ML20140H243
| ML20140H243 | |
| Person / Time | |
|---|---|
| Issue date: | 09/07/1996 |
| From: | Catton I Advisory Committee on Reactor Safeguards |
| To: | Advisory Committee on Reactor Safeguards |
| Shared Package | |
| ML20140H218 | List: |
| References | |
| ACRS-3027, NUDOCS 9705130026 | |
| Download: ML20140H243 (4) | |
Text
. _ _ . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ . . _ . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
2iw racM : !
- casson SEP. 9.*996 t3:M P L 5
, a m, : n , ,, s,,
j TO: ACRS Member:
FROM: Ivan Catton 3 i 1 i
DATE: 7 September 1996 l
SUBJECT:
Condidonal Probability of a steem Genormer Tube Rupture Following a Core Damage Accideas The Issee Given ther a core damage accident has occured and that the system pressure is high, determine what the probabdity that a steam generator tube wiB mpass is. Tbs NRR view is that it should be less than 0.1 to be consistent with requiremsets placed on the conseinment as it plays a i sinular role. This is a defense in depth type statement and comes toen the use of a 104 bypass esquency for an acceptance criterion. Ifit een be shown that this is the caos under the new rule,
! the rule is viewed to be safety neueral. Ifit does not, then the mitigenng edbets of the steem i !
generator internals need to be considered, a much more didicult chore. Here one would anampt l l to taks cruht for the actual Sow rates out of the tube break, the plate out of redio nuclide '
! materials on the sepannors and the remotamos to transpost adbrded by diesence from the tuba l 1
brealt to the outside world. nis is my conjecture on how one should approach the problem l t
i What is the issue before the ACRS? In our letter on the Severs Meidam Reeserch Program, we i' said that the code of choice for this problem. RELAPS/SCDAP, could not do the appropriate j calculations. Such a statement is overty strong as most codes like RELAP5 (the SCDAP part
! plays little role in the issue other then establishing the souros temperature, it has its own j uncertamties) can be tuned to yield adequate results. At the time ofour letter writing, we had not I been made aware of the tuning that had been comed out so that the code could be used. In what
{ follows. I will Srst comment on the tuning that has been done, the highlights of a peer review held
! . at Fauske and Associates and then give my view how one could generate the resuhs that are
! needed by NRR to resolve the issue posed above i
l l The ismas of seem generator tube rupture following a core degradation event first arose in 1983 i as the reauk afeelsuistions done using RELAPS in the Nederlands. It took a long time for the
! nuclear edy ensomunity to come to grips with the problem. The Srst calculations done at INEL
! used an o@ ele shelf version of RELAP5 much like the Dutch did. It became clear that DCH type l accidents in a PWR would be low probabdity and that what would ibil irst (the hot leg, the surge
! line or the steem seasrator tubes) was uncertam. It is a race determined prunardy by our
! knowledge of host transfer and Buid nuxing. This led to a very er:haustave joint experimental study j by NRC and EPRI. The results were clear but qualitative; buoyancy driven recirculation will j effectively heat up the upper parts of the primary system. Another set of expentnants were camed i out, without the participation of the NRC, that were more quanntanve. These experiments were i then used to supply the needed data for code tuning. The experimental results were supplemented
! by use of the COMMTX code. Given this, where are we at? The expenmemal was well scaled i except t'or the fuel rod and steam generator tube sizes. Thmgs like Sow area and pressure drop I were matched to a Wesunghouse PWR. It is not clear that appropriate convergence testing was
)
i 9705130026 960912 PDR ACRS h
l 3027 PDR.
4- _ _ _
L I
l.
j men : Ivan cotton E 9.1996 10:atm p ws m o a! c : tste os 3 ,,3 l
l
! done er the CERWDC to qualify the code reau:ss. This leaves us with the ibeling that the
{ RELAPS/SCDAP ande has been tunad well enough to give one a russonable result. In m this is not enough to make a conditional probability statement and assurance of ssisty l Our statements in the Severe Accident Reneerch letter led RES to most with myself and Tom l Kress and then to convene a peer review gmup made up ofProdmaar Ishii, GrifBek and Viskasta.
