ML20236K514

Forwards Draft NUREG, Beyond DBAs in Spent Fuel Pools (Generic Issue 82) & Discussion of Project Objectives.Matl Provided to Ensure Consistent Basis for Discussions During Preparation of Final Rept

ML20236K514
Person / Time
Issue date:	04/29/1987
From:	Throm E NRC OFFICE OF NUCLEAR REGULATORY RESEARCH (RES)
To:	Perkins K BROOKHAVEN NATIONAL LABORATORY
Shared Package
ML20236K304	List: IA-87-380, Discusses Status of Activity on Generic Issue 82, Beyond DBAs in Spent Fuel Pools. Comments on Draft Rept,Board Notification,Slides Used in Presentation to EDO & Draft Task Action Plan Re Current Activity Encl ML20236K301 ML20236K389 ML20236K514 ML20236K554 ML20236K574 ML20236K582 ML20236K611 ML20236K634 ML20236K650 ML20236K666 ML20236K713 ML20244C851 ML20245A404 ML20245A414 ML20245A438 ML20245A442 ML20245A528 ML20245A532 ML20245A544 ML20245A591
References
CON-FIN-A-3786, FOIA-87-380, TASK-082, TASK-82, TASK-OR NUDOCS 8708070146
Download: ML20236K514 (61)

Text

{{#Wiki_filter:' ~ ~ ~ ~ -~~ Aprl! 29, 1987 l Dr. Kenneth R. Perkins, Group Leader Containment and Systems Integration Department of Nuclear Energy Brookhaven National Laboratory Associated Universities, Inc. Upton, New York 11973

Subject:

FIN A-3786, "Beyond, Design Basis Accidents In Spent Tool Fools (0182)" The enclosed material is provided for your information and use is preparir.g the final NUREC/CR report for the subject project. Enclosure 1 discusses the C) project objectives and highlights most of the concerns covered in our Project Review Meeting on April 21, 1987. Enclosure 2 is an annotated copy of the draft report identifying areas where additional information should be presented. In addition I an enclosing some background material related to the Hatch seal failure event (Enclosure 3) and a copy of a recent submittal from ANO-1 requesting removal of a license condition concerning shipping cask use (Enclosure 4). Note that AND-1 is referencing Ceneric Letter 85-11. I believe that the enclosed material does not change our perception of the work needed from BNL over the nest two months. The purpose for providing the material is to assure that we have a consistent base for discussions during the preparation of the final report, b* Edward D. Thron, Task Manager { Reactor and Plant Safety Issues Branch j Division of Reactor and Plant Systems Office of Nuclear Regulatory Research l Nuclear Regulatory Commission Mall Stop 244 Washington, DC 20555

Enclosures:

(1) Comments on Draft Report (2) Annotated Mark-Up of Draft Report (3) Hatch Seal Failure Information Package ( ANO-1 Tschnical Specification Change for Use of Spent ruel Shipping Cask 8708070146 870004 l PDR FOIA SHOLLYB7-380 PDR N bN ~

(NGG4SDAV f Comments on BNL Cl-82 Draft Report FIN No. A-3786 Fage 1 PROJECT OBJECTIVES need to present the project objectives sa defined in the Technical Th 'e i' a Assistance Statement of Vork. The objective of the study was to provide an estimate of the likelihood and risk (consequences) of a severe spent fuel pool accident, and identify potential mechanisms and conditions under which they may occut. Two major objectives: ( ! ') Reduce the uncertali. ties in the esidat tor. phenomenology - revised rate equation, propagation, ventilation, etc. (2) Review events which could make loss of integrity (loss of water) more likely than VASH-14CD, therefore BNL considered; (a) seismic review because of SEP/SHCP increased hasards e (b) real failure because of Hadden Neck (c) cask drop because of human error (albeit there is a need to re-evaluate against NUREC-0612, A-36 resolution) (d) other events touched on to verify VASH-1400 estimates of events. Vocid objectives of study be more accurately representive by ABSTRACT text: ABSTRACT This investigation provides an assessment of the likelihood and consequences of a severe accident in a spent fuel storage pool - the complete draining of the pool. Potential mechanisms and conditions for failure of the spent fuel, and the subsequent release of the fission products, are identified. Both PVR and BVR spent fuel storage pool designs are considered. Internal and esternal events and accidents are assessed. Conditions which could lead to failure of the spent fuel Zirealey cladding as a result of cladding rupture or as a result of a self-sustaining oxidation reaction are presented. Propagation of a cladding fire to older stored fuel assemblies is evaluated. Baen' fuel pool fission product inventory is estimated and the releases and consequences for the various cladding failure scenarios are provided. Possible preventive or sitigative measures are qualitatively evaluated. The uncertainties are large, perhaps greater than a factor of ten, and areas where additional evaluations are required are identified. Beyond DBAs in SyPs* )

Comments on BNL Cl-82 Draft Report FIN No. A-3786 Page 2 FAILURE RATES, ESTIMATES AND HUMAN ERROR Unsupported tellance on NUREG-0933 data should be reduced /caitted, and other sources should be used for these data or at least support the data. Prioritisation studies are intended to be scoping. To this point, the BNL draft should be considered a scoping, perhaps conservative, estimate of both the likelihood and risk. VASH-1400 assumption are also scoping in nature. A scoping study is likely to be biased in a conservative manner when best estimate data is not readily available, If a conservative assumption is " acceptable" in terms of consequences then no additional work is necessary. When something

occurs, for example the event frequency is found to be

larger, then additional justification is sometimes needed to support or change previously acceptable, conservative assumptions.

There is a definite need to clean-up values which are stated to be " arbitrary" or are unsupported. An assessment of the importance of these assumptions is needed. O NEW ISSUES IDENT! TIED BY REVIEVERS The staff review of the draft report identified other potential spent fuel pool issues dealing with events which could occur with or without water in the pool. The question concerning pool cooling or aske-up fo!!owing beyond design basis s eismic ev er,t s was r aise d. These issues shocid be addressed or noted as beyond the scope of this study. C0KMENTS ON HEAVY LOADS Tor information, and possible inclusion in final report, the following facts are provided:

1. There are planned shipments of spent fuel from Oconee to McGuire.

2.

Barry, participating in DOE dry storage program, performed drop analysis before cask handling was allowed.

3. ANO-1, participating in high burnup program, has recently submitted Technical Specification change to remove condition prohibiting cask movement and states Generic Letter 85-11 as basis. 1

4. There are only 13 cask currently licensed for shipping.

1 Beyond DEAs in SFPs"

Comments on BNL Cl-82 Draft Report FIN No. A-3786 Page 3 CENIR!C ASSESSMENT VERSUS SURROCATE PLANT APPROACH The project objective was to assess the issie for two plants which where judged to be representative of the acre vulnerable designs, however the course of the investigation has suggested that the selected plants may not represent a bounding case. Site seismic hasards, pool site and plant specific features may be better than others. Thermal power ratings are lower than most plants. While is is important to look at Octual facilities and designs tc assess an

issue, it should be possible to present the information in a more generic manner and identify those features which can alter (increase risk) conclusions.

Suf f i cien'. information should be available from work performed to present results 'n terms of risk versus feature. For esample,. seismic hasard and pool f r a g i l l '. y : Upper, median and lower hasard curves with two fragility estimates would provide upper and (reasonable) lower bound seismic failure estimates for three sites. While Cinna has distinct beneficial features, the generic PVR would be at -Q higher risk for turbine missiles and loss of cooling. The source term, total pool inventory, for the surrogate plants is probably an underestimate of the actual full pool situation for most plants. Consider the cattler, Hatch a dual-unit pool, which can hold ten cores versus the three core inventory of Millstone. Consider also higher bornups and higher power levels of mest plants. The risk profile currently presented is a "present" risk proflie, la that most pools have lower or similar inventories. Higher bornap rates could inpact the cooling time requirements to prevent or reduce propagation, longer decay times would be needed. The BVR and the PVR have distinct operating characteristics and pool designs (location) which, based on the available information, suggests that (with the possible esception of salsaic structural failures) BVR may be at lower risk. A better description of why BVR risk is less than PVR should be provided: (a) BVR power density is lower, burnup of 30,000 NVDINTU vs 40,000 NVD/KTU (i) lower likelihood of fire, lower decay heat levels (ii) assembly design easier to cool (b) BVR pool inside large containment structure (1) ventilation rate may reduce likelihood of fire (11) if fission products release, structure may retain large fraction la presenting the risk proflies, it is necessary to qualify results in terms of implied assumptions: PWR: full pool inventory with 1.0 probabilty of a fire. BVR: last discharge with 0.08 probability of fire. Note it is not possible to arrive at 0.08 from reported values, which range f rom 0 to 0.5 and are noted to be 0.28 or about 20 per cent in other sections of the draft report. Beyand DBAs in SFPs" 3

1 Comments on BNL Cl-82 Draft Report FIN No. A-3786 Page 4 i

SUMMARY

OBSERVATIONS 1 The BNL project objective was to verify, and to the estent practical, reduce the uncertainties perceived in the Zirealoy osidation rate equation as originally used by Sandia. A literature search was used to. Identify data related to Zircalcy-Air reactions at elevated temperatures. It was concluded, by

BNL, that the osidation rate equation and temperature at which self-sustained osidations occurs were reasonable.

In addition, BNL evaluated the !!kelihood of propagation from assembly to assembly because of perceived conservatisas in the original Sandia work. It was concluded, by BNL, that the !!kelihood of propagation was a real phenomenon. Decay power

levels, rack geometry and air flow rates were all found to be important.

Energyadditionby radiation heat transfer, for reasonable source temperatures (1700 C), was found to be sufficient for some cases. The estantion of the fire to the entire pool inventory was based on uncertainties in the location of severely degraded (relocated or molten) fuel rod and fuel pe!!st geometries, areas beyond the capability of the methods used to evaluate the likelihood of fires. The methods available for the study of the fire are judged to be conservative, but without y additional development, the methods as used by BNL are not judged to be overly conservative. Refinements in the methods would probably not alter the conclusion regarding the likelihood of a fire, however the amount of fuel involved could be studied for comples fuel pool loading patterns. The accident event evaluation was not intended to be equivalent to a PRA study typically performed for severe core accident evaluations. The intent was to identify possible accident sequences not previously considered in VASH-1400. BNL identified three such sequences: (a) beyond SSE earthquakes, (b) reactor cavity seal failure and (c) cask drop due to human error. It is also laportant to note that BNL was looking for sequences which could result in the complete draining of the pool either by loss of cooling, loss of make-up, or due to gross structural failure of the pool structure. The estimated frequencies for the event sequences are judged to be conservative. Failures rates used are typical of high estimates used in prioritisation studies, human error rates are judged to be conservative and recovery actions are not addressed. While the project objectives were achieved, the presentation is als-leading. It should be concluded that the study has indeed verified that there may be a high risk associated with at reactor storage of spent fuel and that there are areas which need additional consideration and study before a conclusion can be drawn. Beyond DBAs in STPs" I

Consents on ENL Cl-82 Draft Report FIN No., A-3786 Page 9 POSSIBLE ADDITIONAL VORK (EICLUDING FOOL STRUCTURAL EVALUATION) Fool rack geometrier need to be reviewed to determine materials used, assembly spacing (loading), size on inlet flow hole (important to fire analyses) and for potentsai assessment of rack response to earthquakes beyond SSE. Pool designs need to be reviewed for evaluation of cask drop, including A-36 features. fall ors have submitted information in this area.) A thermal analysis of Zircaloy fire dynamics is needed. How long would fire take, could fuel heatup sufficiently to release volatile fission products? What is real estent of firef Relocation of heat source for failed rods needs to be assessed. To what estent is older fuel involved in fission product release? An assessment of retention of fission products on structures (intact or co!!apsad) is needed. A better assessment of failure rates and human error is needed. A scoping study i C7 of seismic induced failures needs to be considered. Potential addition of spent fuel pool and support systems to Seismic Margins Program. The SFUELIV code could be modified to a!!cw modeling of comples storage geometries (Iow density regions and high density regions). Other conservative

features, such as neglecting air flow around assemblies, could be addressed in a revised computer program.

Beyond DBAs in STPs"

EN ct osV*s D. i ly< NUREC/CR-BNL-NUREG-BEYOND 1)ESIGN-BASIS ACCIDENTS IN SPENT FUEL POOLS (GENERIC ISSUE 82) V.L. Sailor, X.R. Perkins, and H. Connell Containment & Systems Integration Group and J. Weeks Materials Safety Application Group Department of Nuclear Energy Brookhaven National Laboratory Upton, New York 11973 (DRAFT) January 1987 Prepared for Division of Safety Review and Oversight Office of Nuclear Reactor Regulation U.S. Nuclear Regulatory Commission Washington, DC 20555 Contract No. DE-AC02-76CH00016 FIN A-3786

iii ABSTRACT f This investigation has provided an integrated assessment of the risk 'o beycnd design basis accidents in spent fuel pools for two surrogate Plants (a l PWR and BWR). The investigation included an assessment of,hItiating fre-analyses of the accident progression including )he fission product l

quency, releases and health consequences.

The estimated health consequences were p found to be about 12 person-rem /Ry and 130 personpreb1/Ry for the BWR and PWR plcats, respectively. These estimated risk re'sults are comparable to the p estimated risk posed by severe core damage ' accidents and appear to warrant further attention. However, the un,certai nty in this estimate is large (grcater than a f actor of 10) and 'p'lant specific ~ features may change the results considerably. ,/ / Preventive and mit,igative measures have been evaluated qualitatively. It is suggested that for plants with similar risk potential to the two surrogate / plants, the onfe.easure which is likely to be effective in reducing risk is l utilization,cf low density storage racks for recently discharged fuel. How-ever, befbe such preventive measures are implemented a complete plant spe-cifi isk assessment for pool related accidents should be performed including ' structural fragility analysis of the pool itself. j Ws on,esy % p, v;Jes a a s s ess,. e4.] +l. a I;k e l: L~d o-d '"~'*)<***' D sh A s e vere a cc idea e' a sj,*.A -[,sl s.l.-s3e p.,. l-Ue ca-yete <lr';~j e -f 1 l A 7.. L r.,4 c.A J ~.J., :s, <,s e.s; x.,,, .p p -. 1 n. ry.e a + ct, 2n =l Msr9ve,fre{esa s,-f Ne.$ss:

JesJ, ed.

yLfL in R And f%v Q. s j,ed f s I yo.I desig o. ,.,gg!sideed. ,,, e Oh

• et ce ledte el 1.cl exfee,.sI eve.ds a,) a a e ;,l,.,{ 3 a,e onpu,J.

ca,);4;,,, w I,;,,j, ,,,, g g 1

z. estay elaJJ; 9

.d4 ra y-s,h;,e/ ox; j,g., 4 juis Sa..t-e.I ne

p. roe 4ed.

Pnfsy o ok a cladd;5 -hie -l. .Iden -{ e I u.

. (d} ',,

N n ontellies is e v alvede J. Se4-(vsl pool hs'slos ( J.od snve-I ; e:Le des od die re le<> e ) ,d p o

c... ye a >

f,. - k vs - t,.a $cln p.,JeJ. Posu 41e

s. e...;,, o,,.

p et a,h v e. o - J ,. *y k >,, e * >s e es yt; fa-t;,ej ev l leJ. a ce Io y y perk s cyr e ak. -%, -[s k -} / *)

• 'd

%. v,u %k

c. c e y

vw . 9. ;<. : y a d); 4;., J e n i k *- c i deAp< J. ]

S-1 BEYOND DESIGN-BASIS ACCIDENTS IN SPENT FUEL POOLS (GENERIC ISSUE 82) pflM p/I

SUMMARY

S.1 INTRODUCTION Generic Safety Issue 82, "Beyond Design Basis Accidents in Spent Fuel 7 a L; h Pools," was assigned MEDIUM priority in November 1983.1 In^+115 prioritiza-j em:.I e a .- s IJo a tion, the NRC stafgtock-asc0unt of two factors that had not been considered o g in earlier risk assessments:2 ' Aten Tc 1. Spent fuel is currently being stored rather than shipped for repro-g wcx cessing or repository disposal, resulting in much larger inventories d% of spent assemblies in reactor fuel basins than had previously been rA M' anticipated; and, Gym'd 3 To 2 2. A theoretical model .4 suggested the possibility of catastrophic Zircaloy fire, propagating from assembly-to-assembly in the event of 0 complete drainage of water from the pool, p ce%. lyech,c) 40 m W4)N-lh r. 9 7 S.I.1 Previous Investigations ys gfl f.g. WA3H-14m me g,/ l/3 c.a. 2 The Reactor Safety Study which did act tcke accouat of the twc facter ad concluded that the risks associated with spent fuel storage were ex-tremely small in comparison with accidents associated with the reactor core. That conclusion was based on design and operational features of the storage pools which made the loss of water inventory highl unlikely. Tn a d 4. u .,4_; m s 4 % <J & + J ia p.I sa.r A, w id L< I; :4td a h.A o.,edL;.J ef a core. Subsequent to the Reactor Safety Study, A.S. Benjamin et al.3." inves-tigated the heatup of spent fuel following drainage of the pool. A computer code, SFUEL, was developed to analyze thermal-hydraulic phenomena oct.urring when storage racks and spent assemblies become exposed to air. Calculations with SFUEL indicated that, for some storage configurations and decay times, the Zircaloy cladding could reach temperatures at which the l t

l S-2 (/mly ce & ex/ & s exothermic oxidation would become self-su ining with resultant destruction l l of the cladding and fission product re. ase. The possibility of propagation l to adjacent assemblies (i.e., the c dding would catch fire and burn at a hot enougn temperature to heat neigh ring fuel assemblies to the ignition point) i was also identified. [In such case 3 the entire inventory of stored fuel could become involved. Cladding fire, of this type could occur at temperatures well L q below the melting point of 'the 00 fuel. The cladding ignition point is about 2 900*C compared to the fuel melting point of 2880*C. S.I.2 Related Events fu - There is no case on record of a significant loss of water inventory from a domestic, commercial spent fuel storage pool. However, o recent incident occurred at the Haddam Neck reactor that raised concern ab ut the possibility of a partial draindown of a storage pool as a result of seal failure in the A refueling cavity at a time when the transfer tube gates to the pool were open, l or when transfer of a spent fuel assembly was in progress.3 JL k n.[al:.$ co;4y -[l.>)<J. l The Haddam. Neck incident occurred during preparations for refueling An g inflatable seal bridging the annulus between the reactor vessel flange and the reactor cavity bearing plate extruded into the gap, allowing 200,000 gallons of borated water to drain out of the refueling cavity into the lower levels of the containment building in about 20 minutes. Gates to the transfer tube and the fuel storage pool were in the closed position, so no water drained from the pool.6 More recently a pneumatic seal failure in the Hatch spent fuel basin which released approximately 141, 000 gallons of water resulted in a drop in hf water level in the pool of about five feet 7 al 4 A k -Ile I** I bilm 'i-NIM M ! /) g MdA was a.t 4b sa as a4 H4/4 Nett. k S.I.3 Report Objective W The objective of this report is to provide an T tegfal assessment of the f 3 potential (6f beyond design basis accidQ i$ spent fuel pool s. The risk risks are defined in terms of R6Anv /'o o' DMwwfo C Wm N '" j D%- /voT 6E o

