ML20238F365

Containment Integrity Research Program Plan

ML20238F365
Person / Time
Issue date:	08/31/1987
From:	NRC OFFICE OF NUCLEAR REGULATORY RESEARCH (RES)
To:
References
NUREG-1264, NUDOCS 8709160104
Download: ML20238F365 (38)

Text

-

NUR EG-1264 Containnlent Integrity Research Program Plan U.S. Nuclear Regulatory Commission Office of Nuclear Regulatory Research y o o ,,,

l ci.x,;

e,,,,,-

8709160104 870831 PDR NUREG PDR 1264 R

NOTICE Availability of Reference Materiais Cited in NRC Publications Most documents cited in NRC publications will be available from one of the following sources:

1. The NRC Public Document Room,1717 H Street, N.W.;

Washington, DC 20555

2. The Superintendent of Documents, U.S. Government Printing Office, Post Office Box 37082, Washington, DC 20013-7082

3. The National Technical information Service, Springfield, VA 22161 Although the listing that follows represents the majority of documents cited in NRC publications, it is not intended to be exhaustive.

Referenced documents available for inspection and copying for a fee from the NRC Public Docu-ment Room include NRC correspondence and internal NRC memoranda; NRC Office of Inspection and Enforcement bulletins, circulan, information notices, inspection and investigation notices; Licensee Event Reports; vendor reports and correspondence; Commission papers; and applicant and licensee documents and correspondence.

The following documents in the NUREG series are available for purchase from the GPO Sales Program: formal NRC sta" and contractor reports, NRC sponsored conference proceedings, and NRC booklets and brochures. Also available are Regulatory Guides, NRC regulations in the Code of l Federal Regulations, and Nuclear Regulatory Commission Issuances.

Documents available from the National Technical Information Service include NUREG series reports and technical reports prepared by other federal agencies and reports prepared by the Atomic Energy Commission, forerunner agency to the Nuclear Regulatory Commission.

Documents available from public and special technical libraries include all open literature items, such as books, journal and periodical articles, and transactions. Federal Register notices, federal and state legislation, and congressional reports can usually be obtained from these libraries.

Documents such as theses, dissertations, foreign reports and translations, and non NRC conference proceedings are available for purchase from the organization sponsoring the publication cited.

Single copies of NRC draf t reports are available free, to the extent of supply, upon written request to the Division of Information Support Services, Distribution Section, U.S. Nuclear Regulatory Commission, Washington, DC 20555.

Copies of industry code and standards used in a substantive manner in the NRC regulatory process are maintained at the NRC Library, 7920 Norfolk Avenue, Bethesda, Maryland, and are available there for reference use by the public. Codes and standards are usually copyrighted and may be purchased from the originating organization or, if they are American National Standards, from the American Natioru Standards Institute,1430 Broadway, New York, NY 10018.

NUREG-1264 Containment Integrity Research Program Plan Manuscript Completed: July 1987 Date Published: August 1987 Division of Engineering Office of Nuclear Regulatory Research U.S. Nuclear Regulatory Commission Washington, DC 20555 r .ey, fo

c. /

ABSTRACT This report presents a plan for research on the question of containment per-formance in postulated severe accident scene los. It focuses on the research being performed by the Structural and Seismic Engineering Branch, Division of Engineering, Office of Nuclear Regulatory Research. Summaries of the plans for this work have previously been published in the " Nuclear Power Plant Severe Accident Research Plan" (NUREG-0900). This report provides an update to re-flect current status. This plan provides a sumary of results to date as well as an outline of planned activities and milestones to the contemplated com-pletion of the program in FY 1989.

iii

TABLE OF CONTENTS Page 1

l ABSTRACT..................................................... iii 1

l FOREW0RD..................................................... Vii

1. TECHNICAL ISSUES........................................ 1-1

2. REGULATORY NEEDS AND RESEARCH CBJECTIVES................ 2-1

3. SCOPF 0F RESEARCH....................................... 3-1 4 RESULTS TO DATE......................................... 4-1 4.1 Containment Model Tests............................ 4-1 4.2 Penetration and Vaterials Tests.................... 4-4 4.3 Electrical Penetration Assembly Tests.............. 4-6 4.4 Purge and Isolation Valve Tests.................... 4-8 4.5 Containment Buckling Tests......................... 4-9

5. RESEARCH PLANS AND SCHEDULES............................ 5-1 5.1 Containment Model Tests............................ 5-1 5.2 Penetration and Materials Tests.................... 5-5 5.3 Electrical Penetration Assembly Tests.............. 5-5 5.4 Purge and Isolation Valve Tests.................... 5-5 5.5 Seismic Effects on Containment Performance. . . . . . . . . 5-6 REFERENCES................................................... R-1 v

FOREWORD This rercrt presents a plan for research on the question of containment per-formance in postulated severe accident scenarios. It focuses on the research being performed by the Structural and Seismic Engineering Branch, Division of i

Engineering, Office of Muclear Regulatory Research. Summaries of the plans for this work have previously been published in the " Nuclear Power Plant Severe Accident Research Plan" (NUREG-0900). This report provides an update to re-flect current status. This plan provides a summary of results to date as well as an outline of planned activities and milestones to the contemplated com-pletion of the program in FY 1989.

Research efforts on containment integrity have had, from inception, the benefit of comments and suggestions from a peer review panel composed of recognized experts from industry and universities.

Publication of this plan will make more visible to inaustry and other inte-rested persons what our objectives are and how we are approaching the work in these important areas. It is intended that this plan will periodically be updated; therefore, comments on this plan are welcomed f rom all quarters.

Comments need not be restricted to activities for the period covereo but may include comments on omissions or what might be considered for the longer term.

