ML20214A728
| ML20214A728 | |
| Person / Time | |
|---|---|
| Issue date: | 08/02/1986 |
| From: | Catton I Advisory Committee on Reactor Safeguards |
| To: | Boehnert P Advisory Committee on Reactor Safeguards |
| Shared Package | |
| ML20214A645 | List: |
| References | |
| ACRS-CT-1854A, NUDOCS 8705190566 | |
| Download: ML20214A728 (7) | |
Text
%
2 '
ET-/9598 l "
TO:
,FROM:
Pcul B Chn:rt Ivan Catton 2 AuguOt 1996 !W 5//hgr, /
EUBJECT: EPRI Workshop on Water Hammer in Nuclear Power Plants, July 9-10, 1986, Boston,d>ssachusetts The Electrical Power Research Company sponsored a workshop on Water Hammer in Nuclear Power Plants. The workshop i"s attended by f' representatives of 25 utilities, several A&Es, 'wo of the NSSS vendors (Westinghouse and Combustion), and representatives from several engineering consulting firms. The NRC representation was minimal. The utility representatives expressed a need for guidance in dealing with water hammer. It was noted that water hammer is still one of the main culprits responsiblo f or plant Cquipment damage and loss of availability. No plant system seems to be immune with RHR systems being the most troublesome. Damage ranges from trivial to severe. Further, it is 4 commonly held view that only ten percent of the events are caccht by LERs.
Nevertheless, feo of the utilities present had designated personnel responsible for water hammer elimination or control.
The f ollowing paragraphs contain summaries of the various
' presentations made at the workshop.
Power, EFR!, gave an interesting and comprehensive overview of water hammer in LWRs. He noted that a significant number of water hammer Cvents have been reported during the past decade or so. The water i hammer events involved piping systems and components attached to
- them as well as piping system internals and components used to Cupport piping systems. Although most of the evonts have resulted in relatively mince damage, there have been several recent events I
, resulting in major component damage and equipment system loss of function. In spite of this, Power felt that Water Hammer had not '
yet had a safety impact.
Approximately thirty fluid systems in the four reactor designs have the potential for water hammer events and ten of these perform safety functions. A number of new challenges were giveng 1) It was noted that recent experience shows subsystems designed to prevent or mitigate water hammer becoming initiators
'Cf water hammer events. 11) The new operator actions dictated by severe accidents may lead to a circumstances favoring water hammer. 111) Off-normal plant configurations need more analysis to ferret out potential water hammer initiators. iv) Removal of restraints, pipe whip protection and seismic snubbers will place new emphasis on normal piping system load limits. Power made a ;
plea in summarizing his overviews we cannot let NRC say "thou
}
Chall not have water hammer" as water hammers will happen. }
- I n o The water hammer data base developed by EPRI was described by
'E
'O
- House 11T1). The data base has a more sound base than that of "
NRC derived solely from LERs. EPR!*used LERs, INPO data combined l5 cith plant visits to bring it all together. 1985 was found to ,, _
,,\4 ja cet a new record f or water hammers. This was probably due to new
}$@g cnd different kinds of testing required as well as the number of ! k 1
'oog plants in operation. If normalized, the number of events per plant
$"m year seem to be generally trending downward but not dramatically
$gy Co. The frequency seems to have asymptoted at 0.2/yr and 0.1/yr
,(i.\(l
.bok mau for PWRs and PWRs respectively. A close look will show, however, . >>
l that the frequency for PWRs seems to have hit a low in 1901 of '
J 0.05/yr which is then followed by a three fold increase over the .;
next four year period. The frequency looks relatively constant i
! for BWRs since 1977.
! , l'
N.
. . . N
.g.
Cf the se events occurring during the 1981-1985 period, 60% resulted in pipe support damage (hangers, anchors and unubbers), 17% resulted in component damage (piping, pumps, valves) and the remainder caused no damage. Ten percent of the events resulted in a reactor trip ,
(0.02 scrams por year per plant due to water hammer) pnd 7X 1ead to plant shutdown.,'The following tables summarize the results of the EPRI study of mate'r hammer.