1 i i r S-3 { { the probabilities of various initiating events that might compromise i the structural integrity of the pool or its cooling capability. the probability of a system failure, given an initiating event, fuel failure mechanisms, given a system failure. potential radionuclides releases, and consequences of a specified release. This study generally follows the logic of a typical probabilistic risk analysis (PRA); however, because of the relatively limited nunber of potential accident sequence e analyses ere atk[simpli fied. wkM c~ldcewl4i., 4le clui,,;g y&pl4 S.I.4 Spent Fuel Storage Pool Designs o The configurations of spent fuel storage pools vary from plant to plant. In BWR's, the pools are located within the reactor building with the botton of the pool at about the same elevation as the upper portion of the reactor pres-sure vessel. During refueling the cavity above the top of the pressure vessel is flooded to thc same elevation as the storage pool, so that fuel assenblies can be transferred directly from the reactor to the pool via a gate which sep-arates the pool from the cavity. In PWR plants, the storage pool is located in an auxiliary building. In some cases the pool surface is at about grade level, in others the pool bottom is at grade. The refueling cavities are usually connected to the storage pool by a transfer tube. During refueling the spent assembly is removed from the reactor vessel and placed in a contain-er which then turns on its ride, moves through transfer tube to storage pool, set upright again and removed from the transfer container to a storage rack. Various gates and weirs separate different sections of the transfer and stor-age systems. More details concerning various configurations are given in Section 2.3 and Table 1.1. p# rop g6 Ng Mp S.I.5 Selection of Surrogate Cases for More Detailed Studies Two " older vintage" plants were selected t erve as BWR and PWR surro-gates for more detailed studies. The choice, Millstone 1 and Ginna, were based primarily on such factors as availability of data and the relative familiarity of the proj ect staff with the various candidate sites. The

S-4 1cm WiM i - operating histories of the two surrogate plants were modeled to obtain a g en.., realistic radioactive inventory in-the various spent fuel batches. } ryw 5.2 ACM LEifl INITIATING EVENTS AND PROBABILITY ESTIMATES /tCo i,Ama l a d erkral eve.ds' ' f Accident initiating events that have been considered include U pool heatup due to loss of cooling water circulation capability. structural failure of pool due to seismic events or missiles, partial draindown of pool due to pneumatic seal failure, and structural failure of pool due to a heavy load drop. Estimates of the likelihood for each of these initiators are provided in g5 h" Section2.bt oncluded that e dominant tiators are stp etural 1-b 8 g ures res ng from a mic eve -2x10-5/Ry) and 'fie'avy Wad drops 0-5/Ry).] Uncertainties in the probability estimates are quite large, gad ' being at least-an order of magnitude in either direction. In the case of pM4 te seismic events, the seismic hazard and structural fragilities both contribute rF r to the uncertainty range. For heavy load drops, human error probabilities-and A# .y, structural damage potentials' are the primary sources of uncertainties. NM. n {PO,. S.3 EVALUATION OF FUEL CLADDING FAILURE } The SFUEL computer code developed at Sandia National Laboratories (SNL) l pw (,A ' by Benjamin et al.,5 analyzes the behavior of spent fuel assemblies after an accident has drained the pool. The analyses predict that self-sustaining oxi-dation of the Zircaloy cladding (i.e., a cladding fire) would occur for a wide range of decay heat levels and storage geometries. Several limitations in the SFUEL analyses had been recognized in Reference 3 and have been addressed in a modified version of the code, SFUELIW.4 The BNL evaluations of SFUELIW have led to the conclusions that the modi-fied code gives a reasonable estimate of the potential for propagation M a cladding fire from high power to low power spent fuel and thap the code pro-vides a valuable tool for assessing the likelihood of a e IstNdpMy., fire for a variety of spent fuel configurations in the event that the pool is drained. 4 %p ly.~cd7 ~ g N a' ') Srvit. c a..,4 h n LM 4L:,4, w, if_

S-5 S.4 CONSEQUENCE EVALUATION 2re. Radioactive releases % estimated for the two surrogate plants for,p4 vel l cladding failure scenarios predicted by SFUEL calculations, Svolvig c laddig j $;m. PirLi desu e eve.h wke.- 4 a, J ee leve 4Le. .4, 4,. c. K. c.laddin,1,,, E,,,, jf, j, 7

c.,f v, e m,.

Lv. b Q. r.J g 7Radioactive Inventories ) m y 5.4.1 The radioactive inventories contained in the spent fuel pool s (as of April 1987) for Millstone 1 and Ginna were calculated using the ORIGEN2 com-1 puter code,s based on the operating histories of each of the plants (Appendix A). The calculated data included the 1987 inventories for each fuel batch discharged at each refueling over the operating history. ,f;y, ( 4 / '.;^,, S.4.2 Release Estimates ,f con, Q Fractional releases for various groups of radionuclides were timated based on the physical parameters characterizSg the SFUEL failur scenario. Thus, four source terms were estimated corresponding to the r accident scenarios; Sees Nv Iv:n (c!d 5 h b v'n'~s *~d & b l *)

• ~d %
• sesiv;9 c.IsJhm ra(4vn r:R.J a 4:r.).

1' S.4.3 Off-Site Radiological Consequences g A, Off-site radiological consequences were calculated using the CRAC2 com-N put r code.8 8ecause 'of several features in the health physics modeling in C#"" to the CRAC2 code, the population dose results appear to be of limited value, h The most meaningful measure of the accident severity appears to be the inter-p# diction area (contaminated land area) which in the worst cases was about two 1 L A,e.I uvw - : r., j [ orders of magnitude greater than for core-melt accidentJ o " prompt fatali-ties" were predicted and the risk of injury was negligible. 'g g kHL sileh. * * * 'k S.5 RISK PROFILE ,, _,,,,,, N o 8 r +^ # I 4La4 x e The likelihood and consequences of various spent f"al pool accidents have been combined to obtain the risks which $re summarized in Table S.I. As noted above, the population dose results are of limited value because they are driven by decontamination levels assigned within the CRAC2 code. Thus the

S-6 s p1 land interdiction area is included in Table @.1, as a more meaningful repre-sentation of severity. The uncertainty in each of these risk indices is esti- [ mated to be an order of magnitude in either direction and is due principally ,3 to uncert.41nty in the fragility of the pools and uncertainty in the seismic 4 or oF hazard. H3 * ' tu nw F (a P W Table S.1 Estimated' Risk for the Two Surrogate Spent Fuel Pools % r,.. from the Two Dominant Contributors g gu. (2 ) M ~~ Spent Fuel Interdiction 4r od"

• Accident Pool Fire Health Risk Risk 0,0f FW Initiator Probability /Ry (Man-rem /Ry)

(Sq. Mi./Ry) c m v.Gur

• Seismic induced cm., 4o PWR pool failure 1.6x10-5 37 8.4x10 4 g

y g. Seismic induced at fjiY. BWR pool failure 1.8x10-6 4 7.6x10 5 Cask drop

• induced gg PWR pool failure 3.1x10-5 71 001 Cask drop
• induced cs BWR pool failure 2.5x10 6 6

1.1x10 4 rg a o*

• After removal of accumulated inventory resumes.

(Note that many new plants ~ have pool configurations and administrative procedures which would preclud$.Icthis failure mode.) swm w ak m subwg sea no % w our Aco ~ w The overall risk due to beyond desi n basis accidents in spent fuel pools d #" v for the PWR surrogate plant is about 30 person-rem /Ry and about iperson- ^'M 7* rem /Ry for the BWR surrogate. These estimates are comparable to present esti- "[hm mates for dominant core melt accidents and appear to warrant further attention F-m # on this basis alone. However, the unique character of such an accident (sub- ( P v' b { stantial releases of long lived isotopes) makes it difficult to compare to (( reactor core melt accidents. The exposure calculations are driven by assump-p Au ' tions in the CRAC modeling and the results are not sensitive to the severity 5W)# of the accident. In terms of interdiction area this type of accident has the f potential to be much worse than a reactor core melt accident. p f,,r. g CR vie-Note that the risk results are calculated for two surrogate plants and k: W 4""'C'" {may not be applicable to generic pool types. Wi 3 w -,a p ce4 w.u u,)M'> v.a ? ~ I L

S-7 S.6 CONSIDERATION OF MEASURE WHICH MIGHT REDUCE CONSEQUENCES cw 5,Je< d A number of potential preventive and mitigative measures have been-pee-pos+4-but the only one which is judged to provide a substantial measure of risk reduction is a modification of the spent fuel storage racks themselves. For those plants that use a high density storage rack configuration Improve-ment in the air circulatio'n capability is estimated to result in risk reduc-tion up to a factor of ten. Les:7 os rmree S.7 References for Summary ](( lm 3ym/ 1. "A Prioritization of Generic Safety Issues," Division of Sa.fety Technolo-gy, Office of Nuclear Reactor Regulation, U.S. Nuclear Regulatory Commis-sion, NUREG-0933, December 1983, pp. 3.82-1 through 6 2. " Reactor Safety Study, An Assessment of Accident Risks in U.S. Comercial Nuclear Power Plants," U.S. Nuclear Regulatory Commission, NUREG-75/014 (WASH-1400), October 1975, App. I, Section 5. 3. A.S. Benjamin, D.J. McClosksy, D. A. Powers, and S. A. Dupree, " Spent Fuel Heatup Following Loss of Water During Storage," prepared for the U.S. Nuclear Regulatory Comission by Sandia Laboratories, NUREG/CR-0649 (SAND 77-1371), May 1979. 4 N. A. Pisano, F. Best, A.S. Benjamin and K.T. Stalker, "The Potential for Propagation of a Self-Sustaining Zirconium 0xidation Following Loss of Water in a Spent Fuel Storage Pool," prepared for the U.S. Nuclear Regu-latory Commission by Sandia Laboratories, (Draft Manuscript, January 1984) (Note: the project ran out of funds before the report was pub-lished.) 5. IE Bulletin No. 84-03: " Refueling Cavity Water Seal," U.S. Nuclear Regu-latory Commission, Office of Inspection and Enforcement, August 24, 1984 6 Licensee Event Report, LER No. 8/,-013-00, Haddam Neck, Docket No. 50-213, " Failure of Refueling Pool Seal," 09/21/84.

S-8 7. Nucleonics Week, December 11, 1986, pg. 3-4, j 8. A.G. Croff, "0RIGEN2: A Versatile Computer Code for Calculating the Nuclide Composition and Characteristics of Nuclear Materials," Nuclear Technology, Vol. 62, pp. 335-352, September 1983 - t 9. L.T. Ritchie, J.D. Jofinson and R.M. Blond, Calculations of Reactor Acci-dent Consequences Version 2, CRAC2: Computer Code User's Guide, prepared by Sandia National Laboratories for the U.S. Nuclear Regulatory Commis-sion, NUREG/CR-2326 (SAND 81-1994), February 1983. 1 o I e .--____.__.-s

1-1 1. INTRODUCTION lfh Generic Safety Issue 82, "Beyond Design Basis Accidents in Spent Fuel f [,"rh' Pcols," was assigned MEDIUM priority in November 1983.1 In its prioritiza-tion, the NRC staff took account of twn factors that had not been considered in earlier risk assessments:2 1. Spent fuel is currently being stored rather than shipped for repro-cessing or repository disposal, resulting in much larger inventories of spent assemblies in reactor fuel basins than had previously been c.ved anticipated; and, 2. A theoretical model 3 suggested the possibility of catastrophic Zirca-O loy fire, propagating from assembly to assembly in the event of com-plete drainage of water from the pool. 1.1 Previous Investigations The Reactor Safety Study 2 (which did not take account of the two factors 33 above) concluded that the risks associated with spent fuel storage were ex-jA tremely small in comparison with accidents associated with the reactor core, f That conclusion was based on design and operational features of the storage pools which made the loss of water inventory highly unlikely, e.g., The pool structures were designed to withstand safe shutdown earth-

quakes, The fuel racks were designed to preclude criticality, Pool design and instrumentation precluded inadvertent and undetected loss of water inventory, Procedures and interlocks prevented the drop of heavy loads on stored assemblies, and The storage structures were designed to accommodate the forces and missiles generated by violent storms.

Probabilities of pool failures due to external events (earthquakes, mis-siles) or heavy load drops were estimated to be in the range of 10-6/ year.

1-2 Radioactive release estimates were based on melting of 1/3 of a core for var-ious decay periods, with and without filtration of the building atmosphere (see Ref. 2, Table 1 5-2). Subsequent to the Reactor Safety Study, A.S. Benjamin et al.3 investigat-ed the heatup of spent fuel following drainage of the pool. A computer code, SFUEL, was developed to a"nalyze thermal-hydraulic phenomena occurring when storage racks and spent assemblies become exposed to air. The computer model thkes into account decay time, fuel assembly design, storage racks design, packing density, room ventilation and other variables that affect the heatup of the fuel. g .g Calculations with SFUEL indicated that for some storage configurations and decay times, the Zircaloy cladding c d reach temperatures at which the exothermic oxidation would become self ustaining with resultant destruction of the cladding and fission product lease. The possibility of propagation to adjacent assemblies (i.e., the adding would catch fire and burn at a hot enough temperature to heat neig oring fuel assemblies to the ignition point) was also identified. (in suchdash the' entire inventory of stored faal could become involved. Cladding fires of this type could occur at temperatures well below the melting point of the U0 fuel. The cladding ignition point is about 2 900*C compared to the fuel melting point of 2880*C. Uncertainties in the SFUEL calculations were primarily attributed to un-certainties in the zirconium oxidation rates. Further work was done to refine the SFUEL computer model and to compare calculated results with experimental data." These more recent results have generally confirmed the earlier concepts of a Zircaloy fire which, given the .-ight cor.ditions, will propagate to neighboring assemblies. However, compari-sons to out-of-pile heat-up data have not shown good agreement with the code. The aM noted that more work in several areas was needed to define more precise 1 the conditions and configurations which allow or prevent propag;. tion. I WL * < Sa d e, / Sg % sv 1 w+ m l~e e k. ,.p, y p,s y y 4,,, .N, sLA J~ r"*~#, 7 / j _w ~ 7*~ a / w... I' Td' I,

1-3 Several studies have been conducted on alternative spent fuel storage concepts. Among' these is a report published by the Electric Power Research Institute (EPRI), which applies pro abilistic risk assessment techniques to several storage concepts.5 - his study does not directly address Generic f Safety Issue 824 heer,-Tt does provide useful insight on appropriate analy-tical methodology as well'as useful data on an in-ground (on-site) storage i pool. d 1.2 Related Events There is no case on record of a significant loss of water inventory from a domestic, commercial spent fuel storage pool.10wevert One recent incident h occurredattheHaddamNeckreactor[6eiraisedconcernagoutthepossibility p of a partial draindown of a storage pool as a result ops 8aYfak5ure in the refueling cavity at a time when the transfer tube gates to the pool were open, or when transfer of a spent fuel assembly was in progress.6 The Haddam Neck incident occurred during preparations for refueling. An inflatable seal bridging the annulus between the reactor vessel flange and the reactor cavity bearing plate extruded into the gap, allowing 200,000 gallons of borated water to drain out of the refueling cavity into the lower levels of the containment building in about 20 minutes. Gates to the transfer tube end the fuel storage pool were in the closed position, so no water drained from the pool.7 Rowagerd hd these gates been open at the time of the leak, and l had they not been closed within 10 to 15 minutes, the pool would have drained J to a depth of about 8.5 feet, exposing the upper 3 feet of the active fuel re-gion in the spent fuel assemblies.7 Also, had the transfer of spent fuel been f in progress with an assembly on the refueling machine, immediate action would 7[of ' have been necessary to place the assembly in a safe location under water to /fst limit exposure to personnel. rAe use. Am i/c Jefied 4G a9<d of,j The em t*-[*

• c- / ['

sea' sa..u. k. 4 au ,,,a in nana i nW7. 3 y (e.1 h, %. o u tje.4{. a,uk.4ebl Gr.,be:s %< lM " frLol: C* *d The NRC Office of Inspection and Enforcement required all licensees to promptly evaluate the potential for refueling cavity seal failures.6 pg, sponses indicated that the refueling cavity configuration at Haddam Neck is un.ique in that the annulus between the reactor flange and the cavity bearing plate is more than 2 feet wide. I'n most plants this gap is only 2 inches

1-4 9 / Wl wide.Le) About 40 operating (or soon to operate) reactors use inflatable pc seals. However, because of design differences, the Haddam Neck failure does grF not appear to be directly applicable to the other plants. It is noted that 5.P. f....ts have pemanent steel bellows seals to fill the gap between the reactor flange and the cavity bearing plate. This issue is discussed more fully in Section 2.3 1.3 Risk Potential The risk potentials of "Beyond Design Basis Accidents in Spent Fuel Pools" are defined in terms of the probabilities of various initiating events that might compromise the structural integrity of the pool or its cooling capability. the probability of a system failure, given an initiating event, fuel failure mechanisms, given a system failure. potential radionuclides releases, and consequences of a specified release. fey This study generally follows the logic of a typical probabilistic risk go analysis (PRA); however, because of the relatively limited number of potential p END accident sequences, the analyses are greatly simplified. 1.4 Discussion of Spent Fuel Storage Pool Designs and Features The general design criteria for spent fuel storage facilities are stated g inAppendixAof10CFR50, and are discussed more fully in Regulatory Guide d$ 1.13.% o b w FS The pool structures, spent fuel racks and overhead cranes must be design-1 ed to Seismic Category I standards. It is required that the systems be de-signed (1) with capability to pemit appropriate periodic inspection and test-ing of components important to safety, (2) with suitable shielding for radia-tion protection, (3) with appropriate containment, confinement, and filtering systems, (4) with a residual heat removal capability having reliability and testability that reflects the importance to safety of decay heat and other

