemen:ed pro:ecures :: ensu-e that ne -a:er in :ne A / syste.
b ree.ains cool to prevent steam formation.
These prever.: ve a:: ions sa:ci: en.
sure nat the ATW pumps are available o perf:rm :neir safety fune-icr. until
- ne :he:h valves are redesignec or repla:ed :: c:rre:: :ne iea(a;e p-::*ie.
-- im;r:ve One perfcrman:e cf ne cic;;r-cpe ratec s t* *e: a-e a*.5: ence r-
-"-e ine licer.:ee's act' ens a: pear at:e; abie anc no a::i 'onal a::d:ns are
.ay.
Le'.'evec ne:essary a: :nis time'.
t
_ - -. =
- ~ ~..
j.
t,.
=
'i.'
t ; ~ ' : - - ~~
r-'
p.E tu.".EC ATI 3h5 ' D U>C-E
l-
~
' ~ '
i--
- a::
Tr.e Office of inspe: tion and Enforcement should consider issuin; an info:-.a-icn y
n:-4:e.: i nf: t. otner licensees of the potential for loss of AFh' capabili:y cue l
ta:(.eaka:e an:i stea:n formation in the AFW system.' -
a
- n -he near future, AEOD should complete a case stucy to evaluate the generic impli:atiens for all FWRs and identify and establish the bases for changes to I
te:nni:ai 'spe:i fic ations.
In addition, the require:7.ents to; include the AFh' e
~
g p -.: :is:harge tactor-operated and check valves in the inservice, testing pro-i t
. -. gr. a.ms sh:uld be evaluated.' :
4 -
.--,= ;
.. :a.
[ '[
,1 ' ' "'.,, 7
{'
- 7,* ;
l 2~
/S
, y
~
e P
p M
4, g
g g
gg p
e 4 h
, p.
b
- - y a *
- i, t
h 4
'+
'Y e
L-V.'.
O
[
m 4
f w
l' t
s P
i F.
l t
1 p
i.
.P s
~
D m
b a
l
. ~
4 i
L L-a _
.._