Action Prior to How did we get int'o l the event the problem? ,
pump start
- 16 pro;edure 24 water valve cycled to component failure 12 cteam valve cycled 9 inadequate component 3 turbine start 8 operator error 3 unknown 15 unanticipated response 3 unknown 13 Pre-water hammer Mechanism causing ccnditions water hammer j __...________.__......_____ .. _________________________
Cteam in water line 17 steam bubble collapse B ccndensate in steam line 16 steam water entrapment 20 l
cir in water lino 10 void line fill 12 Oingle phase 6 water column separation 2
) water in steam line 3 classic water hammer 6 unknown 6 unknown 10 Overall cause of Corrective measures to water hammer events
- water hammer events cdministrative 2 procedure change 32 ccmponent 11 operator training B dcsign 21 modify supports 5 operator 12 modify controls 2
, procedure 23 modify piping 9 l unknown 11 unknown 10 o some events have multiple causes Water Hammer events summary SYSTEM DWRs PWRs Total RHR 0 5 13 Fcedwater 3 0 11 turbine steam supply 5 7 12 ECCS 5 3 8 M2in steam supply 3 -
1 4 PWR SGBD F 3 3 BWR RWCU 1 2 2 i
i S:rvice water 1 2 3 Other 4 l
58 I
Gr1ffith (MIT)
N3. .. .
R cocrch Progrca". govoHeo u2Cd p:ptr titled " Structuring O Water Hammer ;
,co examples. The Indian Point Indicn Paint, Millctona, and SONSS-1; feed ring water hammer was . ,4 pcrticularly interesting because it demonstrated how one should cddress such problems. A simple experiment was devised and a th; cry to explain it was developed. The results were used to :(
a; Cavelop a checkJ11st; is the pipe h irontal?
- s. '
is there a po un'tial for WH7 '
check metastable limits developed during the laboratory study.
Tha laboratory study yielded a 1ist of " fixes" (some of which may have been self evident to those familiar with plant operations);
reduce the subcooling, reduce the length of the horizontal run, ,
increase or decreene the pipe diameter, ,,
codify operations, ,
modify piping, cdd non-condensibles, tilt pipe to decrease interfacial area.
How the fixes should be carried out was only established for the grometry investigated, a horizontal run of pipe. The Millstone WH was similar tn the SONGS-1 incident. A leaky check valve allowed tho water in a line to drain down. The empty line has vertical as well as horizontal segments with Ts and elbows. The modeling done 4 for the horizontal line is only of limited usefulness here. One is
) 02 liquid Oft with assuring sensing thatinthe device theline always remains full or installing line.
fThoMillstoneWHmostlikelyresultedbecause the line fill rate
- wao too high leading to rapid line pressurization and resulting
,,high
. prcssur water subcooling. To understand how to avoid too rapid a
! peccess.ization, oneaneeds With such model,aonemodel of the could line c.ght repressurization consider certain
( p3tential fixes:
toggle pump on starting to restrict fill rate,
{ new starting box so startup is gradual,
- Cut throttling valve in series and open it slowly,
, pre-heat core spray water.
Th3 benefit of the model is being able to quantify the results of a dix or intervention strategy.
I Briffith gave an outline of an approach that seems very sensible
- 1. " Survey" the waterhammer events to get as good a description as possible of the accident sequences.
- 2. " Identify" the processou leading to water hammer.
- 3. " Describe" these processes mathematically. .
- 4. " Recommend" a variety of interventions.
6 ThJ first two steps in the recommended process have been done, in ,
p;rt, by the EPRI study.
Limited work has been done on the third Ctep and a number of eaamples of how intervention has been achieved Cxist in the industry. Unfortunately, the utilities do not publish th] results of f gheir ef f orts and many WH problems are resolved each time they arise.
A-Uffer (Quadrex Energy Services Co.) gave his views of " System C:nfiguration Considerations". His presentation was the common C:nne of WH preventions pump water uphill in a system without high
- points or vent them, pump steam down hill while minimizing low prints or drain them. It is unfortunate that plant piping systems (C
- olgnersdonotgivemuchconsiderationtowaterhammer.