1-5 residual heat removal, and (5) to prevent significant reduction in fuel stor-age coolant inventory under accident conditions.9 1 The configurations of spent fuel storage pools vary from plant to plant. { Table 1.1 lists various information about the pools for licensed plants. In BWRs, the pools are located within the reactor building with the bot-tom of the pool at about the same elevation as the upper portion of the reac-tor pressure vessel. (For example, at Oyster Creek the bottom of the pool is at elevation 80'6", and the top at 119'3". The water depth is 38 feet.) Du r-ing refueling, the cavity above the top of the pressure vessel is flooded to the same elevation as the storage pool, so that fuel assemblies, can be trans-ferred directly from the reactor to the pool via a gate which separates the ce pool from the cavity. In PWR plants, the storage pool is located in an auxiliary building. In some cases the pool surface is at about grade level, in others the pool bottom is at grade. The refueling cavities are usually connected to the storage pool 4 by a transfer tube. During refueling the spent assembly is removed from the reactor vessel and placed in a container which then turns on its side, moves through transfer tube to storage pool, set upright again and removed from the transfer container to a storage rack. Various gates and weirs separate dif-ferent sections of the transfer and storage systems. More details concerning various configurations are given in Section 2.3. 1.5 Selection of Surrogate Cases for More Detailed Studies ef I F Two " older vintage" plants were selected to ve as BWR and PWR surro-gates fcr more detailed studies. The choices, Millstone 1 and Ginna, were -h:2 meWt -:rbitrarilh based primarily on such factors as availability of data and the relative familiarity of the project staff with the various candi-date sites. The operating histories of the two surrogate plants were modeled to obtain a realistic radioactive inventory in the various spent fuel batch-es. Details of the modeling procedures and a listing of the calculated radio-nuclide content are presented in Appendix A. g p gy M r.+- ffc-n.~ 4 l os N D

) 1-6 It should be noted that both surrogate plants have relatively large in-ventories of spent fuel assemblies in their spent fuel basins. 1.6 De m t Content Accident initiating events and their probabilities are covered in Section 2. Fuel cladding failure' scenarios based on the SFUELIW Computer Code are evaluated in Section 3. Included are sensitivity analyses of the failure scenarios arising from uncertainties in Zircaloy oxidation reaction rate data, and hardware configuration assumptions. Section 4 presents data on the poten-tial for releases of radionuclides under various cladding failure scenarios and compares the projected releases with releases associated with severe core accident sequences. In Section 5, risk profiles are developed in terms of D person-rem population doses for several accident sequences. Section 6 considersmeasuresthatmightmitigatehyonddesignbasisaccid p l d es:- uy a_-) / c b <. l p p b 1.7 References for Section 1 1. "A Priorit.ization of Generic Safety Issues," Division of Safety Technolo-gy, Office of Nuclear Reactor Regulation, U.S. Nuclear Regulatory Commis-sion, NUREG-0933, December 1983, pp. 3.82-1 through 6, 2. " Reactor Safety Study, An Assessment of Accident Risks in U.S. Commercial Nuclear Power Plants," U.S. Nuclear Regulatory Commission, NUREG-75/014 (WASH-1400), October 1975, App. I, Section 5. 3. A.S. Benjamin, D.J. McClosksy, D. A. Powers, and S. A. Dupree, " Spent Fuel Heatup Following Loss of Water During Storage," prepared for the U.S. Nuclear Regulatory Commission by Sandia Laboratories, NUREG/CR-0649 (SAND 77-1371), May 1979. 4 ti. A. Pisano, F. Best, A.S. Benjamin and K.T. Stalker, "The Potential for Propagation of a Self-Sustaining Zirconium 0xidation Following Loss of Water in a Spent Fuel Storage Pool," prepared for the U.S. Nuclear Regu-l atory Commissi on by Sandia Laboratories, (Dra f t Ma nusc ript, January 1984) (Note: the project ran out of funds before the report was pub-lished.)

1-7 5. D.D. Orvis, C. Johnson, and R. Jones, " Review of Proposed Dry-Storage Concepts Using Probabilistic Risk Assessment," prepared for the Electric Power Research Institute by the NUS Corporation, EPRI NP-3365, February 1364 6. IE Bulletin No. 84-03: " Refueling Cavity Water Seal," U.S. Nuclear Regu-latory Commission, Off' ice of Inspection and Enforcement, August 24, 1984 7. Licensee Event Report, LER No. 84-013-00, Haddam Neck, Docket No. 50-213, " Failure of Refueling Pool Seal," 09/21/84 r p [ Licensee Responses to NRC IE Bulletin.No. 84-03. /0[ Code of Federal Regulations, Title 10, Part 50, " Domestic Licensing of Production and Utilization Facilities, Appendix A, ' General Design Cri-teria for Nuclear Power Plants,' General Design Criterion 61, ' Fuel Stor-age and Handling and Radioactivity Control'." ll }d. U.S. Nuclear Regulatory Commission, Regulatory Guide 1.13, " Spent Fuel Storage Facility Design Basis," December 1981. y., %d cpn Fr-s? up h ic, 4 8. "Ge.,a.:. I s < tia - =y u. f C., A f S sVA *~ " k 7~ r s,.c is, vi .t.,, o, -...u a n-. ~ L o ven.3'LF $ H 12. i k -L, 'b:-ssAn-Of};<. oj f'vJa' b '& dd Yele p. L k,., pg c., vay /3,1987 y 1 _-_.~J

2-1 2. ACCIDENT INITIATING EVENTS AND PROBABILITY ESTIMATES 2.1 Loss of Water Circulating Capability The spent fuel bcsins of U.S. nuclear power stations contain a large in-ventory of water, primarily to provide ample radiation shielding over the top of the stored spent fuel. Some typical pool dimensions and water inventories ere shown in Table 2.1. The heat load from decay heat of spent fuel depends on decay time since the last refueling. Heat loads for the entire spent fuel inventory of the two older vintage surrogate plants are shown in Table 2.2 (data ' extrapolated to the 1987 scheduled refuelings). The cooling systems provided for : pent fuel pools typically have a capacity in the. range of 15 to 6 3 g 20x10 Btu /hr (4.4 to 5.9x10 kw). In the event that normal circulation of the cooling water is disrupted, e.g., due to station blackout, pump failure, pipe rupture, etc., the water temperature of the pool would steadily increase until bulk boiling occurred. (Note: In a situation where the stored inventory was small, an equilibrium temperature, below the boiling point, would be reached at which surface evap-oration balanced the decay heat load). Thermal-hydraulic analyses of the consequences of partial or complete loss of pool cooling capability are a routine part of the safety analysis re-ports required for licensing and amendments thereto, Generally, these analy-ses consider several scenarios ranging from typical to extremely conservative conditions. A sampling of conservative results for several plants is given in Table 2.3. The data clearly demonstrate that the time interval from loss of circulation until exposure of fuel to air is quite long. Even in the most Ipw pessimistic case cited in Table 2.3 (Docket No. 50-247), the water level in I M* n Lhe pooi wuuld drop only about 6 inches per hour. Thus, there (ppears to bD "(3, v w considerable time available to restore normal cooling or to implement one of'i several alternative backup options for cooling. For licensing purposes, it has been accepted that the time interval for restoring cooling manually from available water sources is adequate without requirir)g active (automatic) redundant cooling systems. I

2-2 However, in considering the prioritization of Generic Issue 82, "Beyond Design Basis Accidents in Spent Fuel Pools," the NRC staff recognized that there is a finite probability that cooling could not be restored in a timely manner.2 The case treated in Ref. 2 was for a BWR. The estimated frequency for the loss of one (of two) cooling " trains" was taken to be 0.1/Ry (the value assumed in WASH-1400).3 This combined with the conditional probabili-m ties of failure /non-availab'ility of the second " train" yielded a combined fre- [h quency of a pool heatup event of 3.7x10 2/Ry. (This estimate appears to be somewhat conservative since no " pool heatup events" are on record after -10 p} 3/ reactor years of accumulated experience). I To escalate from a " pool heatup event" to an event which results in fuel D ** damage requires the failure of several alternative systems that are capable of supplying makeup water (the kriR and condensate transfer systems, or, as a last resort, a fire hose). Estimated frequencies of failure for each of the alter-natives, combined with the frequency of a pool heatup event, resulted in an estimated frequency of 1.4x10 6/Ry for an accident initiated by loss of spent fuel pool cooling. f Originally, the spent fuel pool at the Ginna plant had only one installed ' ggvA cooling train with a " skid-mounted" backup pump and heat exchanger.

However, pa A a second cooling train was to have been installed in 1986.

Because of the M# third option for cooling at Ginna (the skid-mounted system) the probability estimate for an accident initiated by a pool heatup event should be reduced to 5x10 7/Ry, i.e., about a factor of 3 smaller than for the BWR case analyzed in Ref. 2. 2.2 Structural Failure of Pool m ge

,o c, %,,y,

,, 0 1 A 3-12 Because of the massive reinforced concrete structure of LWR spent fuel storage pools, designed to Category I seismic criteria, initiating events that would lead to a structural failure are extremely unlikely. On the other hand, a structural failure that resulted in rapid and complete draining of water from the pool would have serious consequences. Probabilities of events that might result in loss of structural integrity are estimated in the following two subsections.

2-12 1 Therefore, higher weighting factors were assigned to the " Zion" curves as shown in Table 2.5. Based on an upper limit PGA cut-off of 1.259, the mean probability resulting from the convolution of the single hazard curve with the 'd-

T M fragility curves was F = 1.6x10-5/ year (Ginna).

f The difference between the estimates for Millstone and Ginna, 2.2x10-5 vs.1.Cx10-5, should not be regarded as highly significant, but more as an in-f dication of the sensitivity of the results to the weighting factors assigned { to the fragility curves. ) 2.2.2 Structural Failures of Pool Due to Missiles \\ Missiles generated by tornadoes, aircraft crashes or turbine failure ( A >l could penetrate the pool structure and result in structural failure. 0.sW t TH W The probability of tornado missiles depends on the frequency of tornadoes at the site, the target area presented to the missile and the angle of im-pnr

• pact.

An analysis made by Orvis et al.25 for an average U.S. site derives a probability of <1x10-8/ year for structural loss of pool integrity due to a tornado missile (Ref. 25, pg. 4-44). Similarly, the analysis for structural failure of a pool from an aircraft P crash yielded a probability of <1x10-10/ year (Ref. 25, pg. 4-58). The damage caused by Missiles generated by turbine failure depends on the 1 gge orientation of the turbine axis relative to the structure, as well as the fre-p >' quency of turbine failure. An analysis by Bush yields a probability of ib o -4x10-7/ year for spent fuel pool damage from a turbine failure missile.26 g in the case of Ginna, the probability would be several orders of magnitude smaller (i.e., essentially zero) because the spent fuel pool is shielded from turbine missiles by the primary containment. 2.3 Partial Draindown of Pool Due to Refueling Cavity Seal Failures On August 21, 1984, the Haddam Neck Plant experienced a failure of the refueling cavity water seal, while preparing for refueling The water level I p SAA u:a % <e b l.4 es My

2-13 in the refueling cavity dropped by about 23 feet to the top of the reactor vessel flange within 20 minutes -- a loss of approximately 200,000 gallons, or a leak rate of about 10,000 gallons per minute.27 At the time of the event, refueling had not begun. The gates of the transfer tube connecting the re-fueling cavity to the spent fuel storage pool were closed. Although the seal fail'ure did not result in an accident or in the release of radioactivity, the incident raised the question of whether similar failures might occur while spent fuel was being transferred or while transfer gates to the spent fuel basin were open, either case of which might result in exposure of spent fuel to air and possible clad failure. All licensed plants were instructed to evaluate the potential for and D consequences of a refueling cavity seal failure.27 Refueling cavity seals, seal the gap between the reactor vessel flange and a flange on the inner periphery of the reactor cavity, or the floor of the cavity. BWR's have a permanently installed stainless steel bellows to seal the gap, and are, thus, not subject to failure of the Haddam Neck type. p d' Many PWR's seal the gap with gaskets held down by a bolted flat steel ( ring. Such systems have experienced difficulties in achieving tight seal be-y#1 ~ cause of surface irregularities and small vertical and concentric offsets in the two flanges. Consequently, many plants have converted to inflatable (pneumatic) rubber seals. Al so, i t should be noted that pneumatic rubber seals are often used to seal the gates in transfer tubes or canals. Licensee responses to the IE Bulletin indicate that the Haddam Neck cavi-ty configuration is unique in that the width of the annular gap between the reactor flange and the cavity flange is about two feet, whereas, in most plants the gap is of the order of <1" to -3". As of summer 1985 some 45 units used pneumatic seals in the refueling cavity.28 f i 1

2-14 Typical pneunatic seals are illustrated in Figures 2 1.10. There are many variations in the details of the designs, e.g., some plants have various types of retainers to support the rubber seals (e.g., see Figure 2.10), others rely on the rubber seal alone (e.g., see Figure 2.9). According to the re-sponses of the licensees, even if a pneumatic seal should deflate, the leakage would be expected to be small or negligible, because the wedged shaped upper section would naintain a g'ood seal (refer to Figure 2.8), i.e., the deflated seal would not distort enough under the hydrostatic head to extrude through the gap. Aside from the Haddam Neck 1984 incident, a few cases have been reported in which inflated seals have failed, either in the refueling cavity or trans-fer gates. None of these events had significant radiological consequences. Several such events are listed in Table 2.6. It is likely that this list is not exhaustive. To the best of the authors' knowledge no data base has been compiled (or is availabie) of the failure rate of pneumatic seals and their pressurizing systems of the types used in nuclear power plants, or of similar seals used in non-nuclear industries. Based on the limited experience cited in Table 2.6, the historical fail-ure rate in seals / systems is in the range of -1x10-2/Ry. Because of ad-vances in design, increased awareness and surveillance, the present failure rate is. estimated to be an order of magnitude smaller, i.e., -1x10-3/Ry. As is obvious from Table 2.6, a seal failure does not necessarily result in the rapid loss of water inventory from spent fuel transit or storage loca-tions. The limited experience indicates that the most probable time for a refueling cavity seal to fail is shortly after installation., while the cavity volume is being filled with water. According to the analyses supplied by licensees in response to IE Bulletin No. 84-03, the failure of a pneumatic refueling cavity seal in most PWR plants would not result in massive leaks because of the relatively narrow gap to be sealed and the geometric shape of the seal. Also, leaks from seal failures in transfer tube / canal gates would be limited, in most cases, because the leakage would be into a confined volume, e.g., from the storage pool into a drained up-ender sump. Taking l these factors into consideration, it is estimated that the frequency of a

2-15 i serious loss of pool water inventory resulting from a pneumatic seal failure to be in the range of -1x10-5/Ry. Evcn a large loss of water inventory from the spent fuel pool does not l necessarily result in uncovering and subsequent failure of fuel. Most spent fuel basins are constructed with weirs below the transfer gates which preclude complete drainage of the po'ol, even in the event of a catastrophic Haddam Neck type failure with the transfer tube / canal gates open. In most cases, the wa-ter level would remain a foot or more above the active zone of the spent fuel assecbli es. In a few cases, the upper several inches of the fuel could un-cover. (Note: Licensee responses to IE Bulletin 84-03 did not always provide infomation about the elevations of weirs and tops of stored assemblies.) In the event of a draindown of the pool to near the top of the fuel as-semblies, there would still be time (1/2 to I hour) to close gates and restore a supply of water before the residual water inventory reached the boiling point. However, as noted in one licensee response, even if the fuel remained covered " dose rate in the vicinity of the spent fuel pool would, however, be high, complicating recovery from the event.=2s A pool heatup event similar to the partial draindown scenario described above was considered by the NRC staff in Ref. 2. A conditional probability for failure to restore adequate makeup water was taken to be 5x10 2, based purely on judgement. Because of higher radiation levels in the partial drain-down scenario, it is estimated that the probability of failure to restore adequate makeup water to be somewhat larger, i.e., -1x10-1 Given all of the above, the probability of a pneumatic seal failure which results in exposure to air of stored spent fuel with resulting clad failure is esticated to be of the order of P - 1x10 6/Ry.

2-16 l 2.4 Pool Structural Failure Due to Heavy Load Drop i WASH-1400 considered the probability of structural damage to the pool due tv tr.e cropping of a fuel transfer cask (Ref. 3, pg.1-97). In the analysis, it was anticipated that one spent fuel shipment per week would be the equilib-rium shipping rate. The estimated rate for a drop resulting in pool failure x (for a single unit plant) w'as 4.5x10-7/Ry. The above frequency was based on a crane failure probability of 3x10-6 per operating hour. It was further assumed that each lift was of 10 minutes duration and for a 10 second period per lift the cask would be in a position to cause gross structural damage to the pool wall if a crane failure oc-curred. Human error was not considered. .(n/. M Since spent fuel is not currently being shipped, this hazard does not ex- #.45'f ist at the present time. However, at some point in the future, spent fuel 0%l,, will have to be removed from the reactor pools, either to some onsite storage 9,, ,f. ) facility, or eventually to a high level waste repository. At that time, the I,,) J~p frequency of removal of spent fuel will be correspondingly greater. , (p AM co lM' m .},q,l*W Orvis et al.2s have reexamined the cask drop probability and have used ,A c. - the following probabilities: gM, Mechanical failure of crane = 3x10 6/ operating hour O Electrical control failure of crane = 3x10-6/ operating hour Human error = 6x10 4/ lift. As can be seen, human error dominates the Orvis estimates for probability of a cask drop. The Orvis datum for human error was based on a study by Garrick et =1.30 wMeh concerned human reliability in the positioning of heavy objects. The applicability of the Garrick study to crane operations is not obvious. Nevertheless, a human failure rate in the range of 10-3 to 10-4 per operation appears to be consistent with data listed in the NRC handbook on human 31 reliability analysis for cases in which the operation has one or more people who serve as " checkers" and involves some degree of personal risk to the operating personnel.