Please address comments directly to:

James F. Costello Division of Engineering Office of Nuclear Regulatory Research Mail Stop NL-007 U.S. Nuclear Regulatory Commission Washington, DC 20555 vii

1. TECHNICAL ISSUES Risk analyses indicate that containment performance plays a dominant role in assessing of risk associated with severe accidents. A key insight emerging from the research on accident releases is that the niode and timing of con-tainment failure are very important in determining accident consequences.

Early failure without other mitigating factors can result in large radio-ac tivity releases, while delayeo failure of even several hours can signi-ficantly reauce the amount of radioactive material available for release.

Hence, the ultimate concern of the containment performance issue is how well the contailmient , conservatively designed for a postulated loss-of-coolant accident (LOCA), can withstand the pressure and temperature associated with severe core damage accidents. For scenarios in which containment integrity is maintained, consequences are small. In those scenarios leading to containment failure, consequence predictions depend on both timing and type of failure.

The manner by which it fails would influence the amount of airborne radioactive materials that could be released outside the containment. Knowledge of the time interval during which containment leaktight capability is ensured is important because, if the time interval between the release of radioactive material and containment failure is long, substantial fission product deposi-tion will occur within containment. Furthermore, the mode in which containment fails, i.e., gross failure versus leakage through foilure of penetrations, could influence the amount of radioactive material inside the containment that would be released outside the containment.

Two basic models can be used in severe accident risk estimation to characterize the loss of containment integrity: the " threshold" model and the " leakage-before-failure" model. Virtually all risk assessments performed to date have used the threshold model, which defines a threshold pressure, with some as-Lociated uncertainty range, at which the containment will fail with the po-tential atmosphere (which may contain a large amount of fission products).

If the containment pressure loading is calculated to be below the threshold pressure range, the containment is considered to be intact and the offsite consequences are therefore quite low.

Safety Study (Ref. 1). This model was used in the Reactor Some recent analyses (Ref. 2) have pointed out that gross containment failure is not the only pathway to substantial offsite consequences and that significant leakage on the order of 100 volume percent per day or more may result (if it occurs sufficiently early in the accident sequence) in substantial offsite releases of radioactive fission products.

The leakage-before-failure model provides a means of accounting for this ccndition when performing risk assessment analyses.

There are, however, significant uncertainties associated with predictions of leakage from containments. The technical basis for containment design was in-tended to ensure very low leakage under postulated LOCAs. No explicit con-sideration was given to performance under severe accidents. The analytical methods and test procedures used for design are aimed at ensuring that leaks are prevented, not predicting the onset or growth of leakage. Consequently, attempts to introduce a more realistic leakage model into predictions of 1-1

containment performance have essentially required the development and veri-fication of new techniques to predict severe accident loadings on and inelastic response of containment structures. Some elements of this new approach are

• further along in their development than others. The following sections indi-cate the current status of the development of tools necessary for confident prediction of containment performance under the pressure and temperature levels ,

associated with severe accidents.

Predictions of the pressure end temperature histories inside a containment during a severe accident are the subject of current state-of-the-art research. -

Sources of containment loading resulting from a severe accident depend on the sequence and the progression of the accident, including phenomena such as:

(For more details on these loads refer to the cited sections of Appendix 0 to '

Ref. 3.)

1. Containment pressure loading from prima ry system blowdown during and -

af ter core degradation, including decay heat [J.5]; ,

2. Pressure loading from hydrogen burn [J.4];

3. Pressure loadin from high-pressure melt ejection: (a) from a steam spite alone, (b f om a steara spike and direct containment heating, and j (c) from metallic-chemical redction and direct containment heating, including hydrogen phenomena [J.5];

4. Pressure from noncondensible gases generated during core-concrete interaction [J.6]; and

5. High temperatures late in the transient that could induce failure [J.2]. ,

There is still significant uncertainty in the scientific community about the adequacy of understanding of the phenomena leading to some containmerit loads.

• Nonetheless, confident predictions of containment performance under bounding estimates of the loads can serve the useful purpose of pcrmitting distinctions among scenarios likely to lead to early, late, or no containment failure.

Finally, there are two significant f ailure trades of containment thet are not within the scope of this research plan but are the subject of other research programs. First, there is the possibility, in some scenarios involving melt-through of the reactor vessel, of a leak path developing as a result of core debris contacting the containment (Ref. 4). Secondly, there is the possibility of pre-existing leak paths that could lead to fission product release from the containtrent if an accident were to occur (Ref. 5). A diagram is shown in Figure 1 that puts the research program on containment integrity in thc context of other efforts in research on sever'e accident consequences, 1-2

l1ll1 ij i l

s e

d o

M el n sy r

a io t

eo t l tn c len ler e ee i dd d ds v oun on e i Mepf r Moee ee oerly P dnb l

gd e b o l

a k e sa nl n e lotla a t c

n hni a sa v amfela ev eie v e et r sa L (t r ei n u hnf i

Tf o q e e r s e n v o e C S f

o t

x t

e n

s o e C d

o in M t n

e e r e r

e y u la u r i t

t n

t a hcs ht m l

u r e r gi e le ia F

s s

g e ml p e ~c e u nal v o a p E e t inanm ir mus ohav r r t

t f r op I n

er e g

a s hc o usee k e h n n t

er ldn c ntet yre ll t

e h t, g n t

n ond ,s emo le o Ic r a

e e y, n dp gr s hsp g m u omry cin o oo ion sns n e n m ma lie oe s i

a r na hid n

ia f

o or er ,it a

le hng r n klat s

leicar f

otir ac it is e t

t a t t n t ef r

u , r ,t sei eri x e

R ut s tl rt rt n n ge o

te nu Mcb oo C

o liahr s

fe e ,n , e lenk eic ru lat c ce n e luem ns e r

t y

t ol e iaeie C Fsp pp Sob Lep Fps P i

r .