It was o, interesting to note how Griffith's modeling of the horizontal line
{ wa3 able to quantify the cualitative observations made hv uffer. "
~
4p Brooks (EPRI) discussed the effect of check valve performance e'n water hammer. He described the various kinds of check valves and concluded that they better not leak. The SONGS-1 study of check volves is more inf ormative and is discussed below. ;
9:
- The "Open Discussion" following the formal presentations was very i n t erest i ng. In response to the moderators questions about utility cancern, of the 25 utilities present, 7-8 utilities are addressing water hammer 1 utility has an engineer assigned ,
,, 7-8 have operator training that addresses water hammer 3-4 have cheel. valve programs in place 4-5 lool on check valves as an active component.
One of the reascns for the lack of interest is the NRC pronouncement that water hammer is not a safety issue.
It was felt that the real cost of water hammer is loss of operations.
Some enamples given were the two Main Yankee events that resulted in oix weela loss of operation and the Pilgrim events that would have rosulted in 1-2 weel's had they not already been down for other rcasons.
It was noted that NUREG-0500 requires a WH to be catastrophic (not cy word) for it to be considered. As a result the data base is biased towards events that shouldn't happen. Many design cccommodated water hammers should be considered. For example, is a chattering valve a water hammer 7, is a design load water hammer a water hammur or does one have to tear something loose?. A view worth paraphrasing here was given by the engineer from Seabrook.
The engineer from Seabrool A lot of time was spent figuring out how to design and avoid water hammer. System designers of NSSS do not pay attention until too late in the game so we wind up with a "fix". We wind up using non-safety grade steam traps etc. to avoid toter hammer. They are put in after the fact and as such are
'. Cometimes less than they should be. It is a bad state of af f airs-having to engineer around a problem that could have been precluded in the initial design.
There is disagreement between system designers as to what a system designer should consider normal or abnormal. As a result, the plant cngineers fix many of the problems during the pre-operational phase cnd they never show up in the data base. The category " operational transients" is not included in piping design. This becomes a question for licensing what is preferable, fixing it after or
- preventing it? With many of the utilities showing little interest cnd having weak engineering staf,f, this means continual trouble 1cter on. In GE reactors procedural problems dominate the WH initiators is this design or system complexity? Some utilities have installed category !! non-invasive annunciator type empty line indicators and have f ound problems ahead of time. It is interesting to note that SONGC-1 has put their empty line sensor cn the shelf because NRC says it will not be very useful.
u m
-c
. ~ .
Ono can calcuhate a peat pressure for a given steam volume'bh .
cocuming the pressure in the bubble is zero. The pressure ec1culated will be the maximum one can achieve and will yield s COtastrophic results. Measurements, however, show that the pressure p 0k is typically about 1/3 the theoretical maximum value. Out of '
400-500 events the peak pressure will be close to the theoretical conimum. It seems to me that one ought to consider water hammer in the same sense as soismic loads have been considered. A design D0cis water hammer is nended f or the piping system designers.
GJveral water hammer over.ts wore described by the utilities.
H:rness (WPEES) described the feedwater thermal stratification induced pipo doflection and its remedy (stronger supports). It will not be further discussed here as it is not a water hammer GvCnt.
i J nes (YAEC) summari:ed modifications at Yankee to control feedwater I hccmer. The steam generator feedring loop seal was modified to
- rCduce the hori: ental length of the line that could empty. The feed
, water uses topping heat during low flow to reduce subcooling during tho initial restert stages. A bypass was added to better control f,
tho feedrato during low flow conditions.
s
! Quinlan (Northoact Uttlities Service Co.) described a potentially f'. Covere water hammer condition at M111stono Nuclear Power g Station-Unit 7.. The condition arises when a pump is restarted r.
cnd water rejoins as a result of a vapor space occurring during
(,
pump shutdown. The rejoining takes place with a large amplitude pressure spile. The vapor space was formed in the relatively high olevation piping near the building air-conditioning water coolers
- citer the servico water system pump tripped. It was found that I Oignificant peal pressures resulted f ollowing pump start if the MOV downstream of the pump was more than 20% open. This result followed a lengthy esportmental study of the actual system. The i cddition of a small air vent line on the inlet piping ?.o the ccoler mitigeted the hammer. Unf ortunately the study was crudely instrumented anJ it in difficult to generalize the results to Cther piping systems. The study does however clearly delineate j the parameters one ought to look at in a general way to come up cith meaningful design guidance. It was noted that 9 months were !