2-17 Obviously, not all human failures associated with the lifting and moving of a spent fuel shipping cask would result in structural damage to the pool. The section of the pool where the cask is set down has an impact pad'to absorb the impulse of a dropped cask. Accidents in unloading the cask from or se W reloading on the transport vehicle would not involve the pool. pe A4 (x 4 A > Only horizontal moveme'nts of the cask above a structurally critical sec-g [w tion of the pool would pose the threat of structural damage. As noted above, Ar d WASH-1400 assumed that'~ the sensitive section is the vertical wall at the pool 6; 6' edge. It was implicitly assumed that all load drops on the pool edge would result in structural failure. This assumption appears to be too simplistic h.c and consequently too conservative for the following reasons: g! o many " load drops" would be partially attenuated by crane mechanisms which limit descent rates, and reduce impact energy, in case of some "off-center" hits, the full potential impact energy would not be absorbed by the pool edge (cask tilted, one end strikes floor first), and account should be taken of exterior cask fittings (e.g., cooling vanes) which absorb some impact energy. No rigorous structural analyses have been perfomed to scope the range of damage to a pool edge from a cask drop. In the absence of such analyses, it l has been necessary to estimate the conditional probability of catastrophic p7 structural damage given a cask. drop in the vicinity of the pool edge. It is estimated that the conditional probability is less than 100% and greater 1%. A conditional probability of 10% has been y) selected for the hazard calculation and 100% and 1% used for definTng the range of uncertain-ties. Since human error, rather than mechanical or electrical failure .pnears to dominate the hazard arising from shipping cask movements, the various steps in the crane operation have been identified in Table 2.7, which also lists the types of human error associated with each step. The distribution of failure

2-18 f requency in the various steps has been estimated and listed in the last col-unn of Table 2.7. (This distribution was subjected to " peer review" by BNL rigging personnel and managers who oversee operations of this type.) It will be noted that most steps in the crane operation do not jeopardize the structural integrity of the pool. Only in steps 5a and 5b (see Table 2.7) could the cask strike the ' pool edge. An accident of the type listed in 5a (horizontal movement with cask not high enough to clear the pool edge) would prebebly net cause serious damage because of the limited kinetic energy of the cask associated with the slow velocity of horizontal crane movements.

Thus, only step Sb in Table 2.7 is considered in the hazard calculation.

For purposes of calculating the cask drop hazard, i.e., the probability of catastrophic structural damage to the pool resulting from a cask dropping on the pool edge, the assumptions listed in Table 2.8 were used. Table 2.8 also lists the uncertainty ranges for each of the parameters. The results are as follows: ) M Probability of structural failure due to cask drop on pool edge caused by 4 mechanical or electrical failure of crane = 3.5x10 7/Rn pg W Probability of structural failure due to cask drop on pool edge caused by ' g human error = 3.1x10 5/Ry. If the failure rates summarized in Table 2.8 are assumed to be statis-tically independent, then the uncertainty in the overall failure rate is domi-nated by the uncertainty in the probability of pool failure. Thus the overall uncertainty is about a factor of ten in either direction. ?,5 Su~ury of Accident Probabilities Cu l The probability estimates made i Sections 2.1-2.4 are summarized in Table 2.9. These include only thos accidents thet result in the complete loss of pool water inventory. It ir be seen that shipping cask drop result-f l ing from human error and se mic induced failures dominate in the hazards. As WMH 0% -Tr M o C- ) INeleM C H Mh u I O t *'S A/sr fa Pm @ v ewf g 4 - / ep,

2-29 Table 2.9 Summary of Estimated Probabilities for Beyond Design Basis Accidents in Spent Fuel Pools Due to Complete Loss of Water Inventory Estimated Probability /Ry Accident Millstone Ginna Loss of Pool Cooling Capability 1.4x10 6 5.7x10 7* Seismic Structural Failure of Pool 2.2x10 5 1.6x10 5 Structural Failure from Tornado Missiles <1x10-8 <1x10-8 Structural Failure from Aircrash <1x10-10 <1x10 10 Structural Failure from Turbine Missile 4x10-7 -0** g. Loss of Pool Water Due to Pneumatic Seal Failure 0 1x10 6 Structural Failure from Cask Drop! 3.1x10-5 3.1x10-5 1After removal of accumulated inventory resumes.

• With credit for third cooling system.

Other PWRs which typically have two spent fuel cooling systems would have an estimated fuel uncovery freque,ncy of about 1x10-6/Ry.

• • Typical PWRs may have a failure frequency due to turbine missiles on the order of 4x10~7 but Ginna's pool is shielded from the turbine.

.\\[/qNr,[ (A /L vclues. bda>k

• I'^^^

7/ ~ / 3 uaw cn we um y, g& V y (W"A Cy( Qm$ O b < ~<7 y f,; / <,L g/ Il G *^'y or 7 )

3-1 3. EVALUATION OF FUEL CLADDING FAILURE 1 Two previous studies,2 have analyzed the thermal-hydraulic phenomena assuming a complete drainage of the water from a spent fuel pool. The pre-l vious section addressed the possible mechanisms for such an accident and pro-vided estimates for the accident frequency. This section provides a reevalua- ' tion of the basis for the SNL results.1,2 o,.;3;.I c,,. L n,.. g,, I @' ' ' ' ' ov'"'S J 3.1 Summary of SFUEL Results Ma' d W' N 4/ => Tr.4%,,, J J. m e-s* M The SFUEL code was developed by Benjamin et al'.1 to analyze the behavior of spent fuel assemblies after an accident has drained the pool. The results reported in Reference 1. indicated a wide range of decay power levels for which c' self-sustaining oxidation of the cladding would be predicted. Several limita-tions in the SFUEL.model were identified and addressed in a subsequent invas-tigation.2 But comparisons to small scale experiments were not very success-ful. 3.1.1 Summary Model Description The SFUEL code was developed at SNL and is described in Reference 1. Basically it is a finite difference solution of the transient conduction equa-F tion for heating of the fuel rods considering: The heat generation rate from decay heat and oxidation of the clad-ding. Pyy, Auss L es Radiation to adjacent assemblies or walls. A h n.w i " %s Ch IM/%st., s Convection to buoyancy-driven air flows. {[,~'{M /4E cm.a-n vnm ? Tha key assumptions in the analysis are: f om' I 1) The water drains instantaneously from the pool. f m we'n i 2) The geometry of the fuel assemblies and racks remains undistorted.

I 3-2 3) Temperature variations across the fuel rods are neglected.

4) The air flow patterns are one-dimensional.

5) The spaces between adjacent basket walls are assumed to be closed to air flow.

After the water is drained from the pool the. fuel rods heat up until the buoyancy driven air flow is sufficient to prevent further heating. If the decay heat level is sufficient to heat the rods to about 900*C, (1650*F) the oxidation becomes self-sustaining. That is the exothermic oxidation reaction provides sufficient energy to match the decay heat contribution and the temperature rises rapidly. O Reference 2 modified the SFUEL code to increase calculational stability and assess propagation of Zircaloy " fires" from high power to low power assem-blies. The SFUEL code was also modified by Pisano et al.2 to eliminate un-realistically high temperatures

• by non-mechanistically removing each node as it reaches the melting point of Zircaloy dioxide (2740*C or 4963*F).

In the present investigation, the oxidation cutoff has been reduced to 1900 C (The f W H Y ? N - -cy l.' L I melting point of Zircaloy 3450*F).

P m-W p

e,x s y e H c.- A /, > =s 6-p,I,),, n...... J Ie 7 3.1.2 C1ad Fire Initiation Results g .g.J. S' ***) b

• An extensive review of the cladding oxidation models used in SFUEL,1,2 is given in Appendix B:

1. The likelihood of clad fire initiation is not very sensitive to the oxidation equation. 2. The oxidation equation used in SFUEL is a reasonable representation of the data.

• Since the code does not explicitly treat melting of the cladding, tempera-tures as higt) as 3800'C were predicted.2

3-3 3. The likelihood of clad fire initiation is most sensitive to the decay heat level and the storage rack configuration (which controls the ex-tent of natural convection cooling). The critical conditions for clad fire initiation are summarized in Table 3.1. Note that for the old style cylindrical fuel racks with a large inlet orifice (3 inch diameter)~ the natural convection cooling in air is predicted to be adequate to prevent self-sustaining oxidation (cladding " fires") after 10 days of decay for BWR assemblies and 50 days for PWR assemblies. However for the new high density fuel racks, natural convective flows are so restrict-ed that even after cooling for a year there is potential for self-sustaining oxidation. As pointed out by Benjamin et al.1 there are a number of modifica-tions to the fuel rack design which would enhance convective cooling and re-duce the potential for cladding fires. However, the limited flow area of the high density designs make it difficult to ensure adequate cooling by natural convection of air. For the assumption of annual discharges, the critical decay time can be transl ated into a likelihood of cladding fire for a complete loss of pool water inventory. For the critical cooling times given in Table 3.1 the proba-bility of self-sustaining oxidation is approximately: h0 kl Do ya oo w M, 0.0 to 0.5 for BWRs with low density storage racks, g O. OI } yo @. l^' y ) W v .* J u.l (sW 0.0 to 0.7 for PWRs with low density storage racks, and L w' W bf 1.0 for PWRs with high density storage racks. 3.1.3 Clad Fire Propagation 1 The SNL investigations,2 of spent fuel behavior after a loss of pool integrity accident (assumed to result in complete drainage of the pool), iden-tified a range of power levels necessary for the initiation of self-sustaining clad oxidation and substantially lower power levels at which adjacent fuel bundles would oxidize once oxidation had been initiateo. However, the phenom-enology of propagation is not well understood and there -i-s-considerable j WAS I -7 'IWII /l kg F-c o,* jw s ~, ] Lx< _w 45sev W / n-o~. qi._m.

3-4 l uncertainty in these estimates. Benjamin et al.,1 Pisano et al.,2 Han3 and Johnsen" have pointed out a number of limitations in the previous analy-ses.1,2 In order I put the present results in perspective it is worth f mentioning the most important limitations; c / v l.. + 4e he, d. W 4 s ds ed. v, v.4.., kes : c-Lly 1. The oxidation egaation allows oxidation to continue beyond 1900*C (3450*F) where cl'ad melting and relocation is expected. PBF and KfK tests show clad relocation at temperatures in the range of 1900 to 2200 C but the analyses have cal cul ated temperatures as high as 3500*C (6330'F) without accounting for ciad and fuel melting. At such high temperatures the radiation heat flux becomes very large and it is believed that the potential for propagation to adjacent bundles will be overestimated, e -in erderi provide more realistic estimates of the potential for h oxidation propagation, BNL has chosen to terminate oxidation at the Zircaloy melting point. $ %Q d Ov " * >re d

• c a p e rs,J ak

] ZLe cM uM-l n.,n 2. The SFUEL code had not yet been validated successfully against fuel 2 rod oxidation data. A preliminary comparison against SNL data was only partially successful. The SFUEL code has been compared to the SNL data in a separate sec-tion (3.2) and key portions of the code have been validated. Specif-ically, the axial heat up (without oxidation) and the temperature at which self sustaining oxidation is reached has been validated. If a pKu low power spent fuel bundle heats up to within one or two hundred 'C hy of self-sustaining oxidation due to its own internal energy there is Rll 0+ u a high likelihood that the additional energy from an adjacent high power bundle will be sufficient to bring it to the initiation point. C' 6 'M i fpW o 3. The reaction rate equation has been criticized as being too low for long term exposure at low temperatures (when oxide layers may flake off and expose fresh Zircaloy). gver.hpendix B has shown that 1 de SFUEL calculations are not very sensitive to the low temperature r oxidation rate. l ) resp h.- 4 ) h 0

4 3-5 4. The lack of a fuel and clad melting and relocation model has also been criticized. 1 De/eiopment of realistic degraded fuel models is beyond the scope of the present investigation. However, we believe that the modified 4 2 SFUEL code (SFUEL1W ) has sufficient flexibility to estimate the im-portance of oxida' tion propagation. b ox An ia-- - % 04 m. tv le 3 p y y v ~,, y n neau w 3 \\ 5. Johnson criticized the clad failure criterion used in the SNL ana-f lyses.1,2 He noted that the clad failure could occur at tempera-gl tures as low as 650*C if the thermal loading is sustained for several humW h ouwvity P/wu h-we hours. p7 pw In view of the large uncertainty in the thermal behavior, we agree that a prediction of temperatures in excess of 650*C should not be viewed as successful cooling of the assembly. At these temperatures cladding failure and fission product release is very likely and the potential for cladding " fires" is high due to the effects of asymmet-ric heating (from adjacent high power bundles). %> caser de-e cN5 pa,..hl a C Lp 4, R <pW b an 1 4..~< ev-Propagation of cladding " fires" by particulate (i.e., spallation) or zir-conium vapor transport has been investigated and eliminated in an approximate separate effects study by Pisano et al.2 However, propagation due to the heat flux (radiation and convection) from adjacent bundles to occur even to very low power assemblies (at power levels correspon ing to 3 years of decays). >d Mo { d^y Mj Y 5pu zn ] e The purpose of this section is to establish the range of conditions for which propagation is predicted to occur. Both the power of the initiating bundle and the power of the adjacent bundles have been varied as well as the l ventilating conditions of the spent fuel building. I l yyfA-s 1 It should be emphasized that SFUEL does not address the question of Zir-j caloy oxidation propagation after clad melting and relocation. For recently Fe4 discharged fuel (less than 90 days), or for severely restricted air flow 4/h (e.g., high density PWR spent fuel racks) the oxidation reaction is predicted g l

3-6 i to be very vigorous and failure of both the fuel rods and the fuel rod racks is expected. Thus a large fraction of the fuel rods would be expected to f all to the botton of the pool forming a large debris bed. If water is not present in the bottom of the pool, the debris bed will remain hot and will tend to heat the adjacent assemblies from below. The investigation of debris bed for-mation is beyond the scope of the present study, but it appears to be an addi-tional mechanism for oxida' tion propagation. g y ,g 7 7 35 /r mceMM 7p f 3 Results As pointed out in Section 3.1.1,self-sustaining oxidation initiation is 8 not very sensitive to the oxidation rate equation but it is dependent upon the calculated air flow (related to flow resistance) and the power level. BWR v spent fuel with its low power density and open flow configuration must be re-cently discharged (within about 3 months) for self-sustaining oxidation to be initiated and unless it is a very high power bundle (discharged within 10 days or less) there is only a slight chance of propagation to older low power fuel bundles. rM5 6 g However, PWR spent fuel racks typically have a higher power density and g more flow restriction, thus self-sustaining oxidation-sen be initiated in fuel O that has been discharged for one year or more. ges;h%) Two fuel building ventilation conditions have been investigated as de-scribed below but it must be recognized that both of these assumptions corre-spond to very idealized conditions that are unlikely to be duplicated in an (actual accident. Rather these idealized conditions are provided to demon-strate theM of the various assumptions. For a beyond design basis seismic event, that catastrophically f ails the po,ol, it seems likely the f ail-ure of the fuel building may also occur. Mowe(e'r[ Benjamin et al.1 have shown 2 that a very large hole (at least 77 ft ) dust be opened ie=ned*r to approxi-p mate the perfect ventilation case.

3-7 Perfect Ventilation Under the perfect ventilation condition it is assumed that the fuel buildin5 is maintained at ambient conditions by a high powered ventilation i system (note that the flow rate must be much higher than typical gas treatment ( systems) or by a large opening (greater than 77 ft ) in the building. Oxygen 2 is' not depleted and the air entering the pool is assumed not to be heated' by I the hot gases exiting the fuel assemblies. The conditions necessary to initi-ate self-sustaining oxidation ur. der perfect ventilation condi' ions were sum-marized in Table 3.1 for three typical fuel rack configurations. Note that these are " borderline" conditions in that a slightly lower power level or a larger inlet hole size would be predicted to prevent self-sustaining oxidation from occurring. Note that the " critical" conditions outlined in Table 3.1 do o egle not imply that fuel rod failure is not predicted for power levels below these Fg conditions. The power level must be reduced substantially (about 20%) to en-v' sure that the predicted clad temperature is below 650'C (the minimum tempera-l ture at which clad failure and fission product release is likely to occur). 8 5 yf Na s*~y o F k For power and flow conditions that are only slightly below the " critical" conditions it should be obvious that the heat flux from a much higher power adjacent bundle would have the potential to push the "non-critical" fuel over the self-sustaining oxidation threshold. Thus the only real propagation ques-tion is. whether recently discharged (high power) spent fuel will radiate suf-ficient energy to initiate self-sustaining oxidation in low power fuel bundles l that have been cooled for one or more years. In this context two limitations of the SFUEL1Wi code should be noted: 1. The fuel storage racks are assumed to be immediately adjacent 50 that no air flow between racks is allowed. (The numerical approach used i to calculate the heat transfer is numerically unstable if fiow is allowed). j j 2. All fuel storage racks are assumed to be identical so that the ques-I tion of propagation from high power cylindrical racks to low power high density racks cannot be addressed. I

3-8 The first limitation probably represents current storage practices where a number of fuel pools are approaching their design capacity. However, the question of providing deliberate cooling channels between recently discharged f u p' and the older fuel cannot be directly addressed. Based on engineering insight, it appears that, under the idealized perfect ventilation conditions, the provision of an air space of 6 to 12 inches around the periphery of re-cently discharged fuel would minimize the likelihood of oxidation propagation to low power spent fuel assemblies. (Note that the code does allow for an air space adjacent to the pool walls and 6 to 12 inch'es is found to be adequate if flow through the bundles is not restricted.) Since high density fuel storage racks are predicted to cause sel f-sustaining oxidation even af ter storage for one or more years, it seems clear that it would be undesirable to store spent fuel in high density storage racks if it has been discharged within the last two years. (It may be worth noting that current practice restricts the storage density of low burnup fuel due to nuclear criticality considerations.) Thus the question of propagation from cylindrical fuel racks to high density fuel racks should be addressed, but the second limitation mentioned above precludes intermixing of the storage rack configurations. The propagation results with perfect ventilation are summarized in Table 3.2 for the high density rack configuration described in Reference 2. Note that the lowest power (11.0 kW/MTU) for self-sustaining clad oxidation corres-ponds approximately to fuel that has been discharged for one year, but the oxidation reaction will generate sufficient energy to propagate to a fuel bun-die that is about 2 years old (6.0 kW/MTV). For a fuel assembly that has been discharged for about 10 days (90 kW/MTU) the high decay heat level causes ex-tensive clad oxidation in the high power bundle and a somewhat higher propen-sity to propagate to low power fuel assemblies (as low as 5 kW/MTV which cor-responds roughly to a 2-1/2 year old discharge). The propagation results for a low density fuel rack (cylindrical) with a 3 inch diameter inlet hole is summarized in Table 3.3. Note that the range of power for the high power assembly is limited due to the improved free convec-tion within 'this type of fuel rack. Thus self-sustaining clad oxidation is