gh le ec t r ib , na s

s ) I e

g t s o ne P. eR mt ie nn tad i a nc n oc e CA m 1 o

n t t e

r e n

.n en t

h e u e ed g P e m r mi i t

n r n oia un cc t i a s

t n

F ct aa e e nn r t eni s v d eo pos e ic ltonc mc ba ic c i en m A mh ti n hg fit dti is e ow nis e r

la s awed s e s t es e v r c eu r ed uin g

r e pd s r o e S so s t e

L ir Dp ri se le Phb b

i s

s o

P g

,lll

2. REGULATORY HEEDS AND RESEARCH OBJECTIVES On August 8, 1985, the U.S. Nuclear Pegulatory Commission issued a policy statement on severe accidents (Ref. 6). The policy statement provides criteria and procedural requirements for the licensing of new plants and sets goals and a schedule for the systematic examination of existing plants. On the basis of available information, the Commission concluded that existing plants pote no undue risk to the public and the Commissicn sees no present basis for immediate action on generic rulemaking or other regulatory changes for these pl6nts because of severe accident rish. However, the Commission emphasized that systematic examinations of existing plants are needed, en-couraged the development of new designs that might realize safety benefits, end stated that the Commission intends to taf e all reasoncble steps to reduce the chances of occurrence of a severe eccident and to rritigate the consequences of such an accident, should one occur. The NRC staff has developed an im-plenentation program for the policy statement. The program will accomplish the goals of the policy staternent within the schedule specified (Ref. 7).

For existino nuclear power plants, the Commission specified the formulation of a systematic approach to an examination of each plant now operating or under construction for possible severe accident vulnerabilities. The systematic approach will be developed curing the ? years following issuance of the policy sta terrent. The examination of each plant will be perfonred by the licensees.

Vulnerabilities ictntified by this process will be evaluated by the licensees in deciding whether corrective actions are needed. Any generic design chenges thet are icentified by the NRC statf as necessary for public health and scfety would be required through rulemaking.

The Severe Accident Policy Implementation Program provides for coordinated efforts to ensure the fulfillt.ent of the policy contained in the policy state-ment. The implementation program incorporates several major elements. One element is to forrtulate an integrated, systematic approach for exar,nning each nuc1(ar power plant now operating or under construction for possible signi-ficant risk contributors that might be plant specific and might be missed witheut a tystemetic search. The examination will pay specific attention to contaitrent perforr.ance in striking a balance between accident prevention and consequence mitigation.

The major source of risk to the public from the operation of nuclear power plants stems from eccidents trat lead to a containment failure. The regulatory concern is that the failbre rhodes and associated load levels for containment structures cannnt be predicted with any real conficence by the methods useo for esign. This is especially to if the contemplated f ailure mede is lo-calized leakage. Both assessrrents of the risk posed by lcads outside the cesign basis enc estimates of the effectiveness of proposed mitigative steps rec,uire an ability to pred'ct the way in which a containment will fail.

?-1

A final issue stems f rom the results of probabilistic risk analyses of nuclear power plants that have included attempts to mocel earthqueke risk. In those unalysts, the possibility of earthquakes larger than the design besis is included among the potentiel initiatcrs of a severe accident. Typically, it is found that very large and unlikely, but not impossible, earthquake occur-rences are among the dominant contributors to the probability of a severe acci-dent. A coherent picture of the risk associated with those accident scenarios requires an improved understanding of how tolerant the containment pressure boundary is to carthquokes much larcer then the design basis. The fundarrental question is: Will containment performance in a severe accident be signi-ficantly degraded if the accident were initiated by a large earthquake?

The research objectives are:

1. To identify likely failure modes of the wide va riety of containment designs under the loadings associated with the dominant severe accident scenarios.

2. To verify the ability of state-cf-the-art calculational methods to predict actual failure modes and develop improved calculational techniques, if necessary.

3. To provide a basis for reducing the uncertainty in estinctes of contain-ment failure times used in consequence calculations.

4. To assess the ability of different containment designs to sustain challenges from very large earthquak(s.

/

2-2

l

3. SCOPE OF RESEARCH Research on containment failure modes is based cn the observation that excessive leakage can occur, basically, from four sources:

1. Failure of the shell, either the containment shell itself in the case of steel containments, or the liner in the case of concrete containments;

?. Leakege at large penetrations as a result of the inelastic deformations and/or degradation of seals and gaskets

3. Leakage at electrical penetrations due to degradation of materials under the high temperatures associated with accident scenarios; and

4. Leakage through valves due to pressure and temperature effects.

Suntaries of results to date and planned efforts are presented in the following sections. Research related to shell failure or deformations of penetrations  ;

rests on analyses of and experiments on model tests of actual containment designs. These tests involve pressurization up to failure levels under ambient temperatures. Since seal and gasket materials are adversely affected by the temperatures associated with severe accidents, separate tests focusing on the development of lea! age are perforn,ed under pressure end temperature ccoditions, usually at full scale. Experiments to examine the possibility of developing leakage through electrical penetration assemblies and valves also require experiments under temperature ard pressure conditions at full scale.

3-1

4 RESULTS TO DATE 4.1 Containment Model Tests Four 1/32-size steel models of three configurations were built and tested at Sandia National Laboratories in 1982 and 1983. The first configuration, of wh1ch there were two models, was termed a clean shell. Geometrically, the clean shell is a right circular cylinder with one end welded to a hemispherical steel dome and the opposite ord welded to a thick base ring, which in turn is bolted to a rigid testing fixture. The diameter of the cylinder was about 43 inches (1.1 m) and had a height of 65 inches (1.f,5 m) including the dome. The thickrcss of the cylinder and dome material was about 0.045 inch (1.15 mm).