Cpent trying to resolve a similar problem at Seabrook. !
An engineer from Bechtel gave an ALES view of WH. Bechtel uses RELAP5 to,de$ ermine loads for system design. This is done eithout knodledge of the f act that codes like RELAP5 do not hendte condensation wull . The engineers response to why was that "thats 411 we have" and NRC uses it. ,,
I 0
_g_ i Chiu (SONGS-D gave an excellent presentation describing the recent chcck valve gpduced water hammer problem. The details will not be..
given here as W presentation is scheduled to be made to the full, .
ccmmittee at the next meeting. At one time BCE was going to .a inctall an empty line indicator and control room annunciators. .
Thoy purchased the sensors and have now put them on a shelf rather then install them. I asked them why they would do such a strange thing. I was told that NRC says the sensors are not needed because tho additional chuci valves will be suf ficient to insure that line decinage does not occur again. It will cost on the order of
- 1,000,000 to install the sonsors and route the instrument lines to the control room. Without the benefit of an NRC concurrence on thoir usefulness they cannot move the cost to the rate base. NRC
- to requiring SGLD valve position indicators be routed to the ccntrol room. It seems to me that piggy backing a few more lines thould no be overwhelmingly expensive. It would be interesting to find out how the NRC arrived at the conclusion that the empty line censors would not be cost beneficial (another example of treating j Ccch aspect of the problem separately?).
I
' An event at Diablo Canyon had an interesting cause. There were two WH events in the AFW system. They occurred during benchmark testing i Cf the Terry Turbinen. During operetton at 72%, something occurred thct lead to isolating the MSL. When the MGIV isolates, so does tha steam trap. Steam line condensate got into a line where it cculd be accelerated. No, one was aware of the resulting water h mmer when it occurred. A number of design changes were required i to fully eliminate the problem.
Uilliamson FEELO) described a recent Salem event. A motor drivsn i
check valve was inadvnetently lef t partially open. The 50 forceil l bct water back into the condensato system. Vaporization (flashing) yiolded steam that was condensed by the cold water intermittently
,, introduced by the condensate pump. The steam bubble was at a high I
pctnt. Pressure would drop when condensation occurred and the ccndensato pump would como on. A new bubble would be formed and the increasing pressure would repeat the cycle. A lot of the damage fcund was a recult nf the piping supports boing designed for uni-directioral flow. Nc consido-ation was given to possible
,,,ficw transients. Certain aspects of the operating procedures were . ,
fcund to be missing. Powering up the reactor followed very ccmplete procedures whereas reduction in power didn't call f or c1csing certain block valves resulting in full reliance on check volves. This as been remedied. ,
To conclude the meeting EFR! organized " Breakout Bessions" to try and gain guidance for futu o direction in their consideration Cf water hammer. Each of the breabout sessions summarized their daliberations in the form of a series of statuments. The summaries cro reproduced as best as ! can recollect as f ollows:
.- _ - _ _ . _ - . _ - _ _ _ _ - _ - . . - - ---_..__.---______..---_.----..--_.___.-...-_-_--.-l
i
- 1. Prevention, Mitigation and Accommodation Broup
?
Prevention operator procedures caintenance improvements improve valve testing methods design revlefons and improvements Mi ti gati on ,
training operators design improvements Closure time ,.relanation Accommodation classic event-design for it large rare event-can't de anything about it Technology experimental studies of condensation induced events I handbool. on water hammur analyzer -PC computer ,
- 2. Operational and Design Change Insights Group Check valves should only be used for gross flow reversal i
not long term. Where necessary replace them with gate valves. ,
Piping layouts should be evaluated for water hammer potential.
Their are not too many passive devices available so forget it.
Do not use check valves f or long term isolation.
Their in to much burden on the operators.
Provide guidelines review procedures use orifices for continuous blowdown where appropriate.
- 3. Research and Technolog) Needs Group Research necessary to write a handbook on water hammer for practical application in the operating plant environment,ent.
In tummary, the industry usems to recognize the importance of water 8
hammer. They plan to try and do something about it. I get the (Ctrong) feeling that they do not want NAC interfering with how th y plan to do it. The ir.pression one gets is that the industry recognizes the implications of the water hammer issue whereas the NRC does not. ,
. ..