3-9 initiated at decay power levels at or above 30 kW/MTU (corresponding to about 90 days of cooling). Assuming that more than one discharge per year is un-likely, the adjacent low power assembly must be less than or equal to about 19 l agi!v ULO cays of cooling). Thus propagation only occurs for fuel that has been discharged less than 1 year with initiation from fuel that has been dis-charged within 2 weeks. For a PWR cylindrical fuel rack with only a 1.5 inch diameter flow hole, the air flow is much more restricted and the possibility of propagation is stronger as indicated in Table 3.4. For the 1.5 inch hole size propagation is predicted to occur for cooling times as long as two years. Inadequate Ventilation As previously mentioned the case of perfect ventilation implies a very high ventilation rate that is not normally possible. Benjamin et al.1 extend-ed the SFUEL code to consider limited heat removal to just keep the spent fuel building at constant pressure. Details of the modeling are described in Ref-erence 1, but the main result of the model is that the fuel building atmos-phere heats up (due to decay heat and the chemical energy of oxidation) and i found that the heat-up of the building in-the oxygen is depleted. Benjamin creased the likelihood of self-sustaining oxidation (i.e., decreased the decay power level necessary to initiate self-sustaining oxidation). This section is intended to address the question of whether limited ventilation also increases the likelihood of propagation to low power bundles. Table 3.5 provides a summary of propagation runs under inadequate venti-lation conditions. For the analyses the high power assemblies are modeled to represent approximately 1/3 of the core for 1000 MWe plant and the fuel build-3 ing is taken to have a volume of 150,000 ft. The results given in Table 3.5 indicate that propagation is no more likely with inadequate ventilation than with perfect ventilation. In fact propagation does not occur for several con-ditions listed in Table 3.5 for which propagation was predicted with perfect ventilation. Although this result is somewhat surprising, it is simply a re-sult of the oxygen depletion calculation. That is, the oxidation of the

3-10 1 recently discharged assemblies uses up the oxygen supply before the lower power assemblies can be heated to the point of self-sustaining oxidation. In view of the potential for fuel building f ailure due to either the assumed initiating event (e.g., a beyond design basis earthquake) or the rapid building pressurization from Zircaloy combustion and decay heat, BNL considers the oxygen depletion calculation to be unrealistic. Thus, in spite of the many uncertainties, the perfect ventilation model is expected to give the best approximation for the potential for propagation. Iqursrw : ka n o m /Fve1tw FWc 7~ sp A o uaur Am o u e A C.s ' vwn ' ot o ? Conclusions Regarding Propagation

f. g. o m g n> a w w m n>,t m ~.r ranc au c. '

l. 4[3l 2 3,l4jf N"*'

Based on the previous results we have concluded that the modified SFUEL 2 code (SFUEL1W ) gives a reasonable estimate, of the potential for propagation of self-sustaining clad oxidation from high power spent fuel to low power spent fuel. Under some conditions, propagation is predicted to occur for spent fuel that has been stored as long as 2 years. The investigation of the effect of insufficient ventilation in the fuel building indicated that oxygen depletion is a competing factor with heating of the building atmosphere and propagation is not predicted to occur for spent fuel that has been cooled for more than three years even without ventilation. These results are in general agreement with the earlier SNL studies.1,2 3.2 Validation of the SFUEL Computer Code 1 The SNL investigations,2 of spent fuel behavior after a loss of pool integrity accident (assumed to result in complete drainage of the pool), iden-tified a range of power levels necessary for the initiation of self-sustaining clad oxidation and substantially lower power levels at which adjacent fuel bundles would oxidize once oxidation had been initiated. However, an attempt 2 to validate the code was orsly minimally successful in that the post-test ana-lyses were able to match the heat-up rate in helium (without oxidation) but the SFUEL code over-estimated the temperature transient after air was intro-duced.

i ) l 3-11 1 I The objective of this section is to use the revised 3 oxidation rate equa-tion in SFUEL to analyze the SNL small scale tests to aid in validating the SFUEL code. The SNL tests are described in Reference 2, but in order to put Len results in perspective several important conditions should be high-t.n e li5hted: 1. The test was of a small bundle of electrically heated rods (9 rods) with a short length (38 cm). 2. In order to achieve self-sustaining clad oxidation (~>850*C) the rods were heated with a very low fluw rate of helium before air was admit-ted to the test assembly. O Under these test conditions the dominant heat loss is via radiation whereas for tht; postulated accident the dominant heat loss is via free convec-tion. These test conditions lead to laminar flow (a Reynolds number of about 100) in which oxygen diffusion to the cladding surface limits the reaction rate. Only one test (6) had a sufficiently high air flow rate to allow vig-ourous oxidation. Since the free convection and radiation calculations in SFUEL,2 were 1 inappropriate to the test configuration, Pisano et al.2 created a stripped down version called CLAD 2 which used a matrix inversion routine to calculate radiation losses. After several preliminary attempts to analyze the helium portion of the tests we concluded that there were several errors which led to underestimation of the convection portion of the heat losses. Since helium has a much higher heat capacity and conductivity than air it appears to contribute to establish-in; the initial conditions. In order to provide an adequate simulation of the j initial steady-state portion of the test we made two modifications to the CLAD I code: i 1. Include helium properties with a switch to air properties at the start of the transient, i

3-12 2. Include an energy balance on each gas control volume to force conser-vation of energy. With these changes we were able to obtain an adequate simulation of the i initial portion of the tests. Using this revised version of CLAD with the Weeks' oxidation correlation,3 analysis of both the helium and the air por-tions of the test looked r'easonable, but still tended to over-predict the' peak temperatures during oxidation. In order to bring the calculations into rea-sonable ' agreement with the small scale data the Weeks' correlation has been reduced by a factor of four (note that this corresponds approximately to the data scatter). Results o The revised CLAD code has been used to analyze the SNL small scale exper-iments Tests 4, 5 and 6. The other three tests were intended to simulate propagation with nonuniform heating and structures that CLAD was not capable of modeling. The CLAD results for Test 4 are compared to the data in Figure 3.1. These results still tend to overpredict the temperature in the center of the test rod, but give reasonably good agreement at the top of the rod where radiative heat iosses are large. The peak temperatures calculated by CLAD are summarized in Table 3.6 and compared to the peak measured temperatures for the three tests. Note that CLAD still overpredicts the peak temperature for the low flow rate test (4 and

5) but gives good agreement with the high flow rate tests where adequate oxy-gen is available.

It should be noted that this " oxygen starvation" phenomenon appears to be a result of the extremely low laminar flow where oxygen must diffuse to the clad surface. CLAD includes an oxygen depletion calculation but assumes that all the oxygen in each volume is immediately available at the surface. 3.3 Conclusions Regarding SFUEL Analyses After an extensive review of the SFUEL code and comparison to the SNL small scale experiments, BNL concludes that the code provides a valuable l l

3-13 tool for assessing the likelihood of self-sustaining clad oxidation for a variety of spent fuel configurations assuming that the pool has been drained. The SNL cmall scale data provida a reasonable degree of validation for the heat-up and oxidation models, but the results are extremely sensitive to the natural convection calculation which has not been v611 dated. When oxidation is terminated at the Zircaloy melting temperature (assum-ing that the molten Zircaloy is relocated), oxidation propagation only occurs for spent fuel bundles which are already approaching the " critical" conditions for self-sustaining oxidation (see Table 3.1). However, this finding does not mean that oxidation propagation is unlikely. On the contrary, for some high density storage configurations the " critical" conditions are approached for spent fuel that has decayed for two to three years. Thus clad " fire" propaga-tion appears to be a real threat but the basic question remains as to what are the " critical" conditions for 1nitiation of oxidation and what the uncertainty is for a given spent fuel configuration. The critical conditions are summar-ized in Table 3.1 for several typical spent fuel racks. While the heat-up and oxidation models have been validated to a limited extent by the SNL data (see Section 3.2), the authors believe that the largest source of uncertainty is in wm7 the natural convection flow rate. It is recommended that these free convec-Ig ~ W NT tion flow calculations be verified against large scale data. Preferably the ,,s; data would be obtained from spent fuel assemblies in typical storage racks (both high and low density). O hg. g e-' 3.4 References for Section 3 1. A.S. Benjamin, D.J. McCloskey, D. A. Powers, S. A. Dupree, " Spent Fuel Heat-up Following Loss of Water During Storage," NUREG/CR-0649, March 1979. 2. N.A. Pisano, F. Best, A.S. Benjamin, K.T. Stalker, "The Potential for Propagation of a Self-Sustaining Zirconium 0xidation Following loss of Water in a Spent Fuel Storage Pool," Draft Report, January 1984 3. J.T. Han, Memo to M. Silberberg, USNRC, May 21, 1984 4 G.W. Johnsen, Letter to F.L. Sims, EG8G, Idaho, April 4,1984

1 4-1 4 CONSEQUENCE EVALUATION A PWR and a BWR reactor were selected for risk evaluation based on a pre-liminary screening l of perceived vulnerability and the spent fuel poo'l inven-to ry. The reactors selected were Ginna and Millstone 1. Both are older plants that were built before the current seismic design criteria were promul-gated and have relatively farge inventories of spent fuel. 4.1 Radionuclides Inventories The radionuclides inventories for both the PWR and BWR pools were calcu-lated using the ORIGEN2 Computer Code 2 for the actuhl operating and discharge histories for Ginna and Millstone 1. The ORIGEN2 program in use at BNL was g verified by comparison with results obtained at ORNL for identical cases.3 A description of the assumptions and methods of analysis is given in Appendix A along with the detailed results for each species. The results for the risk significant species are summarized in Table 4.1 (Millstone 1) and Table 4.6 (Ginna). For both plants, the noble gases and halogens in the spent fuel inventor-ies are a small fraction of the inventory in an equilibrium core at shutdown except for freshly discharged fuel, but cesium and strontium are more than three times the equilibrium inventory (see Tables 4.1 and 4.6). 4.2 Release Estimates The fission product release fractions have been calculated for two limit-b*c ing cases in which a Zircaloy fire occurs: In Case 1, the clad combustion is , (h assumed to propagate throughout the pool and the entire inventory is in-volved. In Case 2 only the most recently discharged fuel undergoes clad com-bustion. I The release calculations for Cases 1 and 2 make the assumption that if the spent fuel pool suffers a structural failure, coolant inventory will be totally drained, i.e., the leak rate will greatly exceed makeup capability

4-2 even if the coolant systems are still available. The probability of Zircaloy fire and fission product release has been determined from BNL calculations de-scribed in Section 3. imer kr a cladding fire to occur the fuel must be 0 recently aischarged (about 10 to 150 days for a BWR and 30 to 250 days for a l PWR). This leads to a conditional probability for a Zircalcy fire of for a B'.iR and.40 for a PWR. If the discharged fuel is put into high density racks the critical cooling time is increased to one to three years and the conditional probability of a Zircaloy fire is increased to a virtual certain-q ty. A reevaluation of the cladding fire propagation estimates indicates that ) there is a substantial likelihood of propagation to other fuel bundles that have been discharged within the last one or two years. Subsequent propagation g to low power bundles by thermal radiation is highly unlikely, but with a sub- ] stantial amount of fuel and cladding debris on the pool floor, the coolability of even low power bundles is uncertain. 0W 4.2.1 Estimated Releases fcr Set f-Sustaining Cladding 0xidation Cases (Cases DD 1 and 2) WQ <-asq 'Ty, to t M,7 IV Cn Ty,. r.- I b", Q1~ As discusse in Section 3.1 there are broad range of spent fuel rage 0,4 conditions for which self-sustaining ox) tion of the cladding . occur if the water in the pool is lost. ifr Ginn)with high dens racks the condi-gy; tional prob, ility of a cladding " fire" is redipetr to be nearly 100% while for Mills 6ne pthe probability is abou 204. If self-sustaining oxidation g, occurs the fuel rods are predicted to reach 00 to 2100 C over a substantial b'# portion of their length. At these temperatures, the release fraction is pre-dicted to be substantial. ., } Rcqh estimates of the fractional release of various isotopes have been ] presented in an attachment to Ref. 4 Included in the estimates were noble gases (100%), halogens (100%), alkali metals (100%), tellurium (2 to 100%), ] bariu:. (2%), strontium (0.2%) and ruthenium (0.002%). i 1 Estimated release fractions of other isotopes are given in Table 4.2. These estimates are based on various considerations, including experimental

4-3 { data (tellurium), location of the isotopes (whether in cladding as activation products or in fuel pellets as fission products), and melting / boiling points of the el+ nent or oxides of the element. Comments on the estimates list ed in Table 4.2 follow: Tellurium: The releas,es shown assume the lower limit of Ref. 4 based on the tellurium

• release model recently proposed by Lorenz, et al.5 The low release value assumes that a fraction of the Zircaloy cladding relocates (melts and flows downward) before oxidation is j

compl ete. s j Alkali Earths: Because of the high boiling points of the oxides of Sr and Ba, it is estimated that only a very small fraction (2x10-3) of these elements of fission product origen in the fuel pr11ets es-cape. It is estimated that 100% of the activation product Sr-89 and Y-91 contained in the Zircaloy cladding are released as aerosols. Transition Elements: It is estimated that 100% of the transition element activation products contained in the cladding are levitated as aerosols of the oxides (smoke). Note that the small release frac-tion of Zr-95 (0.01) takes into account the large inventory of fis-sion product Zr-95 trapped in the fuel pellets. It is assumed that only 10% of the activation products in the assembly hardware escapes (see Table 4.2, Fe-55, Co-58, Co-60 and Y-91). The Co-60 fraction is corrected for its small content in the cladding. Antimony: It is estimated that 100% of the SB-125 is roasted out of the fuel pellets, because of its high mobility. Lathanides and Actinides: A negligible release of the oxides of the 1athanides and actinides is estimated because of their chemical sta-bility, low vapor pressures and ceramic characteristics. 1

4.a Case 1: Case 1, the " worst" case, assumes an accideat that results in a Zircaloy fire that propagates throughout the entire spent fuel in-ventory in the pool, and that the accident occurs 30 days after the s eanor was shut down for discharge of the last fuel batch. The esti-mated releases of radionuclides are listed in Table 4.3. These were cbtained by combining.the "30-day" inventory given in Column 3 of Table 4.1 with the release fractions listed in Table 4.2. Case 2: Case 2 assumes an accident that results in a Zircaloy fire that involves only the last fuel batch to be discharged, and that the accident occurs 90 days after the reactor was shut down for fuel dis-charge. The estimated releases of radionuclides are listed in Table 4.4 These were obtained by combining the inventory in the last fuel batch (data tabulated in Table A.6 of Appendix A) with the release fractions it Table 4.2. 4.2.2 Estimated Release for low-Temperature Cladding Failure (Cases 3 and 4) For a less severe accident in which fuel is exposed to air but does not reach temperatures at which a Zircaloy fire ignites, it is assumed that the cladding on many fuel rods will fail (i.e., develop leaks) resulting in a re-lease limited to the noble gases and halogens. Two limiting cases have been considered: ' 15 Case 3: in which the entire pool is drained but the decay time since the t l Y4 last discharge is one year, and 507, of the fuel rods suffer clad rupture. gpv 1 Case 4: in which the pool drains to a level that exposes the upper por-J I tion of the fuel assemblies, the decay time for the last discharged fuel 30 DM b:tch is 30 days, no Zircaloy fire occurs but all of the fuel rods in the / Mre#' last discharged batch rupture. f pp ys (/ M s"7 The estimated releases for Cases 3 and 4 are given in Table 4.5. g4 TlW,

4-5 4 4.3 Off-Site Radiological Consequences 4.3.1 _ Scenarios for Consequences Calculations The off-site radiological consequences have been calculated using the CRAC2 computer code.6 The scenario used in the CRAC2 calculations consisted of the following conditions: 'vam 40A a generalized site surrounded by a constant population density o@ % persons per square mile; e se n s. generalized meteorology (a uniform wind rose, average weather condi-tions);and O gpg 4,, - omisw .f n.w a 'r m n er st/STd.V 7 the population in affected zones was relocated after 24 hours. The radiological effects were calculated out to a distance of 50 e.c 500 f milesy cu u A'h rAJ h a-f** A

• W-4 P/. - J -

y w CRAC2 calculations were made for a range of possible releases as de-scribed in Section 4.2. The consequences are summarized in Table 4.,7. 4.3.2 Consequence Results There are several unusual characteristics of a spent fuel accident that cause somewhat surprising results in the radiation exposure calculations. Specifically, the radiation exposure is insensitive to fairly large variations in the estimateo release. This is due principally to the health physics assumptions within CRAC. For the long lived isotopes (predominantly cesium), the exposure is due mainly to exposure after the area is decontaminated and people return to their homes. The CRAC code assumes that decontamination will limit the exposure of each person to 25 rem. Thus, for this type of release the long term whole body dose is limited by the population in the affected sectors (about 0.8 million people in the 16 sectors for a 50 mile radius) to 6 about 3x10 person-rem (only 3 of the 16 sectors are downwind).

4 4-6 The extreme cases (IA; immediately after refueling and 1B and IC; with the total fuel pool inventory involved) result in much higher releases but no significant change in population dose. 1 A more sensitive indication of the consequences for a spent fuel accident is the interdiction area (the area with such a high level of radiation that it is assumed that it cannot 'ever be decontaminated). As indicated in Table 5 the worst spent fuel accident is calculated to result in an interdiction area of 224 sq. miles. d about orders of ma Mdhigher an t W ~ 41 # f, int 1rrdict ttm--are computed for eactor core mel cidents out 1 10 se" mW CA [ 70 po nM 4.4 References for Section 4 O 1. BNL Memorandum, V.L. Sailor and K.R. Perkin to W.T. Pratt, " Study of Be-yond Design Basis Accidents in Spent Fuel Pools," May 81985. 2. A.G. Croff, "0RIGEN2: A Versatile Computer Code for Calculating the Nu-clide Compositions and Characteristics of Nuclear Materials," Nuclear Technology, Vol. 62, pp. 335-352, September 1983. 3. Internal Memorandum, Brookhaven National Laboratory, from V.L. Sailor to R.A. Bari, " Comparison of BNL ORIGEN2 Calculations with ORNL," May 27, 1986. 4. Memorandum of J.T. Han to M. Silberberg, " Response to a NRR Request to Review SNL Studies Regardhg Spent Fuel Heatup and Burning Following Loss of Water in Storage Pool." U.S. Nuclear Regulatory Commission, (May 21, 1984). 5. R.A. Lorenz, E.C. Beahm and R.P Wichner, " Review of Telliurf om Release Rates from LWR Fuel Elements Under Accident Conditions," Proceedings of the International Meeting on Light Water Reactor Severe Accident Fvalua-tion, August 28-September 1,1983, pg. 4.4-1, American Nuclear Society Order 700085, ISBN 0-89448-1112-6. i

4-7 6 L.T. Ritchie, J.D. Johnson and R.M. Blond, Calculations of Reactor Acci-dent Consequences Version 2, CRAC2: Computer Code, User's Guide, pre-pared by Sandia National Laboratories for the U.S. Nuclear Regulatory C =nission, NUREG/CR-2326 (SAND 81-1994), February 1983.