The basemat was not modeled. The second configuration was a ring-stiffened containment that used the clean shell geometry with the addition of ten stif-fening rings brazed to the cylinder wall. The third cr.nfiguration was a penetration model thet is also based on the clean shell geometry. Three penetrations are included in this model, representing two personnel locks and er, equipment hatch. The containment models were pressurized incrementally with nitrogen gas. Strain and displacement data were recorded at each pressure increment. Due to the irherent dangers of pneumatic pressurization, testing was performed remotely in an isolated area. The steel containment models were instrumented with high elongation strain gauges, several displacement gauges, pressure transducers, and thermocouple to acquire data during the testing of the models. A coordinate determination system, which uses theodolites and the principles of triangulation, was also used to measure large displacements.

For the geometries investigated, stiffening rings attached to the containment wall increased the pressure at which the majority of the wall yields and in-creased th( ultimate strength of the vessel. Treating the stiffening rings implicitly, by increasing the thickness of the cylinder wall by a volume equal to the increase in volume represented by the stiffening rings, i.e., smearing of the rings, was found to be a reasonable analytical procedure to predict ultimate capacity.

The penetrations in the penetration model did not have a thickened shell area around the penetration sleeves and no gaskets or seals were included in the penetrations. For these simplistic geometries, the presence of penetrations does not significantly decrease the capability of the containment. The ten-dency of ductile steels to flow elastically mitigates the effect of discon-tinuities such as penetrations. In the absence of severe flaws, estimating failure pressure using an equivalent plastic strain criterion does an adequate job for the simple geometries investigated with these analyses.

Finally, a 1/8-scale steel containment model was tested to determine its response to pressure levels exceedin3 the design basis. Extensive structural analyses of the model were performed prior to the test. A number of pene-trations were present in this experimental model, including operable equip-ment batches with single "0" ring seals, personnel lock representations, and 4-1

-_,__.__._w, -

a constrained pipe. The model was built to ASME Codt specifications with a design pressure of 40 psig. An extensive structural data base was generated auring the high-pressure test of the 1/8-scale steel containment model, which was conducted November 15-17, 1984. Data were recorded at 21 different pres-sure levels up to and including 190 psig, which is 4.75 times the design pressure. The model ruptured after the pressure in the model was increased to 195 psig. No significant leakage was detected up to this point, although the measured displacements around the equipment hatch indicated that leakage was irminent. The membrane strdins, which denote strains in the cylinder away from the effects of penetrations, were between 2.5 percent and 3 percent at 300 psig.

The containment shell is defined as the. pressure boundary of the containment, exclusive of those parts of the boundary made up of penetrations, including k equipment hatches, personnel airlocks, piping, and electrical penetration '

assemblies. A loss of integrity associated with the containment shell must involve a cohesive material failure, i.e., a throughwall crack or tear in the pressure boundary. If a throughwall crack or tear is developed, it is highly likely that rupture (unstable crack growth) will rapidly ensue.

Based on results of tests on stcle models of steel containment buildings that were pressurized to failure, a simple s' rain criterion appears to be adequate {

for estimating the point at which a throughwall crack or tear will develop.

The steel most commonly used in U.S. containments, A516 Gr 70, has exceptional ductility, and very large plastic strains may be developed prior te failure.

The following criterion for evaluating rupture follows from the test results:

Rupture will occur if, at any point on the pressure boundary, the equivalent strain exceeds the material's ultimate strain (the strain at rnaximum load as determined from a uniaxial tensile test). If the strain is primarily due to bending, somewhet higher strain may be tolerable (to account for crack arewth through the thickrtss).

The above criterion should be used only with detailed three-dimensional finite-element analyses, which woula need to include at least major penetrations and stiffeners. Results can te very sensitive to design details such as stifferer patterns arcurd penetrations. Any failure criterion must be ccnsistent with the dudlysis USed to CdlCulate the PCsponse measures used in the criterion. If axisymraetric analyses arr used to predict rupture, the limiting value of strain used to predict rupture must be decreased tc account for the effects cf pene-trations are stif feners.

Fany of the discontinuities in the containment shell, such as the soringline and the bee embedment, result in self-limiting states of stress, and, con-secuently, a (chesive material failure is not libly to initiate in these areas. On the other har.d, large penetrations cause local strain concentrations in the shell thet reduce the capacity relative to an undisturbed shell.

Stifferrrs tend to increase the gerrral yield pressure, but they do not necessarily ircrease the ultimate strength by a like amount.

4-P

There is a large body of evidence that indicates that, if a throughwall crack or tear is developed, a steel containment shell will rupture catastrophically.

Kiefner (Ref. 8) developed Empirical eGuations based on a large number of tests on pressurized cylinders that shew that rupture will cccur unless the initial throughwali crack is the result of a deep surface flaw (depth of the surface flaw must be at least one-half the wall thickness). Four of the five steel containment models tested at Sandia ruptured; in the one exception, leakage from a throughwall crack was sufficient to depressurize the model.

, However, in that particular model, the cylindrical wall had been severely thinneo by excessive grinding adjacent to a repoireo weld, which was equivalent to a deep surface flaw (Ref. 9). The formulas developed by Kiefner explain why rupture did not occur in that model. A final example is the J-integral ana-lysis of the Sequoyah containment conducted by Greimann (Ref. 10). Greimann's analysis showed that a throughwall crack originating in the thinner wall sectior.s near the springline would not be arrested at the thicker well sections used at lower elevations, and he concluded that unstable crack growth (rupture) would occur.

It seems clear that steel containments are highly likely to rupture if a throughwall crack is developed at any point in the containment shell. It is assumed that initial surface flaws do not exceed 20 percent of the shell thickness.

Any penetration that is intended to be used for access to the containment is Cdtegurized as an operable penetration. In u.s. steel containments, this would include equipment hatches, personnel airlocks, escape hatches, and BWR Mark I and Mark 11 drywell heads. The potential for leakoge between the sealing surfeces of these types of penetrations is outlined below.