l At our meeting with RES, I gave RE5 say view on what our conceres ware. I told them that it
! was my view that RELAP5/SCDAP was a ihr too comples a code A r the problem et hand and i that akhough it could be used, unceruimies would adE have to be desk whb. I chose to attend the
! meeting with the peer review group and RES. The messing was opened by Charlie Ta' ider with j some instruccome to the peer review group. The mind est was ciser. No meetico of the need fbr ;
! uncertaisnee was made The peer review smup was asked only to comunemt on whether or not the i l code could do the job. I then asked the peer review group to consider the need to commier !
i uncertainties in heat transhr and mixing in a way thes would aEow one to make a statuneet about i
the probabdity of tube tilure with some level ofconddance. NRR made a presentation and, j although not as ciesriy stated, indicated a need der such a statsmaat. The reestance by RES to 1
doing so was marensely strong. I don't know if this is because they don't undantend that aR expenments and code results are uncertain to some degree and that in the world of risk intemed regulation this must be addressed or that they have made a policy decision not to do so. These was a great deal of discussica es to how dlSouk it would be to do it. I agossed that it would be much lika the CSAU exercies and the response was that we can't spend two or three years on this. Such a response is didicult to undermand because the present problest is rauch4asier to treat than the one dealt with during the CSAU emerase.
Now, what do I think should be done? A simple study unng the RELAPS/SCDAP code should be '
carried out to generate a response surface for the time to thii a steam genormor tube. The parameters one needs to consider are firw and their ranges are hidy wel knowri 1 will list them and give some rational ihr some ranges and ha the number of q=W one might need.
PARAMETER RANGES AND DISTRIBUTIONS Vesant apper pissem temaperature:
It is wet anospeed that the upper pienum will be well mixed. It will be a haa+y osames on the problem and should be well determined by past SCDAP output. One only needs to andmate what rMW to f assocate with the resukmg spread I willleave this to those more ammliar with core degradation calculanons Het leg heat transfer:
Ourside .wrface: The outside of the hotles is insulated and there is nom' dication of how the insulation will perform under severe accident conditions. Without insulation, the bare pipe heat transfer coeScient will be around 25 W/m aK. As an estimate, one might assume a chance of % for the interval 0-5.0W/m2 K, a chance of 1/4 for 2
the interval 5.0-15W/m2K and 1/4 for the interval 15 30W/m K.
hmde .wrface: This is the region of most uncertainty. There is no avadable data or f
l* y ,q caum , ,,$,y y, p sg
- e q theorise that deal with heat transfhr tosa countersunent asigle phans Sow in j thisk waged steet pipe. It is a cosiugate problem with the % g j
j conneerinartent Bow. One could start with the results WJackson's work fo convection a a horisontal pipe. A range d50 to 200% dJackson's resuk could
- be probable. I would assign a 1/4 chenes to the 50 to 100% ruega, a % chance to j the 100 to 150% interval and a 1/4 chenes to the 150 to 200% intervet.
SwP hw: t The surge line is a problem. Tbs heat transhr wGI be a Sandon of whether or not the PORVS opersas and how the Sow sepersees et the aseste as wee as its location around the hot leg. At this moment, I do not have a recommandenon for i its value(s).
i l Stasas gemarator tube best treaster:
J Inndr arjisce: The heat tranadhr coeSciset on the inside dths tube wiB most likely be j
ununportant as it is not the controEng value. The tube sempensure wit track i
pretty close to the steam temperomre as soon as one is away tem the lower abe i sheet. A good value can be obtained toe the work WJackson. He has saadled l mixed convection in vwtical tubes and written many papers during the past tweety
! or so years Resuhs such a these are typiceEy good to witida +/- 20%. Whh this in i
mind, I would assign a % chance to the interval +/20%, a 2/10 chance to the interval tom 0 to 80% of the Jackson correlation and 3/10 to the interval tom l 120% to 200% of the Jackson correission.
j Owaside arJtice: The outside of the steam generator tube is esposed to a large volume of steem. An estimate of the heat trenefer aaamteia=* under such % would be j around 12 W/m'K. A % chance could be applied to the rangs 713 W/m'K, a 1/4 l chance to the ranges 0 7 and 13-20 W/m'K. -
l c
Steam generator let side pienen mixing:
The dietnbution needed fbr the mixing parameter, ( should be based on the Westinghouse i espermeents whers it was found to be 0.7. It is ramaaaahia to assume that 0.7 +/- 0.1 is a i pesshis reage. The Sow rate could be higher in a fhu piam and the muang parameter
! sheeld imemose 4 Sow rate. On the other hand, the sesam generator tubes are not i sesmendenEy scaled and this could lead to a reduction in maang (or maybe an increase).
