ML20003B816

Informs That Addl Factors Re Deliberate Ignition in Ice Condenser Upper Plenum Region Reinforce Earlier Conclusions. Igniters in Upper Plenum Should Be Removed

ML20003B816
Person / Time
Site:	Sequoyah
Issue date:	02/09/1981
From:	Berman M SANDIA NATIONAL LABORATORIES
To:	Murley T, Ross D Office of Nuclear Reactor Regulation, NRC OFFICE OF NUCLEAR REGULATORY RESEARCH (RES)
References
NUDOCS 8102250620
Download: ML20003B816 (5)

Text

._ .. .

.- Sandia National Laboratories mie. February 9, 1981 Atouquorcuo. No. us..,co en as to: Thomas E. Jiurley and Denwood F. D as United States Nuclear Regulatory Cnmmission O 44Au Gw from: ' Marshall Berman, 4441 sag a Deliberate Ignition in the Ice Condenser Upper Plenum Region of Sequoyah In reference 1, Sandia recommanded that the igniters located in the upper plenum of the ice condensers (IC) be removed for the Sequoyah Interim Distributed Ignition System (IDIS). The justification for this suggestion was based on the following reasoning:

We felt that deliberate ignition in the lower com-partment (LC) was desirable, and that serious threats to ~

containment would most likely arise from situations where LC ignition was prevented. Lower compartment combustion would not occur if the LC was steam inerted or oxygen deprived. Small break and large break calculations with the liARCH code shcwed very large and potentially inerting quantities of steam released at times close to the times of hydrogen release (PJ gs. IS, 20, 23, and 25 in Ref. 1).

LC inerting was, in fact, calculated to occur when the fans were inoperative. Since the IDIS system is intended to burn lean hydrogen mixtures, and steam inerting in the II would produce the richest possible hydrogen mixture at the ice condenser outlets, we recommended that the igniters there be moved higher up in the upper compartment.

This veuld result in the burning of more dilute mixtures and would lessen the possibility of inadvertent local l detonations occarring in the ice condanser upper plenum  !

region.

Sandia has continued its investigation of hydrogen mitigation systems since the release of the Sequoyah l draft (Ref. 1). Three additional factors concerning l igniter placement have been identified which tend to reinforce our earlier conclusions. These include a re-examination of the assumptions and calculations reported in 'Ehe first Sandia Drafts, the " anomalous" steam inerting experiments performed by Lawrence Livermore Laboratory, and new detonation calculations with the Sandia CSQ computer code.

I 8 APPENDLX 0

. 1

.1 .

T. E. Murley & D. P. Ross 1

2. February 9, 1981 i

Sequoyah, This meant that we modified our report to reflect PSAR dataBecausel .

2250 lbs of hydrogen,100% metal-water reaction would produce the draft. ccmpared to 1950 lbs reported in hydrogen concentration for a given fraction of metal-This change water reaction.

the MARCH possible calculational for the results with more care than wasSandia a draft report.

.. oscillation behavior has been observed; in Anomalousacalculational particular, numerical of cladding temperature from cycle to cycle has been noted. (an MARCH input assLT.ptions do not order of magnitude)

Also, some of the as we know it. agree with the plant status appears For example, UHI accumulator injection to have been prevented in some runs, although it would be conditions. expected to occur under the prevailing pressure Whether these problems would result or less conservatism is known, however, is in the results is not known. in more What Given that an abundance of steam is a necessary conditionthat the re for zirconium difficult oxidation to imagine anyand hydrogen production, it is accident prcduction and release scenario in which hydrogen quantities of steam. This would not result be accompanied by large calculations. is independent of MARCH Since MARCH, however, also predicts large quantities of steam for much of the S,D and AB calculations, steam a inarting possible, of the if not loweroccurrence.

likely, compartment must be considered l Sandia adiabatic, also modified and improved its prediction of-isochoric deflagration pressures. These c resulted has assumin the predictions of higher burn pressures.hanges NRC

! pressure.gd that 45 psig would be a conservative " failure" l

compartment Considering that burns in the Sequoyah upper it is prudentcan approach adiabatic, isochoric conditions, to keep those concentrations at or below St.

Lawrence experiments forLivermore NRR.2 Laboratories performed some combustion Two to anomalous (i.e., unexplained) tests results.

at high steam fractions led In one, an initial concentration of 12% H2 and 534 steam was not ignited by a continuously activated glow plug as the steam condensed. No perceptible tion.

ignition occurred even as low as 231 steam frac-A second test was conducted where the glow plug was activated 301. periodically at steam fractions from 50% down to No burns were evidenced at any time, even when the w

. ' =. i T. E. Murley & D. F. Ross .:

3. February 9, 1981 fan was switched plausible on before the final sampin. We have a (although unsubstantiated) results based on our studies of fogs.gxplanation ' If a fog is for these composed flame front (aboutof very1fine mm droplets,wide it may vaporize within the and quench the flame. Such o " cone.ansation fog" appears) to have existed in the LLL tank. i Note that test behavior the MARCH in large scales2S i.e., D calculations mimic this fraction (almost minutesto20% 701 {. decreasing over steam an initially high steam a period of about 90

_. procana, If a fog is generated in this likely. then lower (Sandia has campartment inorting becomes even mor e i ocheme in conjunction with deliberate ignition. recommended Droplet fogging as a mitigation cizes, however, would have to be large enough to pass through the flama front without quenching, but not so large that_they would time interval). _ not vaporize within the necessary since the ice condenser upper plenum region contains in the plant,the highest concentration of hydrogen to be found anywhere detonation occurring which might damage containment.we postulated the calculation A CSQ hydrogen waswas performed in which a 20% concentration of region. This ring ignition is conservativecircumferential1y detonated in the u in this region. unrealistic since TVA has placed a large num,ber of ignitersbut not wholly However, code, the artificial viscosity in the all combine to produce detonation pressures lower than theth oteady state Chapman-Jougust values. FurtherdBrm, the rest of the upper compartment was assumed-to contain only air.