In general, leakage from operable penetrations depends on both displacement and performance of the seal. Tests have shown that if metal-to-metal contact is maintained between the sealing surfaces, significant leakage does not occur regardless of the condition of the seal material (Ref. II). Also, because the seals are typically compressed one-quarter of an inch, large separations of the sealing surfaces can arise without significant leakage if the performance of the seal is not compromised.

Relative displacements of the sealing surfaces can be categorized as either separation (relative motion perpendicular to the sealing surfaces) er sliding (relative motion in the plane of the surfaces). Tests on seal and gasket materials (Ref. 11) suggest that there are at least two ways in which per-formance of the seal can be compromised: (1) radiation or tgermal aggrg can result in a loss of resiliency, and (2) high temperatures (500 F to 650 F, depending on the type of material) degrade the seal materials, that is, the materials outgas, then become cry and powdery.

In special cases, leakage may be a function of performance of the seal only or sealing surface deformation only. In personnel airlocks with inflatable 4-3

seals, the door and bulkhead are not designed to be in intimate contact, and, therefore, leakage will arise if the performance of the seal is compromised, irrespective of structural deformations. The performance of the seal may alsc be the only factor preventing leakage if there is significant out-of-flatness ,

between the sealing surfaces. This is a condition that would not be detected by integrated leak rate tests; at design levels, seal performance prevents leakage. However, in RWR Mark I and Mark Il plants, severe accident temper-atures are expected to be sufficient to degrade seals, and out-of-flar ess could result in leakage without additional separation or sliding of the see.ing surfaces. In cases where sliding or separation of the sealing surfaces is very large, leakage can result esen if the performance of the seal is not com-promised.

4.2 Penetration and Futerials Tests Because of the different types of trajor penetrations that exist in LWR nuclear power plants and because of the large number of designs that exist for a given type of penetration, a comprehensive survey was conducted on 48 U.S. plants to determine these variations for all the major penetrations (Ref. 12). The survey that was performed by Argonne National Laboratory includes all con-toinment types, materials, penetration designs, and all types of seals and '

gaskets. Baseo on these survey data, penetrations that are most susceptible to leakage were identified. To better evaluate the relative leakage potential for the different penetrations, a rumber of figure-of-merit analyses were made. These figure-of-merit analyses are based en the structural behavior of the penetration-containment system and on the geometry and material variations of the seals and gaskets that are used in the various penetrations. A com-parative fiaure-of-merit analysis was performed on different types of pene-tration designs to come up with 6 list of penetrations that are most sus-ceptible to leakage at beycnd design conoitions and, hence, may need to be testea to determine their leakage behavior.

Also, since the seal and gasket materials used in the major penetrations are -

expected to be a major cause of leakage of the penetrations under severe accident conditions, en extensive literature survey of the behavior and the capability of the dif ferent seal materials under different environments was conducted. This survey indicated that the performance and the capability of <

all the commonly gused seal raterials are affected by (1) temperature, parti-cufarly beyond 400 F, (2) acing cue to radiation beyond a dose level of about 10 -10' rad, (3) the combination of radiction and temperatures, and (4) the p steam environment (mainly f or silicone rubber). However, very little infor-mation was found in the litereture about their behavior in the environments of interest in severe accidents. Hence, a plan has been put in place to test some of the nore commonly used seal and gasket materials for severe accident con-ditions in order to understano their leakage behavior. Using the information from the survey of the plants and the survey of the seal and gasket materials, a test matrix for seals art gaskets has been developed. Some of the commer ceometries that will t< tested are 0-ring, tongue and groove, double dog-ear, ocuble gumdrop, and inflatable seals. /11 these geometries arc commonly used in large nechanical penetrations (e.g. , equi pment ha+ch, personnel airlock). ,

4-4

The 0-ring type is also commonly used in electrical penetration assemblies.

Sor:e of the materials to be tested are silicone rubber ethylene-propylene type rubbers (EPR/EPDM), Neoprene, and Viton. The test matrix for the seals and gaskets will evaluate the effects of radiation a temperature, linear scaling (i.e. , ef fects of the length of the seal)ging, , cross secticn, rotation between seal mating surfaces, end gaps between mating surfaces. Full-size cross sections of the seels are used in these tests.

Tests gere conducted in ritrogen, air, and steam atmospheres at temperatures up to 700 F and pressure at 155 psia. Both pressurized water reactor (PWR) and boiling water reactor (EWR) severe accident profiles, which are discussed in hference 13, are enveloped by this envirorpent. In a seal test, the initial temperature and pr essure are raised to 360 F and 155 psie.

are held overnight before raising the temperature to b [hesecondftions and holding for 2 hours2.314815e-5 days <br />5.555556e-4 hours <br />3.306878e-6 weeks <br />7.61e-7 months <br /> on successive days. Tests are terminated 500 , 600before

, and 700 700 [F is reached only if a seal fails ard leaks.

Principal results obtained to date show that:

o Seal materials degrade as e iunction of time and ten,perature. Depending en the seal traterial ana environment, degradation took the form of melt-ing, charrino, or brittleness.

o Leakage is lin.ited or prevented when metal-to-metal contact is maintained at the sealing sur face. In actual mechanical penetrations, the niagnitude of leakage under these conditions may be higher because the surface finish and flatness of the sealing surfaces is probably not as good as in our test fixtures.

o The amount of squeeze has no apparent effect on the degradation of the seal. Metal-to-metal contact (30% squeeze) masks seal degradation by pre-venting or significantly limiting leakage, o The degradation of silicore rubber under extreme environments is affected by its aging history. In a steam atmosphere, silicone rubber gaskets that were both radiated and thennally aged performed slightly better than unaged silicone rubber. Unageo silicone rubber gaskets typically failed c

to maintain a seal at about 500 F in a steam gtmosphere. Aged silicone rubber failed to maintain a seal at about 550 F. In both cases squeeze was fixeo. A decrease in squeeze during the test would probably result in leakage at icwer temperatures, o One-quarter-inch EPM /EPDM seals maintain a seal up to 650U F, regardless of aging history.