- Their are noses indication tom the data that penetration of the hot layer under the tube i sheet took place allowing hotter than average gasses to sneer the tubes. As a result, one
! might assign a % chance to the 0.6-0 8 range, a 1/10 chance to the 0.0-0.6 range and a
{ 3/10 chance to the 0.8 to 1.0 range.
I I RESPONSE SURFACE GENERAT10N l
{ To generate a response surface, one only needs to be sure the parameter ranges are covered well
. _ . . . . . - - - - . - _ - . - - . - - .._.. .- _ - .-.- - -.. - .. - - . ... - . ~ _ .- _. .
l*
i l,
- F20M : Ivan Cat *8" SW'. 9.1996 to:23m p 54
- M'OE m. 131e m 52 1 .-
1 i
l enough to essa a surbs. There are a nunbar of ways to do this. Wesinghouse did their u j BE work byassamlag a Enear relenonship, then checidng and % a e w)CA
! leading to their anal response s 1hos. There are censin charamerimics of the poems pob
! might help. One can, Air esempio, essene that the steun generator tube behavior has linie essa on the hot leg. What is done with the surge line asperated Sow heat transdur eShots linie else. M open PORV edists everything. A series af such N wiB lead so a set ofcniadannas l My guess in that the number will be wuS under 200. On c weskmation, the tinne per run is about l 20 nunutes. With carned compounding of rune, my gases is that the emirs ausrase will take on the order of 100 hours0.00116 days <br />0.0278 hours <br />1.653439e-4 weeks <br />3.805e-5 months <br /> of SUN workstation time.
+
}
1 CONCLUDING RElWARKS 4
l The seam generator rule is a good place to mart impleasesing rielt istinned repletion. To do so, t
one needs to incorporate whe is imown or not known in a nesseingAd and scrutable way. This means one needs to bague incorporating decennimistic results into the FRA menu in a meaning 6d
, way. A prescription might be the ""^i.,,.
! I) de8nition of theissus j 2) description of a model for the phanamana in queadon j 3) speciacetion (and jusaScetion of) of uncertainty distremtions der parameters used in the i models
! 4) establieb a ceiculational metrix fbr generadon of a respones surthos l 5) rendom sempting of the model parameters accortung to their distributions
- 6) estimmion of the probability diaribution tom the rueults
, It seems to me that this is a simple and straightfbrward way to deel with many of the problems l where PRA and deterministic results need to be married into a statement of risk.
d
8 -
IN RESPONSE, PLEASE REFER TO: M960826 September 10, 1996 l
MEMORANDUM TO: E- Thomas Boulette, Chairman Nuclear Safety Research Review Committee i
. FROM: John C. Hoyle, Secretary /s/
SUBJECT:
STAFF REQUIREMENTS - MEETING WITH CHAIRMAN OF ;
NUCLEAR SAFETY RESEARCH REVIEW COMMITTEE i (NSRRC) , 2:00 P.M., MONDAY, AUGUST 26, 1996, COMMISSIONERS' CONFERENCE ROOM, ONE WHITE
- FLINT NORTH, ROCKVILLE, MARYLAND (OPEN TO l PUBLIC ATTENDANCE) l The Commission was briefed by Dr. E. Thomas Boulette, Chairman of i the Nuclear Safety Research Review Committee (:NSRRC) , and Dr. '
i David Morrison, Director of the Office of Research, on the
' results of the NSRRC's June 27-28, 1996 meeting.
i The Commission requested that the NSRRC coordinate its activities with those of the ACRS in areas of joint' interest to ensure that
, the activities are supportive and complimentary and not duplicative. The NSRRC should also continue to review the progress of human factors research. The Committee should identify those human factor aspects that can be treated adequately in PRA, as well as those human factor areas where progress for inclusion in PRA is likely. The Committee should also provide recommendations for integrating these human factors
- considerations into PRA methods.
(NSRRC) (SECY Suspense: 3/7/97) cc: Chairman Jackson Commissioner Rogers Commissioner Dieus Commissioner Diaz Commissioner McGaffigan EDO OGC OCA OIG Office Directors, Regions, ACRS., ACNW, ASLBP (via E-Mail)
/0 u n c a, q.