1 All of these latter assumptions tend to reduce the con-corvatism ignition source.introduced by the cylindrical symmetry of the i

toinment for impulsive dome exceeded loading reported theinconservative Refs. 4 and 5. failure criterionCalculated Further cniculations are planned cymmetric versus point-source ignition.

to investigate the effects of

! All of our research to date indicates that upper com-l pcrtment combustion of hydrogen mixtures above 84 should plenum is possible and may be dangerous.bo avoided and that a de the IC UP increase The igniters in without providing any significant safety benefit.the possibility Igniters of a local deton placed higher in the upper burn lean mixtures below St. The TVA and LLL expcompartment should be able to hnve demonstrated this capability in small scale.griments ~

if the igniters cannot burn such mixtures, Indeed, premise of deliberate ignition is questionable.then the Perhaps basic be installed in the upper compartment for the interima more app cystam.

inadvertent ignition i.n.Thet. UC is a large, nearly empty region, and his. . r. egion may be, erp n1$En, ,'

9; 6;g S - '-

. - . = ,

T. E. Murley & D. F. Ross .; ,

4. February 9, 1981 In the absence of hard data, several qualitative argu-ments can be made.

an UC deflagration occurs.The sprays may or may not be on when cause a lean mixture deflagration to propagate back downSpray-induced turb into the IC UP. Although that is an outside possibility, we result believe in that ef more theficient enhanced i turbulence is more likely to with reduced overpressures.gnition at lower this Furthermore, concentrations situation does not present an argument against removal of igniters

. in the IC danger of UP, but rather ignition there. a confirmation of the potential Arguments have also been made that early burning is preferable to later, because uniform concentrations have not had time to build up. Those argu-ments again ignore the experimental data on lean mixture ignition.

ignition. There are also important advantages to delaying Allowing time for the hydrogen to mix will pro-duce high concentrations more dilute mixtures and lessen the probability of remaining in pockets.

may also result in the plating out of some Time delays radioactive aerosols; combustion might also be reignited in the lower compartment (an important question here would be the possibility densers and ofrasultant flame propagation through the ice con-

- ignition in the UC mixture) .

ignition. Jeveral improvements are possible for UC deliberate The most obvious is the inclusion of hydrogen sensors in the UC. and separate controls for igniters placed high hydrogen concentration approached 8%.

These ignitors would be inactivated if the is mechanical separation of regions, perhaps by screensAnother possibility These screens might prevent UC turbulent deflagrations from .

propagating dounward into the IC upper plenum.- Turbulence could also be significantly reduced (if that was shown to be desirable) if the spray systems were not activated prior to a burn '

are with respect to spray activation).(I do not know what TVA's present amarg In summary, we think that: 1. The potential for hydrogen generation previously estimated.and2.combustion may be larger than The codes being amployed for accident analysis may be very inadequate for licensing decisions tuition andand should not engineering be used to replace physical in-judgement. 3. The probability of lower compartment inerting during an accident may be higher because of ccmputational uncertainties and because of the possible mechanism existence of an additional physical for inerting (condensation fog). 4. If a detona-tien were to occur in the IC upper plenum, calculations have tivelyshown that containment failure could not be conserva-precluded.

em- S rq' -

oa S o

__ _ _____.m____.-_______________m .-v- -1 ,

--r-- -9 --- y-m-

.--.7..

T. E. Murley & D. P. Ross 5. February 9, 1981 Deliberate ignition in the upper compartment is a difficult, quantitative problem. Igniters placed high in the upper compartment will probably be beneficial most of the time, but circumstances can be postulated where they might be dangerous. Igniters in the IC upper plenum appear to be justified only 11 one abandons the entire con-cept of lean mixture combustion, and replaces it with early, rapid, rich mixture, local combustion. We think that such a concept is fraught with danger.

If you feel that we can be of further assistence to you, don' t hesitate to call on us.

Retferences

1. M. Berman et al., " Analysis of Hydrogen Hitigation for Degraded Core Accidents in the Sequoyah Nuclear Power Plant", SAND 80-2714, NUREG/CR-1762, December, 1980.

2. W. lowry, " Preliminary Results of Thermal Igniter Experi-ments in H 9-Air-Steam Environments", UCRL-84167, presented l at the.WorRshop on the Impact of Hydrog_en on Water Reactor Safety, Albuquerque, New Mexico, January, 1981.

3. L. S. Rubenstein, " Input to Sequoyah SER on Igniter Systems." -- ~

4. R. K. Byers, "CSQ Calculations of H y Detonations in Zion and Sequoyah", SAND 81-0361C, pTesented at the i

I Workshop on the Impact of Hydrogen on Water Reactor safety, Albuquerque, New Mexico, January, 1981.

5. J. C. Mark, Memorandum for ACRS Members, " Notes on Hydrogen Burn with Igniters", December 4 g 1980.

1 l

l

- _ _ _ ,