As expected, the double dog-ear and double gumdgop EPDM unaged seals testeo t;egan leaking at a lower temperature (600 F) than the 1/4-inch seals. (Typically the physical properties of a given material are degraded as cross section increases.)

4-5

o Samples are aged in a fixture at the same percent squeeze that is used during the severe accident test. When subjected to aging, either thermal or radiation, all materials retain a permanent set. As a result, any subsequent decrease in squeeze (i.e., any increase in gap size) Would result in leakage depending on the degree to which resiliency is lost.

o Silicone rubber degrades at a significantly lower temperature ir6 a steam envfronmentthanineitheranairoranitrogenenvironment (500 F versus 700 F). No silicone rubber seals other than 1/4-inch 0-rings have been tested.

o The degradation of EPM /EPDM is not affected by the atmosphere. One-quarter-igch EPM /EPDM 0-rings typically did not leak at temperatures up to 650 F in nitrogen, air, or steam atmospheres.

o The effect of pressure on seal degradation was not investigated, but it is believed to be insignificant. Leak rate obviously depends on pressure.

Higher pressures may also be more effective in eroding seal material, thus increasing leakage area, o Leakage typically occurs over less than 30 percent of the seal length. It is reasonable to expect that, given time, ercsion of the seal material would cause the leakage arca to increase.

4.3 Electrical Penetration Assembly lests As part of an effort to evaluate " containment integrity" under severe accident ccnditions, electrical penetration assemblies were singled out as having "one of the largest uncertainties associated with predicting the amount of radio-nuclides released" (Ref. 14). This potential led to a study of electrical penetration essemblies by Sebrell (Ref. 15) te assess the leak paths and potential for leakage through electric penetration assen,blies in light-water-reactor (LWR) contairiraert buildings.

Electrical penetration asserablies (EPAs) with the highest potential for leakage are (1) those with organic seals and gaskets, (2) those with elastomer 0-rings on header plates, and (3) those designed for low-pressure capability. Pe-garding the availability of the EPAs, of the identified 18 suppliers, only three suppliers ere active tocey from whcm actual full-size EPAs similar to those supplied to the existing nuclear power plants could be obtained. These suppliers ore D. G. O'Brien, Westinghouse, end Conax.

EPAs wete tested to determine their leakage under severe accident environments typical of both FhRs and BWPs. These severe accident profiles are based on calculations performed under the Severe Accident Sequence Analysis (SASA) program, also funded by the NRC. Where such information is not available, 4-6

Other sources, including data from probabilistic risk assessment (PRA' ttudies of plonts, wert used. The test motrix was:

Severe Accident EPA Plant Environment Conex BWR - Mark I 700U F, 13E psia Wes tir.ghouse BWP - Vark III 400g F, 75 psia D. G. O'Brien PWR 360 F, 155 psio The EPAs were aged for the equivalent of a 40-year life and exposed to radia-tien simulating that of a severe accident prior to the severc accident steam tests. Results are presented in Reference 12 ano are summarized below.

D. G. O' Brier PWR Test The D. G. O'Brien Severe Accident Conditions (SAC) test was completed according to the temperature and pressure accident conditions. There were no detectable leaks through thc FPA during the steam pressurized portions, including none through the header plate 0-ring seals, also called the aperture seals. A post-test air leak rate measurement at ambient temperature and 155 psia produced a leek on the order of 0.13 scc /sec. The electrical system on the EPA for all modules measured 1E+06 ohms to ground at 50 volts, and five out of the eight circuits were passing 0.5 amp to ground af ter 10 days. One-half amp is the maxin,um current permissible in the circuit designed for monitoring the current to ground. The posttest inspection showec' z.11 but one module to be electri-cally faulty, and the preliininary indication is that moisture traveled through the connector (also referred to as a plug) anc grounded out between the module pin and the rattal " mask" that surrounds each pin. It is this bridging with moisture or contaminants that causes the short to ground. There was posttest evidence of significant corrosion and chemical attack during this test. There was also a large amount of polysulfone expansion out from inside the plug bonnets and in some cases seizing between the plug and receptacle. It is important to recognize that this was a tgst that exceeded the design of the EPA. It was designed for 65 psig and 330 F. The g test had a pressure as high ds 155 psia and a temperature as high as about 390 F.

Westinghouse PWR Mark Ill Test The Westinghouse SAC test was completed according to the temperature and pressure accident conditions. There were no detectable leaks through the EPA, including the header-plate 0-ring seals, curing the steam pressurized portions of the test and during the posttest air leak rate measurement at 75 psia.

The electrical performance of the EPA during the steam test was more a func-tien of the EPA cables used rather than the EPA design. The first low re-sistance between an EPA cable and ground occurred in one of the ITT Surrenant thermocouple cables at 64 hours7.407407e-4 days <br />0.0178 hours <br />1.058201e-4 weeks <br />2.4352e-5 months <br /> af ter the start of the SAC test. The re-sistance to ground of one conductor dropped from 3E+06 ohms at 500V to 1.5 4-7

Kohm. The insulator was probably damaged by the 500V potential of the Megohm meter because the resistance dropped so quickly and then slowly recovered to 1E+C6 ohms at SCCV. All the EPA thermocouple insuletors failed in a similar manner; the resistcnce dropped suddenly from a value greater than IE+05 ohms to less than 3E+04 ohms after an IR measurement with the Vegohm meter. The use of the 50V to 500V potential to measure the insulation resistance of a thermo-couple c6ble is a severe test, considering that the cabit will only see a potential of itss than 0.1V. lhe measurement probably overstressed the cable and accelerated the degradation.