5-1 5. RISK PROFILE The likelihood and consequences of various spent fuel pool accidents has itsr. Estimated in the previous sections. The risk is summarized in Table 5.1. As previously mentioned, the exposure results are tied to the health physics assumptions regarding decontamination and maximum allowable exposure. Thus the land interdiction' area is included in Table 5.1 as a more meaningful representation of severity. The uncertainty in each of these risk indices is estimated to be an order of magnitude in either direction and is due princi-pally to uncertainty in the fragility of the pools and uncertainty in the seismic hazard, Note that the risk results are calculated for two surrogate plants and j'? C; w c 6V' e'# may not be applicable to generic pool types. Q s Jesch 5.1 Failure Frequency Estimates W. ]I 5.1.1 Spent Feel Pool Failure Probability The likelihood of the various postulated spent fuel pool accidents was developed in Section 2 and summarized in Table 2.9. The probability is simi-lar to the frequency of dominant core melt sequences for many PRA's. The major contributors are: 1. Cask drop accidents, 2. Seismic induced pool failure, 3. Loss of pool cooling, and 4. Pneumatic seal failure. Note that all of these potcntial accidents are plant specific and their frequency will vary widely from plant to plant. In particular. BWR's do not have pneumatic seals so their failure frequency is zero, h~CO f

5-2 5.1.2. Spent Fuel Failure Likelihood Previous investigations 2 of spent fuel behavior after a less of pool integrity accident focused on the conditions necessary to initiate cladding " fi re s" after a spent fuel pool has drainen. The present project has reevaluated these conditions using the SFUEL code develope ' by SNL. The 2 likelihood of such cladding fires has been assessed in Section 3. For a PWR with high density storage racks, the conditional probability of a clad fire was found to be 1.0 while for a BWR wit ow density storage racks the proba-bility of a clad fire was found to b 0.08 p p w yy g ygg (, g 0.d 5.2 Conclusions Regarding Risk O' N )* o The overall risk due to beyond design bas 1s accidents in spent fuel pools for the PWR surrogate plant is about 130 person-rem /Ry and about 12 person-pe w.:m ofp a rem /Ry for the BWR surrogate. These estimates are c^ p:rs.le te pre 5eni. eni-- [ i 3 A

t:0 for avminant ceic meli enidenis end eppeer to warrant further atten-tion on this basis alone.

However, the unique character of such an accident }; (substantial releases of long lived isotopes) makes it difficult to compare to ? reactor core melt accidents. The exposure calculations are driven by assunp-f tions in the CRAC modeling and the results are not sensitive to the severity rp of the accident.hn terms of interdiction area this type of accident has the T* potential to be much worse than a reactor core melt accidenh un( C 1 The uncertainty in risk in terms of person-rem /Ry is driven principally by the uncertainty in the likelihood of complete draining of the spent fuel pool which is estimated to be at least an order of magnitude in either direc-tion. 5.3 References for Section 5 1. A.S.

Benjamin, D.J.

McCl os key, D.A.

Powers, S.A.

Dupree, " Spent Fuel Heat-up Following Loss of Water During Storage," NUREG/CR-0649, March 1979. l 1

5-3 2. N.A.

Pisano, F.

Best, A.S.

Benjamin, K.T.

Stalker, "The Potential for Propagation of a Self-Sustaining Zirconium 0xidation Following Loss of Water in a Spent Fuel Storage Pool," Draft Report, January 1984 3. " Evaluation of Severe Accident Risks and the Potential for Risk Reduc-tion," HUREG-1150 (To be published). O

5-4 Table 5.1 Estimated Risk for the Two Surrogate Spent Fuel Pools from the Two Dominant Contributors Spent Fuel Interdiction Accident Pool Fire Health Risk Risk Initiator Probability /Ry (Man-rem /Ry) (Sq. Mi./Ry) Seismic induced 1.6x10-5 37 8.4x10 4 PWR pool failure Seismic induced BWR pool failure 1.8x10-6 4 7.6x10-5 Cask drop

• induced PWR pool failure 3.1x10-5 71

.001 Cask drop

• induced BWR pool failure 2.5x10-6 6

1.1x10-4

• Af ter removal of accumulated inventory resumes.

(Note that many new plants have pool configurations and administrative procedures which would preclude this failure mode.) fk SU D NO T ga y w p co,ur Puuma n., + r~. m n n< n y (h'IT Ins I "' ' T *

• F m.-

f-p L. G G-H pa 7 2 Qvw c mw o

6-1 6. CONSIDERATION OF RISK REDUCTION MEASURES Due to diversity in the nature of initiating events for beyond design basis accidents in spent fuel pools, there appear to be several possible ways to reduce the risks. It inust be emphasized that each of the contributors to risk are plant specific and one or more of the risk significant sequences identified in Section 5 may not be important at other plant sites. The following sections discuss the advantage and disadvantages of a number of proposed risk reduction strategies. A cost benefit analysis has not been performed but the estimated risk appears to be large enough to justify further investigation of risk reduction measures. 6.1 Risk Prevention v 1. Reduction of Stored Radioactive Inventory - Most of the consequences of a release of radioactivity from a catastrophic pool accident is associated with the large inventory of isotopes of intermediate half-lives, e.g., Cs-137, Sr-90. Th"e potential release increases approxi-mately in proportion to the number of fuel assemblies in the storage i nventory. One obvious measure for risk reduction is to transfer part of the inventory to alternative storage locations (e.g., see Ref. 1). 2. Air Circulation - The one universal prevention measure is to promote air cooling in the event of loss of water cooling of the spent fuel. The new high density fuel storage racks restrict air flow and make even old spect fuel (one to two years) susceptible to heat-up and self-sustaining oxidation. The older style fuel baskets with large inlet holes (3 inch diameter or more per assembly) allow much freer air circulation. If all recently discharged fuel (less than two years) is kept in low density fuel baskets and they are separated from the wall and the older fuel by a one foot gap then the lik hood of self-sustaining oxidation would be reduced by.a fpctor of 5 or more compared to the high density storage configuration, i pfq& Y )(lD

6-2 3. Additional Cooling Systems - Although loss-of-pool cooling appears to be risk significant, an additional cooling system is unlikely to be cost beneficial (unless the cooling system was substantially more unreliable than the two surrogate systems) An additional cooling sht system would not affect the risk from poop, failure events (seismic or cask drop accidents _). Thus the net risk reduction would be minimal unless loss-of-cooling were the dominant event. 4 Improved Procedures and Eauipment - The likelihood of cask drop acci-dents can be reduced by improving procedures, administrative controls and/or installing more reliable equipment. However, none of these improvements would reduce the risk from the other dominant se-Thus the net risk reduction would be difficult to quantify quences. O on a plant soecific bas tM Iuf! Ip'pe @ N 'be fefM IcYr. N t & s p'.:t: risk evaluation before spent fuel shipment is begun at each site. A key piece of such an evaluation would be a structural analy-sis of the pool response to the loading from a dropped cask Ph 4-Tpp. (%k - f 6.2 ' Accident Mitigation 1. Post-Accident Spray - Water spray has the potential to terminate the progression of a spent fuel pool accident whether or not the pool is intact. However, large quantities of water must be available (it l would be necessary to continue spraying until the pool could be repaired and reflooded) and the eluipment would have to be seismical-p..s ly qualified to a higher (9) level than the pool structure (in order r j for the sprays to have a high likelihood of surviving). Some pools may have fire sprays available in the spent fuel pool building. For those plants without sprays available, it seems unlikely that the ex-pense of a new safety grade spray system could be justified consider-ing the 'large uncertainty in the risk. Temporary fire hoses were suggested by Benjamin et al.,2 but the radiation levels would make such ad hoc measures extremely difficult. Furthermore, if the spray is not initiated before the rods reach 900 C or there is insufficient flow, the water may aggravate the reaction by providing additional oxidation. (The steam /Zircaloy reaction is also highly exothermic.)

6-3 2. Filtering - For those plants with a standby gas treatment system available, operation of the system has the potential to substantially reduce the fission product release from the building. However, the high temperatures and large aerosol production rate would tend to rapidly degrade the ef festiveness of the system. The performance of such a filtering system would be difficult to characterize under fuel pool accident coriditions. It is unlikely to be cost effective to install a new system large enough to handle the worst case spent fuel s pool accident scenarios. ) 6.3 Conclusions Regarding Preventive and Mitigative Measures For those plants which have a similar spent fuel pool risk potential O to the two surrogate plants, the one preventive measure which appears to have a substantial effect on risk (a risk reduction of 5 ac more) is to maintain recently discharged fuel in low density storage racks that are isolated from the rest of the fuel racks by a foot or more of space (to provide free air circulation). However, there may be plant specific features which make a sub-stantial difference in the order of the dominant contributors to risk. There-fore plant specific risk evaluations should be performed before any changes are implemented at a given plant. 6.4 References for Section 6 1. D.D. Orvis, C. Johnson, and R. Jones, " Review of Proposed Dry-Storage Con-cepts Using Probabilistic Risk Assessment," prepared for the Electric Power Research Institute by the NUS Corporation, EPRI NP-3365, February 1984. 2. A.S. Benjamin, D.J. McCloskey, D. A. Powers, S. A. Dupree, " Spent Fuel Heat-up Following Loss of Water During Storage," NUREG/CR-0649, March 1979. l I (w o o% r 3. F et Fi,a ge l...q br H. J.i es - Ce g c la ru t, a.. k. A At bou-l de ~ - s l le 'l A~') 4G. in i r C he A, I ,e k,g.f,,(.fs,, m.m as was ~.a.s (u / 4.,,..,,. 4. ). r~L a 6 &~' Iw-w~ p{helh.,.$ven,.- ,,4 e,.s, y..u e~y y 4 -..,...s 17 L e.,l. Q e h. ) c o. l.I a llw /- < v.* a t le. 1(ed -F.a',.t ,F,7 a & 'yk~ ( u rdh *

[NC 4,G 304 ([ ] REGULATORY INFORMATION DISTRIBUTION SYSTEM (RIDS) .CCESSION NDR: 8701090416 DDC.DATE: 86/12/R3 NOTARIZED: NO DOCKET

• FACIL:50-321 Edwin I.

Hatch Nuclear Plant, Unit 1, Georgia P ower C 05000321 50-366 Edwin I. Hatch Nuclear Plant, Unit 2, Georgia Power C 05000366 AUTH.NAME AUTHOR AFFILIATION GUC WA, L. T. Georgia Power Co. RECIP,NAME RFCIPIENT AFFILIATION GRACE J.N. Region 2, Office of Director SUB JECT: Further response to IE Bulletin 84-03 re refueling cavity I water seal. Clarifies 840927 response re pneumatic seals used i at plant hatch associated w/ fuel pool gates & expansion Joint between Units 1 & 2 in refueling canal. DISTRIBUTION CODE: IE11D COPIES RECEIVED: LTR _. ENCL _ SI ZE: TITLE: Bulletin Response (50 DKT) l NOTES: RECIPIENT COPIES RECIPIENT C OP IES j ID CODE /NAME LTTR ENCL ID CODE /NAME LTTR ENCL ? C7 BWR PD2 LA 1 0 BWR PD2 PD 1 RIVENDARK,C 1 JTERNAL: ACRS WYLIE 1 1 AEOD 1 AEOD/PTB 1 1 IE FILE O2 1 IE/DEPER/EGCB 1 1 IE/DEPER/EGCB/S 2 O NRR BWR ADTS 1 1 NRR JOYCE J 1 1 NRR PWR-A ADTS 1 1 NRR fM L n 1 NRR/DSRO/EIB 1 1 cNRR/wKv/RS 1 NRR/ORAS 1 RGN2 01 1 (TERNAL: LPDR 1 NRC PDR 1 1 NSIC 1 t i i ^ /.Qf 'r \\.) ; FILE 83I-8' M7cg vea4r "O

gf Gr; a 0 we+ Comca,y 123 0,e:toat A.e vo - ** a*:t Ger; a 3:3D&

• • e;*: 4 40.: 9 06526 va u; AO: ess Ms: C".ce Sa 49!

C 6t:a Geo'; a 30302 Georgia Power ba l;%la..a. sa<ew - DEC31 nij:n6 """ n ' i l a-n.ce 5 ; SL-1753 0560H 1 l December 23, 1986

REFERENCE:

U.S. Nuclear Regulatory Commission RII: JNG Office of Inspection and Enforce ~ nt 50-321/50-366 Region II - Suite 2900 IE Bulletin 84-03 101 Marietta Street, N. W. Atlanta, Georgia 30323 e-ATTENTION: Dr. J. Nelson Grace Gentlemen: Georgia Power Company (GPC) provided a response to IE Bulletin 84-03, 27, 1984. Our Refueling Cavity Water Seal, in a letter dated Septenber response stated that the only pneunatic seals used at Plant Hatch were associated with the fuel pool gates and the expansion joint between units 1 and 2 in the refueling canal. We further stated that these seals were redundant. In order to avoid any possible misunderstand 1ng of our 27, 1984 bulletin

response, the following points of September clarification are provided for your information.

IE Bulletin 84-03 discussed a condition where one pneumatic seal was The seal the only protection to prevent a refueling cavity water leak. arrangement at Plant Hatch is clearly redundant in that failure of one of the installed pneumatic seals (due to puncturing or rupturing), will not prevent the other seals from properly functioning. However, even though the pneumatic seals were redundant, the transfer GPC canal expansion joint pneumatic seals were not single failure proof. has taken actions recently to upgrade the design to include the capability to sustain loss of air to one set of seals witho function. supply from each, unit), such that if the air supply is lost from one unit, the other unit's air supply should properly maintain the function of the joint seals to prevent water loss from the transfer canal. f. j e M+M 904t6 B61223 FDR ADDOK 05000321 0 PDR

~ Georgia Power d U.S. Nuclear Regulatory Commission Office of Inspection and Enforcement Region II - Suite 2900 December 23, 1986 Page 2 If you have any questions regarding this matter, please feel free to contact this office. Sincerely, f 6.k' . L. T. Gucwa JCJ/jhu c: Georgia Power Company U.S. Nuclear Regulatory Commission Mr. J. P. O' Reilly Mr. P. Holmes-Ray, Senior Resident Mr. J. T. Beckham, Jr. Inspector - Hatch Mr. H. C. Nix, Jr. GO-NORMS e 1 1 0560H

Fw G T-h-f, December 4,1986

PRELIMINARY NOTIFICATION OF EVENT OR UNUSUAL OCCURRENCE PNO-II-86-90 This preliminary notification constitutes EARLY notice of events of POSSIBLE safety or

-public interest significance. The infomation is as initially received without verifi-cation or evaluation, and is basically all that is known by the Region Il staff on this j date. FACILITYi. GeoIgia Power Company I Licensee Emergency Classification: L. Hatch Units 1 and 2 i Notification of Unusual Event Docket Nos. 50-321/366 i Alert Baxley, Georgia Site Area Emergency , ~ L General Emergency X Not Applicable AITDISPATCHEDTOINVESTIGATkLEAKFROMSPENTFUELP00k SUBJECTi Region II has dispatched an Augmented Inspection Team (AIT) to the Hatch site to investigate the leak of about 50,000 gallons of water from the Hatch Units 1 and 2 spent fuel pools. Georgia Power discovered the leak at 10:02 p.m. (EST) yesterday, when it was found that water had spilled into an outside area between the Unit I and Unit 2 reactor buildings. Hatch Unit 1 is operating at 100 percent power; Hatch Unit 2 has completed a refueling cutage, but is in cold shutdown. An immediate investigation by Georgia Power disclosed that the ' leak detection annunciator failed to alarm and that the spent fuel pool levels had dropped about five feet. The pools are built so that they cannot be completely drained and the fuel uncovered. Although the pool levels dropped by five feet, the levels did not go below the technical specification limits. Of the approximately 50,000 gallons of water which leaked, between 5,000 and 10,000 gallons were released through the storm drain system to a swampy brea within the owner-controlled property. Plant personnel are building dikes and taking other steps to contain this water. Georgis Power believes no contaminated water has entered the nearby Altamaha River. Georgia Power also believes that leak may have been caused by a loss of air to inflatable seals in the transfer canal flexible-joint seismic area. A valve which regulates air supply to these seals was found shut. Georgia Power is still investigating why the valve tas shut and why the leak detection annunciator failed to alam. Coolant sample analysis by Georgia Power indicates that 1.26 times the maximum permissible concentrations of the following isotopes were released: cesium-134, cesium-137, zinc-65 and manganese-54. The AIT is composed of a section chief from the Division of Reactor Projects, resident inspectors, and a specialist in both radiological effluents and chemistry, and environmental effects. Media interest has. occurred. Georgia Power has issued a press release, and Region II is responding to inquiries. The State of Georgia has been informed and has dispatched a person to take environmental samples.

Georgia Power informed the NRC incident response center of this occurrence by telephone at 1:35 a.m. to(ay. This information is current as of 2:30 p.m.

Contact:

_ R. Croteau, 242-4668 V. L. Brownlee, 242-5563 DISTR 1BUTION: H. Street MNBB Phillips LEW W111ste MAIL: ChainT4n Zech EiD'~ hRR IE hM55 XURIUMB Com. Roberts PA OIA RES DOT: Trans Only Com. Asselstine MPA AEOD Applicable State Cem. Bernthal ELD Com. Carr Air Rights INPO SECY SP NSAC ACRS CA PDR Regions: Licensee: (ReactorLicensees) Applicable Resident Site 9 e 4 /

L- - YE RYAI 9, RT T N EBNT 40 TASN N ONi S T AS 6 STSE O O TA ADCE 2~ ISEC I DMt N EOHGWE VR D M RR T EME G O RN N SRNOTC E PE A RON.NS NA I ENRIDUEITA H P L RCNM.IC OE I T WAEEEAF AH.LCEA O U A WG C CTBTCSSIT7A T 0 SBCEHATN ONAE R A A NIC 3NDAG0 I CHC AE LIET ESWENNA ND1ONGN1 EOT 6LR

PL TN RIEREUE IAII TS D1OR 10U'ADSAORAMETENTSG TKT NANTE

SU FA NWLO AR SNACESSAA EHOCR2IC CNM0 LFST ENAM.RTEM MPIEI1 C N

AA*6A NNEHE I NV S N TLU TO WS El00DOOVTCNT4GENSI IEAFoTN OAOLA EICA IOS3NUII AHLEEAEH DWT G)0TT HFLIE1IL 1 TTORR MT T NE L0 CO O T DFOE N S .N O UTIH0O0MANO ECSSUFTET OFITYYIB HA T0O TSHEICLE SI C0 OtLA SHL 0,P0CDS ETT C ENN ENNTTG TA 1CNEST ETDNLTEU' B1H ONNN D N 0L AAVRA EAG.DIEIAIC TD EOI LKAS5 EON S CSI EL IEUCT OACG UTUTIE TKDA]I L SE.D R Q A A CE NYF NLPEUANETGE D. AN 0EAG LRIL 0EAEPHHEALIOHEE HN)ErS I N EDET0HLBATSL EALTHUP PAEREBET IAFLTN0,TP DR(O TSU HRURUSS SIAE EKRDEE IO SW ACUDtSAE 2DNUMP5SEEAONHZEMDT .I O SEH HV EABIS EHSEFUTYRAA L DNSCT .P N TRR X HVTNL O LEERTDGL EOEO I IETROECE E SFYAWT NENO V RRoSA NV ORHIIOCEL HN DEIIF ITPPNL Y UOETPTHLTIHASWASNNSFE ES AE ,L CHCP WE LTEA EOA IBR CETE NNT HSTAAN BT SWDSPI NT O ETNC.NON TI E IF NE NAOTSEOST R ENTAIE IDMRD OEEH.ETAWTCRENIC EMACHTSP 9 W O NT ECTRLA OEOB E 2CNLtCAE'J ,SR2AELSA EBHTTSPT NNP NILN RRW T RF EONJ VATUNISCSEPS TAAINHOPO N )O TMFOED.IT HA NEAE N ILTEOTT L I E TTRIR PCARRAESLGIPHBAI NLNVCOSSL I I V SAENA3 I E LV ON S UIOR BEIO I E ERTUL LLATAFLSOIDNSEDT N6 (EA AOE EAHNAACWEIRVNN , ECUS R F V SSF/E N6 O8 R PWD TUERWAIVW OT OAAE .R I OOYEO O8 I9 O NO FHA M ELNTTH D M. O,LAT2 T EAEYEHLENC MI M.S HCUEN UHER RST I9 G1 G1 E M O /T YHTHLVTOMEESES A IT EVES E . E PEO1D1N.PTLTPL FGDPLTE AH NIOCR R, R M A PA UISAIR 1 (EcLCIC 4= 4 T GP E E I 2A TL1PRAN OUV.EASNI 1G SI ELT - 2 R 0TLII SOWIETS MH EICSD :NCUTNR C R-TE F

UENAY FSAH ERTNR IIN 2IIAAI EE iE RB O 0OUUFLE TTRRHA A LFHA 1MGCI MHP nB OM 1

F EHDAN IITLF' dFT ROETDETS oM PE N G ERTTE ODAA AO NTO T TOLBIET N PE EC O TNTHOA TOCE Y JON NI AF NKS .I EC RE I AINTTMMNT RFNHOCSIELOA RN2 ICY RE D T LE AIOU OEOAWTPSSMA EO OSDT D G P 3EPNIXRODTT H INCE 4PILLL EN G N I /USECCFCE NSOGD NVOOS / TEE GRE N I: R 2F ENR CNNESTNESZRRLAR 2EANNENRD I: NE C IESWUPTAIE O ILE EI EE ILLNNBIUI NE RT S RSTNPS AMSLETIMDPVDLW IOAA RCS RT OA E N OENAOERAA UAAINUNNEO NHSHHOUCE OA MD D OALBA(LBDTNADGFTASEARP DWICCTDOR MD ~ L A 4 C I 3 4 T G 9 / C O 0 2 E L 7 1 J O 0 B I Y U RD D R A S EAEN E D / CRSA C O N I A I T O FTE1 F2 I FNL9 F9 S T OEE0 O0 S T A TR7 7 T N C YR 0 Y0 N E I TED T E M t'VI U M E I I ODU: D: E G T AQN N G A O QNIE OE A G N HILR HR O N N E E G G N N I I M Y M A T A E I E P L P S I 2 S C 73 O A 6 I2 F N F 6 133 O / 3 E E H00 R R E 20 A55 A A S 5 Y D N H O N E E C U: E R C T: QS L E I AN EN L HD A H SD C T = l l lll l' Ill' llllll

h Gw;a % e Company jrJy/ 333 P.eomvat AveNe At4-ta Geo'; a 30308 f t / ',. 'erorote 404 $26 6526 tof e o 4045 L At anta. Georg's 3C302 Georgia Power - e +~ m - c -- i t o cwe SL-1692 .T$$*;'* ***'# 0963C E December 5, 1986

REFERENCE:

U. 5. Nuclear Regulatory Comission RII: JNG Office of Inspection and Enforcement 50-321/50-366 Region II - Suite 2900 101 Marietta Street, NW Atlanta, Georgia 30323 ATTENTION: Dr. J. Nelson Grace Gentlemen: This letter serves to document those items which we presented to members of your staff in telephone conversation today. The following items describe certain of our planned actions related to the loss of water from the Plant Hatch Unit 1 and 2 spent fuel pools which occurred on December 3,1986. The fuel pool transfer canal bellows seals arrangement is being evaluated to assure that design changes are generated to make the transfer canal assenbly single failure proof. The transfer canal sealing configuration is currently such that no single-failure will result in the loss of inventory from the spent fuel pools and will be maintr.ined in that mode. The secondary containment is designed to minimize any ground level radioactive materials which might result from a serious release of To assure this objective, Technical Specifications provide for accident. periodic operation of the standby ventilation system to assure its ability to draw and maintain an airflow into the building. The specific performance objectives of the Technical Specifications could be affected by changes in the status of elements of the secondary containment boundary. ' As is our practice, Georgia Power Company will continue to monitor these secondary containment elements, including the refueling floor boundary, in order to assure that we meet the performance requirements of the Technical Specifications as plant conditions change. The plant environmental monitoring program has been augmented to I further assure detection of sanmade isotope concentrations and to identify any significant migration. This

program, will be fully implemented by December 15, 1986, and will be available for NRC review upon site inspection.

The program specifically will include sampling of 'l the run-off from the yard drainage system. Monitoring data and analysis r gh S will be maintained at the plant for review by your staff. g1 a

g k i GeorgiaPower h U. d. nuclear Regulatory Comission Office of Inspection and Enforcement Region II - Suite 2900 December 5,1986 Page Two If you have any questions in this regard, please contact this office at any time. Sincerely, pg, c> L. T. Gucwa RDB/lc U. S. Nuclear Regulatory Comission c: Georgia Power Company Mr. i. P. O'Reilly Mr. D. Muller, Project Director Mr. J. T. Beckham, Jr. BWR Project Directorate No. 2 Mr. H. C. Nix, Jr. Mr. P. Holmes-Ray, Senior Resident GO-NORMS Inspector - Hatch 1 0963C

e 5 E. W- ~. E Ca W:.*et WE mE,* e $2w s 5 1. See8 W E i*W f:52t

• .E S:8_,~h.i '2 5 ~8._E,*w v..

.s zvo-e w..w eE g:r t -m-R5ea se s errggs. grWha "giMI:" E* g*o*;&a e e e e-.f:ef:amo>I*w..s . ::=u.ssaa, .sst o -e <s <== wwwm-k .O -rw-vumm E E r~EU ..~kOmWW IN ~ - ~

• 2

" hem -****"kk $r ~SI.2 $U~fu em 83 fi* L,L"E.5~~ E

"~

E ~55g $:e8 Et5EE*ye55-~55$ ">ssR E 8 b...S..~ =.s .. esE> w w s.es so"* E Ee.:E55--wts"'b"~W.W ~"*5r g 8 3 Wsfo o ww s v e'g '.,8ws.s. 5500..$.gsgge-IWa*EN"Ey -Wr==- -.Wwm s as wE 2E*W h m .o w = .w. -e WW"*.T:,5*. git."t *5ewit$.rel-R* ezEm .o w E.E R$E".W.**d.%.5"f.e 2 s 1 553 xe. gys g55"" e

a - -e.-*=jbCgerESgg.

ww-g 8d c Ew S'gg[Wds -5:ase=BE ="E 22jf ES,.twEit=! sswo o --:Eim*Ew! - Ee oW** g oWW " g r esg~.r.,r.newsJ-:=.st"E585"W .a ss.s or

5 s EerWese rtp-E w s..Ea mew a.r:

-u,s .r.s. - .. ~5a e sw s* ,u %v s. -o ~ava >sg.=e!*Ie.Even,= --*-*er-W-* ,g g~s~ amE* s.! san.as.e!v5:=.e p: er*g: = -55_,. ag.: D e=am*We.- ~ .5mi . 55 88:5 " e

5. ~ es re! zWra ~El.gw,...;W gg u=".

r ,la. m --~ s--W"-Ee E"Ro-n-e = sE Eses *sgs g:ss. ek :eseeWW:gts:.. e .seacWe te e = se

5 st

.e.w, fr=ce.<amW ear z-b.: r: En eeer --5 v e Et,", E -w=w.8 "e'u.E I""eW t = W w.- .me-

• t *E***-*#dW :s*"
  • .gw"u wo g
    • * ~=Ew=5.~5 gat w=gh.

= s. w .I 2:*.stM g = e E~EW "EweE':*- --Ems.- E ~""E E-sE b S*~ 8 iErstu"E*.138.I""E"*W>"r"*"E.G" ag a r,wr u .1 W= sew =w.E."ww g% .s .u tw IEsEo:g8s.::.$ewvero5"GEg:8 -ww -s wo -r w

• a c

.g=g. w W" h.b Wgeerm ws~EE5ggw=.ru f* --e~ ww-E 5e E h

• a 5 OwEE "*5~'E Enns fesa=>=vEgg: ass,~t.v rs.e:W=

5. eEe W:-

.fe t* E-55825 ma.e ~gEr" e Ws_ e -sa .W E 3: ag=s-. s:ansgxzzes:8~raeae. y. =w wus- . re 8 s.::. w .E = 3. w ,== 0 5 w k wL 8 WS =g: .e,, 2 E* = 2 g== Eg .W Gog gi R 0* tw. I -= E CI Es-EL = ES B" "d: -= EE -w-u.. E.. m -M W an. =g GE O

• 5 5

$NN E %" u B.E.: E E WW En Wie ~ w g O Eg ~ c I., b. =! E .N ":.g,I a ~w. O N. =.v ? ~;; e a:: / EO 2* 5 g 0 #G i.. KG = E pg 3 EE 15 EE I

at w ac O W Z O w M e w at w .J U CC w as > 2 o > M W 4 w Up4w 64 st O ac 2 w 2.J > 2 3.eU O O D 6-e-2 24 2 a 3.J 4 W w a= N .J 2234 WW@C OMZ

• • O ac.J 24 wCC wk M 303 ac w

+ = > E - - M ec:a -ac

• * = M w>ww24w wO2g2 4 w C D *- as 4 J Z Z w ec C e., at U = M L4 scZw=

xQ>=4>O 4 cc w D v 4.J O M24 W> ac > 2 >cc C Z 4 t= ac

10.

2 e= Z 4 U C *- MO w w=*- 00 I sc CDo 4 < > w as w== 2 Z 4 2 20r>4% 4 Z M M e O u w aa > w2 C w== w w Q > ac 2 > ** M 4 "3 2 2 D 4 er.J ac x I aw w w 4 N.e>( w 2 2 4 > 2 e= c 4 a= k e= e= U.J M *=e e

• - M ac E C 8= 4 w Mwwu2 *= E

'= = = E3*2*hD>w - Om C.J wow Cruw >DIC warwMM2w C2C a4I> w== w w

• dL e= at U 3 MOUOM*3>4 EM>

ZZZEED .J = w ar C WN $= p= C w e= M W 4MMCLwwMEE Okt-N - M ( am 2 M4ENZ-CW 6-O e( M

• Owww Q D Z D a* Q Z ac >

3 2 E.= (D 2 % at w w u

a..J e-w M Z

M EM eh w2 OMM 4 w M M e= > M>J > M ae W M W C at -

c. >

4M4 >4wscWWE .JOC2wE E 43wo M.J O Q as w w O .J a*a2wC44 2 wwokshWw 4OMw-*-w3w 0wWD >= ac Z (**> >2 C Z M M M ac wwWM 3 M M.J U M a= > w>M 4 UMM .J O -C > w EOOMOMWW 2222 (WOwdOWA

• a 2 2 M U Z.J wwo 2 >.J ww2x w

Z W D e= 4 UU C 4 w C w as e-w W 4 4 at O p sc U

• = em e= C.

M.J ea C ac w at ww3ML2O .J.J ac >= w > w M W&ac>C4 M.J U 4 4 at w2 M A34423C www E OMMM p 20W MZZMs W>EwwCE> p M ac e. 3 2

• - O *=.J 2 ac e- > w.J

e. >- to - Z 2 w ac 2 W4IM Cwwo O

ac o OW <>w-C

• a w

>>4m2OcC) b ww D ec MA > WM2EC W2 4 - O ac 4 2 kE>w w C M > *= > M.J a o acMM2w 4 M 4 w o k e es 2@ N

1. = 2 *a wA242 at **

Cm K w 3 44 C2 h10 M Estwe3W2 C C o= Z .J.J ac 4 w CDMck Q4w3M Ow

• = En.

c. 2 *a - c D L w ac w e*oD w ac k c ac e. W E 42 20 Z M w dwo w e= > > W D e= ec

• w U D E *= ac w *= O Z at' U =s t E>Cw4>&M en s *=

C D = 2 O 2 ac H 2 *a Owk214 er e ev M at 2 ac 2 0 > 2 ( D 4wk MM e= > *a 4 3 9= UU at e enJO w w w 4 O o p e* 4 6-w k w k ac o czw o U C (LinwOood acto O e c.J 4 C >>p2 .J 2 > M2www ec o M w m.J DE 2 Jt a= ac M M *= s 2 > < ODEM C6 w 2 4cDUC304WwM ( M ** w t w w (

• 4 >=

wu O ac % C w Dh2 > ac .J.J w ac ac u Z M *= ac w M M E.J O cc O C C 4 0.J 0-ac D D wss w C E O

• =

4wJwaswwE 46E

e..J >- to w er e. w e

n. w as== M w e es w M

• • Uwo
• a u t= 4 >

2

• = 30

>-MwM ac 2 N2 ** C U 0440W Ow ac > w as M W V W w ** ** N42 266wOZ2 Ew U Uw U z u tr..J 4 MJZ 2 M D.J v4M ac > M OCMOCCC .J E CU w ac w 2 *= M *= M OC w 2.e4 2 acacac e - w row w D2444Z+=1 EQ Q < a 3 4 4e-h > E t D u p u so QWEw23M> M> E 4 O .J og w U M e as W U

• "a D t= 4 to at at o D

E 6= 0 M w M c. W O 2 w or 2 O ac D O k sn U .J M 6 cn > e= >= Co .J u t 4 WC2w U >o wMw M >= M ** M i w D 220 e4 O ++ w w Mat 2 U>a m O Ow MQwa 2 E ac .J h. D E Ow h e= e > e=s 2 .J O M U U

== w nn ( en NNN 6 N en .e W n e e e

== o o w w== o in en M at en w 2 2 .4.. w U w U > CM

• • 4 2

or .a K MO >O

OPERATING REACTOR EVENTS BRIEFING 86113 DECEMBER 8, 1986 SIGNIFICANT EVENTS 0 MULTIPLANT ENVIRONMENTAL QUALIFICATION OF AMP SPLICES 1 HATCH 1 AND 2 AIT FOR LEAK.FROM SPENT FUEL POOL HADDAM NECK ELEVATED REACTOR COOLANT SYSTEM (RCS) LEAKAGE CALVERT CLIFFS 1 GAS LEAKAGE TO JACKET WATER COOLING SYSTEM FOR EDG OTHER EVENT OF INTEREST MILLSTONE 2 RESULTS OF EDDY CURRENT TESTING (ECT) 0F STEAM GENERATOR TUBES IN NOVEMBER 1986 - UPDATE t

DECEMBER 8, 1906 ] HATCH I AND 2 - AIT FOR LEAK FROM SPENT FUEL POOL DECEMBER 4, 1986, (E. WEISS, IE) PROBLEM: 141,000 GALLONS LEAKED FROM SPENT FUEL POOL TRANSFER CANAL CAUSES: AIR SUPPLY INADVERTENTLY SHUT TO INFLATABLE SEALS IN CANAL DRAINS ON LEAK DETECTOR INADVERTENTLY LEFT OPEN SIGNIFICANCE: LEAK NOT IDENTIFIED FOR HOURS (AIR SUPPLY SHUT 2200 CST DEC 2) LEAKAGE PATH TO ENVIRONMENT AND CAUSE NOT IMMEDIATELY APPARENT D IF FUEL BUNDLE HAD BEEN IN TRANSIT, POTENTIAL FOR UNC0VERY EXISTS DISCUSSION: UNIT 1 AT 100% POWER THROUGHOUT EVENT; UNIT 2 SHUTDOWN TRANSFER CANAL SEAL IS INFLATED DURING REFUELING INFLATABLE SEAL IS USED IN GAP BETWEEN REACTOR BUILDINGS FOR SEISMIC CONSIDERATIONS DOUBLE INFLATABLE SEAL ON GATE IS IN PLACE DURING REACTOR OPERATION SEAL LEAK DETECTION DRAIN VALVES (F238 AND F239) LEFT OPEN PRIOR TO EVENT - SEAL LEAK DETECTION ALARM DID NOT WORK PROBLEM WITH PRESSURE REGULATOR; AIR VALVE THROTTLED AIRVALVEMOVEDTOCLOSEDPOSITIONWHILERESTORINGFR0!dCLEARANCES 1430 CST DEC 3, THIRD LOW LEVEL ALARM IN FUEL POOL-2200 CST DEC 3, COULD NOT OPEN DOOR TO NITROGEN ROOM, WATER POURING INTO CABLE TRAYS, LEAK FOUND LEVEL DOWN ABOUT 5 FEET ALARM ON SPENT FUEL POOL LEVEL WORKED 17,000 GAL TO RAD WASTE 40,000 GAL CONTAINED BETWEEN REACTOR BUILDINGS 84,000 GAL TO STORM DRAIN AND SWAMP 1.26 X MPC CS-134, CS-137, ZN-65, C0-60, MN-54 DIKES BUILT FROM OUTFALL TO SWAMP RIVER RISING BECAUSE OF RECENT RAINFALL TANKER TRUCKS USED TO REMOVE WATER CLEANUP OF WATER BY RECIRCING THRU DEMINS TO TANK TRUCK