Conax BUR Mark I Test The Conax test, originally scheduled for a 10-day duration, had to be ter-minated after 8 days and 15 hours1.736111e-4 days <br />0.00417 hours <br />2.480159e-5 weeks <br />5.7075e-6 months <br /> because of operational difficulties at the pf facility unrelated to the EPA test. All significant findings had emerged long before the test termination, and there was no need to rerun the experiment.

There was no detectable leakage during or af ter the StC test. The threshold for detection is 1E-05 scc /sec. At about 5 hours5.787037e-5 days <br />0.00139 hours <br />8.267196e-6 weeks <br />1.9025e-6 months <br /> into the test, two EPA coppcr cable irs to ground had dropped to less than 2 Vohms measured with the DMM.

The remeining six cables had irs to ground between 2E+05 and 1E+06 ohms at 100 .

VDC measured with the Megchm meter. By 11 heurs into the test, all EPA copper tables had IPs to grouna itss than 1E405 at 50 VDC n,tasured with the Megohm meter and from 1500 to 1E+06 chms racesured with the DMM. The cetput of all EPA thermocouple cebles agreed with the temperatures nicasured with the moni-torina thermocouple during and after the SAC test. The inner EPA seal failed cbout I hour into the test, while the outer seals remained 1caktight throughout the test.

4.4 Purge and Isolation Valve Tests The Energy Technology Engineering Center completed elevated temperature and pressure leak testing on two 8-inch and one 24-inch butterfly valves typical of containtnent purge aiic vent valves.

The 24-inch valve passed the design basis elcvated temperature leaf test without leeking during or after the test. During the severc accident portion of the elevated temperature leak test, leakage was observed with the valve pressurized at 90 psig. The pressure was applied to the shaft upstream side of the. disk. The pccking was also The leakage started shortly after stabilization at 350' observed F. The to be leakine.

pressure was increased to 120 psig, S' and the packing leakage increased to the pointg that it was necessary to lower the pressure to continue the test. The 350 F hold period with the shaf t upstream was cerrpleted at 90 psig. The cpposite side of the valve was then pressurized and the pre.ssure increased to 120 psig. Durinc a 4-hour hold period at ?f 0'F, random leakage was cbserved; howave. , it was not constant.

After the valve cooled down, ambient leak tests were performed. With pressure epplied te the shaft ups tream side of the disk, seat leakage averaged 3P stcodard cubic feet per hcor (SCfH) at E0 psig and 320 SCFH at 125 psig. With 4

4-8

pressure applied to the cpposite f6ce of the disk, leakage was zero at all pressures.

The pressure-temperature combinations were also applied to the two C-inch valves for shaft-side and nonshaf t-side disk orientations. One of the valves ex-perienced leakage at most of the test points considered while the second 8-inch valve experienced no leakape either at design basis conditions or at severe accident conditions for all disk orientations. For the first 8-inch valve, leak-ages were significantly higher with pressures on the nonshaft side than for the shaft-sideorientation. Specifical,1y, at design basis conditiops (60 psig and 285 F),gthe leakage began at 58 cm*/mfr and infreased to 84 cm'/ min at 60 psig and 310 F. The leakage diminished to 32 cm /mirg when the pressure was in-creased to 90 psig with steady temperature cf 310 F. No leakage was observed for pressures cn shaf t-side orientations. The post elevated temperature test showed the leakage to decrease f rom 52 SCFH (at 50 psig) to 36 SCFF (at 125 psig) with pressure on the nonshaft-side orientation. For the shaft-side orientation, the leakage remained nearly constant at 20 cm /3min when the pres-sure was increased from 50 to 125 psig.

The capability of typical purge and vent valves to function and to prevent excessive leakage when subjected to leads resulting from severe accidents is important to the containment performance program. These results as well as test results to be obtained in FY 1987 will contribute to upgrading the model for estimating containment leakage.

4.5 Containment Buckling Tests Two efforts on the buckling behavior possible in steel containments were com-pleted at Los Alamos National Laboratory. The first, related to the ef fects on buckling Icads caused by penetrations in cylindrical containments, was aimed at the adequacy of the design practice rather typlicit prediction of containment failure, but did yield insights that will have applicability in detailed examinations of possible failure modes under seismic loading. The second, related to the post-buckling behavior of turispherical heads, applies directly to the performance of those steel containments with shallow domes, e.g., BWR Mark III.

The first effort was a comprehensive study to determine the effectiveness of the design of framing near penetrations for the prevention of buckling. Here, framing is defined as the stringer-like structural members that are attached to the shell and/or the reinforcing around and near a penetration where one or more circumferential stiffening rings are interrupted. The relative sizes of the penetrations studied in this work are typical of personnel airlocks and equipment hatches.

While the formation of a buckle does not necessarily mean that there will be a loss of containment, it would introduce conditions into the shell geometry and 4-9

material that could lead to failure. Therefore, in this study, formation of a buckle was the criterion used for evaluating the effectiveness of the

, framing.

1 It is well recognized th6t a sufficiently small penetration in a shell sub-jected to axial compression will not affect the buckling load because it merely represents a small imperfection, and there are usually other imperfections from f abrication or boundary conditions that have greater effects. As the size of the penetration is increased, however, its effect on the buckling f.apa-city will become greater than the effect of other imperfections, and then it will determine the failure load. Tennyson (Ref. 16) and Starnes (Ref. 17) re- ..)

port studies that demonstrate this effect. The sizes of penetrations studied in this project are large enough to reduce the buckling capacity.