HATCH - CONTINUED 2 FOLLOWUP: HATCH AIT SCOPE IS TO DETERMINE ROOT CAUSES OF EVENT AND FAILURE OF ALARM DETERMINE AMOUNT OF SPILL AND RELEASE LICENSEE RESPONSE TO IEB 84-03 AND IN 84-93 POTENTIAL RISK IN DRAINING BOTH POOLS POTENTIAL FOR LOSS OF SECONDARY CONTAINMENT LICENSEE'S KNOWLEDGE OF PRECURSOR EVENTS PRELIMINARY OBSERVATIONS BY AIT RELATE TO: ADMIN CONTROLS FOR PRESSURE REGULATOR AND LACK OF DEFICIENCY REPORT e CLOSING OF AIR SUPPLY VALVE WITHOUT PROCEDURE PROCEDURE TO CAllBRATE LEAK DETECTION PROCEDURE TO CHECK AIR PRESSURE DID NOT INCLUDE TRANSFER CANAL SEALS ' AIR SUPPLY TO SEALS NOT REDUNDANT IE CONSIDERING INFORMATION NOTICE IF CONFIGURATION IS NOT UNIQUE b I 1

f .3 k y r H d a f ~ ab { /e gg 5/.y 1 e> t ~ >g %"e9 e i~<\\ ( lPv; y 5,:a 3 =g e d~ L*rti l* 3 ee , 4

,t-3* e i,

a y 1 \\ 8 ~ ~ s \\ ,g $3 0 @wkis Q 1 Lt t e g = , s. 3M S 9 .... _ - - - - - - - - ~ - - ~ '

(ML9(J A V + REOULATORV IfGORMAT20'd DISTR 2DUTION SYSTEM (R2DS) ACCESE. ION NDR: 8704160;'68 DOC. DATt.: 67/04/07 NOT AR 17 E D: YES DOCKET = FAC IL: 50- 31'J Ar k arisa s Nu c l ear One, Unit 1, Arkansas f>0vt1

t. Light 0 500C3 3.

A U T H. N /.f*0 AUlHOR AFFILIATION CAMPBEt t,7. O. Arkansas Powrr & Light Co. REC IP. NAP 1E RECIPIENT AFFILIATION STOL2,J.l-f'WR Pr o jec t Direc t cra t e 6 SU11JEC1: Application for amend to Licer.se DPR-51, reviving Tech Specs to delete Section 3.8.15 re availiary b1dg crane & associated basis. Fee paid. DISTRIBUTION CODE: A033D COPIES RECEIVED: LTR,_, ENCL _ SI ZE: _3 + 7 TITLE: OR Submittal: USI A-36 Control of Heavy Load Neter Spent Fucl-NUREG-Oe NOTES:AEOD/Ornstein:1cy. 05000313 ) RECIPIENT COPIES RECIPIENT COPIES ID CODE /NAME L1TR ENCL I D CODE / NAM.S. LTTR ENCL tD J 1 PD4 LA 1 O NRR SINGH,A 4 4 PD4 PD 5 5 P I:1 CK, C 3 1 . NTER NAL: ARM /AkF/LFMD 1 O NRR/ DEST /PSD 1 1 NRR /DE ST /RSl4 3 1 f.f 4 R /PMAS / P MS14 1 1 REC FILE 01 1 1 m.c d r.**. 2 1 1

XTERNAL: LPDR J

1 fRRC PDR 1 1 NSIC 1 1 NOTES: T' 2 \\ \\ w ciecf. I 6d L g

1. 04-1130 i

un of A-56 /CL - ST-It ^* I "A T Ua J ca.f(/ '[' ' ' ) i V. 4 ~ v A TOTAL NUMBER OF COPIES REGUIRED. LTTR 23 ENCL I'1

1 4 ) S, ARKANSAS POWER & LIGHT COMPANY CAPIT0L TOWER BUILDING /P. O. BM 551/LITTLE ROCK. AR",ANSAS 72203/(501) 377 2525 1 i T GENE CAMPBELL vee prescent Wclese Operatens 'ICAN048701 U. 5. Nuclear Requ11 tory Commission Document Control Desx Washington, DC 20555 ATTN: Mr. J. F. Stolz, Director PWR Project Directorate No. 6 Division of PWR Licensing - B SUEk'ECT: Arkansas Nuclear One - Unit I Docket No. 50-313 r. o License No. OPR-51 Technical Specification Change Request Regarding the Auxiliary Building Crane " ear Mr. Stolz: i, As you are awere, AP&L is actively n rticipating in the Department of Energy Si (DOE) Extended Burnup Program. Part of this program, in association with the ' ANO-1 fuel vendor, Babcock and Wilcox (B&W), involves the shipment of a maximum of,16 spent fuel pins to th;(B&W Lynchburg Research Center (LRC) for hot cell examination. The DOE has ma e only one licensed spent fuel shipping cask 3'availabic for a very limhed time ( April 15 ,'une 15); therefore, oue to this h significant schedule constre,ipt we Wquest NRC reoperation in prompt review and aiproval of this amendment request. ANO-1 Technical Specification 175) 3.8.15 presently prohibits hanoling of a spent fuel shipping cuk with the Auxiliary Building crane. This constraint was made by ANO-1 T5 Amendment 17, dated December 17, 1976 (ICNA127641), pending evaluation of the spent fue\\ caskWop accicent and the crane dtsign by AP&L and NRC review and approval. At@jl TS Amendment 36, dated October 5, 1978 (ICNA107805), allowed a one Wae exemption during Cycle 3 to allow the crane to handle a cask for removal and sMpment of some irradiated burnabic poison rod y., test assemblies. The NRC evaluation of this issue was beorporated into the review of the generic issue (A-36) related to control of hepvy loads near spent fuel. On December 22, 1980, the NRC issued NUREG-0612, " Control of Heavy Loads at Nuclear Power Plants"; the transmittal also contained interfm actions for implementation and a request for additional information regardjng control of heavy loads. AP&L completed the necessary actions and submittais, and received a Safety Evaluation Report (SER) on October 11, 1984 (OCNA188406) which concluded l that Phase I of the iss.Je for ANO-1 & 2 was acceptable. Completion of Phase 11 was transmitted by Generic Letter 85-11, dated June 28,1985 (0CNA0c620), which f edicated that application for license amendment may be submitted to delete 'related license condidons citing GL 85-11 as the basis..This amendment request ? is therefore submitted to delete TS 3.8.15 and its as ociated si. 3h g / M TP M N $das slo I 'Y c, c o

l. April 7, 19E7 i

AP&L's procedures, load paths, crane equipment ; certification and operator training and other related heavy load handling topics were evaluated as part of the control of heavy loads issue and found acceptable. Spent fuel cask handling is discussed in Section 9.6.2.6 of the ANO-1 SAR. Specialized equipment is being transferred to ANO by B&W from the Oconee Nuclear Station for removal of the selected individual fuel pins from the extended burnup test fuel assemblies. This equipment will be set up in the cask pit, the selected fuel pins will ba removed and then stored in a special basket in the spent. fuel pool. The equipment will be disassembled and removed from the cask pit, then the shipping cask will be moved into the cask pit and loaded with the basket containing the test fuel pins for shipment. Preplanning of the fuel pin transfer and related tasks is already well underway, with appropriate considerations based on AP&L's commitment to ALARA principles. Transportation arrangements, routing, etc. for the shipment are being coordinated by the B&W EC. The DOE schedule limitation on the cask has required a tentative sh Nment date of June 15, 1987. re The circumstances of this proposed T5 amendment request are not of an exigent or emergency nature; however, expeditious handling is requested due to the limited availability of the DOE cask, as discussed above. In accordance with 10CFR50.91(a)(1), AP&L has evaluated the proposed change using the criteria in 10CFR50.92(c) and has determined that said change involves no significant hazards considerations. Also, in accordance with 10CFR50.91(b)(1) a copy of this aaendment request with attachments has been sent to Ms. Greta Dieus, Director, Division of Radiation Control and Emergency Management, Arkansas Department of Health. A check in the amount of $150.00 is included herewith as an application fee in accordance with 10CFR170.12(c). Very truly yours, dWM T. Gene Campbel Vice President, Nuclear Operations TGC:RBT Attachments cc: Ms. Greta Dicus, Director Division of Radiation Control and Emergency Management Arkansas Department of Health 4815 West Markham Street Little Rock, AR 72201

STATE OF' ARKANSAS. ) ) 55 COUNTY OF PULASKI ') I,_T. Gene Campbell, being duly sworn, subscribe to and say that I am Vice President, Nuclear Operations for Arkansas Power & Light Company; that I have full authority to execute this oath; that I have read the document numbered ICAN048701 and know the contents thereof; and that to the best of i my knowledge, information and belief the statements in it are true. f AYM T. Gene Campbell SUBSCRIBED AND SWORN 10 before me, a Notary Public in and for the County and State above named, this / h day of ' 1987. ) Lk. mu O \\ / Notary Publi U My Commission Expires: fflo 1. /wL

3.8.15 Deleted 3.8.16 Storage in the spert fuel pool shall be restricted to fuel assemblies having initial enrichment less than or equal to 4.1 w/o U-235. The l provisions of Specifications 3.0.3 and 3.0.4 are not applicable. 3.8.17 Storage in Region 2 (as shown on Figure 3.8.1) of :.he spent fuel pool shall be further restricted by burnup and enrichmeat limits specified in Figure 3.8.2. In the event a checkerboard stor ge configuration is deemed necessary for a portion of Region 2, vaca.t spaces adjacent to the faces of any fuel assembly which does not meet the Region 2 burnup criteria (Non-Restricted) shall be physically blocked before any such fuel assembly may be placed in Region 2. This will prevent inadvertent fuel assembly insertion into two adjacent storage locations. The provisions of Specifications 3.0.3 and 3.0.4 are not applicable. 3.8.18 The boron concentration in the spent fuel pool shall be maintained (at all times) at greater than 1600 parts per million. f BASES Detailed written procedures will be available for use by refueling personnel. These procedures, the above specifications, and the design of the fuel handling equipment as described in Section 9.6 of the FSAR incorporating built-in inter-locks and safety features, provide assurance that no incident could occur during the refueling operations that would result in a hazard to public health and safety. If no change is being made in core geometry, one flux monitor is sufficient. This permits maintenance on the instrumentation. Continuous ronitoring of radiation levels and aeutron 'iux provides immediate indication of an unsafe condition. The requirement that at least one decay heat removal loop be in operation ensures that (1) sufficient cooling capacity is available to remove decay heat and maintain the water in the reactor pressure vessel at the refueling temperature (normally 140*F), and (2) sufficient coolant circulation is main-tainedthroughthereactorcoretominimizyheeffectsofaborondilution incident and prevent boron stratification The requirement to have two decay heat removal loops operable when there is less than 23 feet of water above the core, ensures that a single failure of the operating decay heat removal loop will not result in a complete loss of decay heat removal capability. With the reactor vessel head removed and 23 feet of water above the core, a large heat sink is available for core cooling, thus in the event of a failure of the operating decay heat removal loop, adequate time is provided to initiate emergency procedures to cool the core. The shutdown margin indicated in Specification 3.8.4 will keep core subtritical, even with all control rods withdrawn from the core Although the refueling boron concentration is sufficient to maintain the 5 0.99 if all the control rods were removed from the core,. only a core k fewcobolrodswillberemovedatanyonetimeduringfuelshufflingand 14 AmendmentNo.I[0 , 56, 57, 76 59a 470460288 B70407 PDR ADOCK 05000313 A

i repl acerrent. The k with all rods in the core and with refueling boron concentrationisaphximately0.9. Specification 3.8.5 allows the control I room operator to inform the reactor building personnel of any impending unsafe condition detected from the main control board indicators during fuel movement. The specification requiring' testing reactor building purge termination is to verify that these components will function as required should a fuel handling accident occur which resulted in the release of significant fission products. Because of physical dimensions of the fuel bridges, it is physically { 1mpossible for fuel assemblies to be within 10 feet of each other while being handled. l i Specification 3.8.11 is required as:

1) the safety analysis for the fuel handlingaccidentwasggjedontheassumptionthatthereactorhadbeen shutdown for 72 hours.

and, 2) to assure that the maximum design heet load of the spent fuel pool cooling system will not be exceeded during a f

full core offload. Specification 3.8.14 will. assure that damage to fuel in the spent fuel pool will not be caused by dropping heavy objects onto the fuel. Specifications 3.8.16 and 3.8.17 assure fuel enrichment and fuel burnup limits assumed in the spent fuel safety analyses will not be exceeded. Specification 3.8.18 assures the boron concentration in the spent fuel pool will remain within the limits of-the spent fuel pool accioent and criticality analyses. REFERENCES (1) FSAR, Section 9.5 (2) FSAR, Section 14.2.2.3 (3) FSAR, Section 14.2.2.3.3 Amendment No., I7, 56, 57, 76 59b

DESCRIPTION OF AMENDMENT REQUEST l The proposed amendment request would revise ANO-1 Technical Specification (TS) 3.8.15 and the related Basis to allow the Auxiliary Building crane to handle a spent fuel shipping cask. TS 3.8.15 presently states that the spent fuel shipping cask shall not be carried by the Auxiliary Building crane pending the evaluation of the spent fuel cask drop accident and the crane design by AP&L and NRC review and approval. The related Basis states that upon satisfactory completion of the NRC's review, Specification 3.8.15 shall be deleted. TS 3.8.15 assures that the spent fuel cask drop accident cannot occur prior to completion of the NRC staff's review of this potential accident and the completion of any modifications that may be necessary to preclude the accident or mitigate the consequences. NRC review of this particular issue was incorporated into the staff's resolution of the generic issue (A-36) related to control of heavy loads near spent fuel. AP&L has completed all actions and submittals required by the issuance of NUREG-0612. " Control of Heavy Loads at Nuclear Power Plants", as evidenced by NRC Safety Evaluation Report dated I October 11, 1984 and Generic Letter 85-11 dated June 28, 1985. Generic Letter 85-11 indicated that application for license amendment may be submitted to delete related license conditions citing GL 85-11 as the basis. This amendment request is therefore submitted to delete TS 3.8.15 and its associated Basis. BASIS FOR PROPOSED NO SIGNIFICANT HAZARDS CONSIDERATION DETERMINATION The NRC has provided standards (10CFR50.92(c)) for determining whether a significant hazards consideration exists. AP&L has evaluated the proposed TS amendment and, on the basis of these three standards as discussed below, determined that the proposed amendment does not represent a significant hazards consideration. A proposed amendment to an operating license for a facility involves no significant hazards consideration if operation of the facility in accordance with the proposed amendment would not: (1) Involve a significant increase in the probability or consequences of an accident previously evaluated; AP&L's procedures, load paths, crane aquipment certification and operator training and other related heavy load handling topics were evaluated as part of the control of heavy loads issue and found acceptable. Spent fuel cask handling is discussed in Section 9.6.2.6 of the ANO-1 SAR, which shows that the cask will never travel over spent fuel. Although cask handling is presently prohibited by TS 3.8.15, ANO-1 SAR Section 9.6.2.6 further evaluates the incredible event of a cask drop accident and shows that the consequences are acceptable. Deletion of TS 3.8.15 and the related Basis to all.pw handling of a spent fuel shipping cask by the Auxiliary Building crane will have no actual impact on the cask drop or any other previously analyzed accident. This amendment request therefore does not involve a significant increase in the probability or consequences of an accident previously evaluated.

(2) create the possibility of a new or different kind of accident from any } accident previously evaluated; The proposed amendment request will allow handling of a spent fuel shipping cask by the Auxiliary Building crane where previously prohibited cue to pending NRC review and approval of the control of heavy loads issue. The cask handling methods and cask drop accident are discussed and evaluated in AND-1 SAR Section 9.6.2.6. The evaluation of the incredible event of a cask drop accident has included all possible equipment f ailures and shown the consequences to be within acceptable bounds. No new accident scenarios can be identified related to the proposed amendment request, therefore this change is bounded by the current analysis in the SAR. The proposed amendment request will therefore not create the possibility of a new or different kind of accident from any accident previously evaluated. (3) involve a significant reduction in a margin of safety. Although allowing use of the Auxiliary Building crane where previously prohibited could increase the possibility of a cask drop accident, the margin of safety is preserved in that the acceptable consequences of the cask drop accident evaluation in SAR Section 9.6.2.6 are not affected by this change. The proposed amendment request will not adversely affect the adequacy and conservatism of the cask drop accident evaluation. The spent fuel cask has been issued and continues to hold an NRC Certificate of Compliance for radioactive materials packages, and the procedures, load paths and equipment to be used for cask handling have been reviewed and approved by the NRC with the resolution of the control of heavy loads issue. Therefore, the cask handling issue at AND-1 cortinues to exibit an acceptable margin of sufety and this amendment request will not involve a significant reduction in a margin of safety. The NRC has also provided certain examples (51FR7750) of T5 amendment requests which are likely to involve no significant hazards considerations. This T5 change request most closely matches example (iv): A relief granted upon demonstration of acceptable operation from an operating restriction that was imposed because acceptable operation was not yet demonstrated. In this case, demonstration of acceptable operation applies to the acceptable resolution of the NRC review of the control of heavy loads issue for AND-1. It should also be noted that a one-time exemption to existing TS 3.8.15 was allowed by the NRC to allow cask handling shipment of some test burnable poison assemblies during Cycle 3 operation at ANO-1 (T5 Amendment 36, dated October 5, 1978). Based on the above discussion and evaluation, AP&L has concluded that the proposed T5 amendment request meets the standards for determining that no significant hazards consideration is involved and therefore concluded that this amendment application involves no significant hazards considerations. __}}