Addition of reinforcement around a penetration is used to increase the strength and buckling capacity, ideally, to the value of the unpenetrated shell. Penetrations in steel containments for nuclear reactors must be rein-forced according to ASME Code rules (Ref. 18), which apply only to pressure loads. Code Case N-284 (Ref. 19) specifies the reinforcement required for buckling strength of steel containments, and it, in part, uses the area re- =

placement rule (/PM) of Reference 18. The question arises as to whether the ARM is edequate for buckling leads and, in addition, whether it applies to I penetrations with diameters over 10 percent of the containment oiameter, which '

Reference 19 does net address. There have been numerous studies of this problem, and the results are not entirely consistent. A few of these re-ferences will be cited here.

m ller (Ref. 20) used Mylar models having containment-like ceometries to study the effect of penetration size and the effectiveness of reinforcing on the buckling strength where the models are loaded in axial compression. Mylar re-mains elastic during the tests and, consequently, the models may be buckled n:any times with little changes in the buckling response. Miller constructed and testec several models, where unpenetrated shells had original buckling strengths of 50 percent or less of classical. Af ter introducing penetrations and using reinforcing of the ARM type, he found that the buckling load in models with penetrations cculd be restored to the value without penetra-tions. Snell (Ref. 21) tested Lexan mcc'els of an airframe structure in axial compression without penetrations and with reinforced penetrations. He also concluded that the addition of reinforcement around a penetration could restore the buckling strength of a shell to its original value.

The problem was studied by Cervantes and Palazotto (Ref. ??) with a numerical raodel based on the Structural Analysis of General Shells (STAGS) program. The linear bifurcation portion of STAGS was used, and both ring-stiffened and stringer-stiffened shells were studied. The work indicated that a reinforcing f rame aajacent to a penetration can increase the buckling load to or above that of a shell without a cutout. Starnes experimental work (Ref.17) was found to agrec well with their analysis.

4-10 a4

Tests were conducted on 6061-T6 aluminum shells having a radius-to-thickness retic of about 67C by Almroth and Holmes (Ref. 23). The shells were tested in the unpenetrated condition, ard with and without reinforcing around a penetra-tion. The buckling strength of the virgin shells was from 40 to 63 percent of classical, and the penetrated and reinforced shells had buckline loads that were always less than the original load. Anelysis with STAGS was in agreement with the experiraental results.

Dove, Bennett, and Butler (Ref. 24) conducted tests on shells fabricated from A3?] stainless steel in e study te evaluate the effectiveness of ARM in re-storing the static buckling strength of the per.etrated shell to the original value. The results showed that for the fabricated shells, that is, shells with Upical geometric iraperfections, the added reinforcing increased the strength; b'u t whether or not the original strength was obtained was obscured by the effects of the imperfections. It was pointed out that the capacity reduction factor used in design would maintain the desired margin to failure. These tests were particularly important because the steel was used for the raedel material and capacity reduction factors typical for steel construction were obtained.

In the series of tests conducted by Baker, Bennett, end Babcock (Ref. 25), the bucklino strength of six ring-stiffened shells having centoinment-like fea-tures was studied. The shells were made from a soft steel; two were unpene-trated and the others had one penetration each, of varicus sizes. Each pene-tration had reinforcing in accordance with ARM and freming representative of that used cn existing steel containment structures. The results of the buckl-ing tests on these shells were quite consistent and showed the following:

1. The buckling strength of the shell having the smallest penetration, cne which dic net interrupt any ring stiffeners, was unaffected by the penetration.

P. The buckling strengths of the shells having the larger penetrations were all reduced from the strength of the unpenetrated shells, the larger the penetration, the larger the reduction in strength.

It was also noted that the type of failure was a localized buckle, usually in a location adjacent to the ends of the vertical freming members. An analysis done by Feller and Bushnell (Pef. 26) for these models agreed well with the results, both for the collapse load and for the buckling modes. They also used STAGS as the numerical tool for performing their analyses.

It is of interest to note that the studies performed using plastic tredels and/or linear aMstic materials show that the ARM reinforcing restores the model to the unpenetrated strength, but tests on steel models are either inconclusive (Ref. 24) or show that the ARM reinforcing plus the framine ror-mally used around large penetrations may either not be adding to the buckling 4-11

strength or might even be decreasing it. Neither the plastic models nor the metallic models used in the references cited had a stress / strain curve repre-sentative of a typical containment stcel. This factor is most likely the cause of the plastic models showing full recovery of buckling strength with the ARM and of the steel models not always supporting this conclusicn.

The final effort (Ref. 27) was undertaken to determine the effects of framing near penetrations on buckling capacity for linear elastic materials. Based on the e<perimental work done for this investigation, several conclusions about the effect of f raming on the buckling capacity of a containment having penetra-tions that have been reinforced according to the ARM method were stated:

1. The presence on an untramed penetration reinforced in accordance with ARM has little effect on buckling capacity as long as the primary membrane stresses remain elastic.

2. Addition of framing up to an "overf ramed" condition has a small effect on buckling capacity as long as the primary stresses remain elastic.

The tests showed a maximum increase in capacity of 11 percent and a maximum decrease of 4 percent, the amount being dependent on penetration size, loca-tion, end loading condition.

The results of the concomitant finite-element analysis led to these conclu-sions (Ref. 27):

1. The addition of framirc to a model having an equipment hatch penetratico caused a change in the buckling capacity of 15 percent. Whether the capacity increases or decreases depends on the loading condition.

2. The primary menbrane stresses do not change significantly with the addition of framing.

A series of carefully planned ar.d conducted buckling tests was recenmended (Ref. 27) f or which the mcdels have containment-like features and the medel material has the same stress / strain curve as typical containment steels. The tests conducted to date have not riet these criteria, and there remain unanswered questions as to the buckling response of such models and for the prototype con-tainment and as to the applicability of the codes used in the design and ana-lysis of containmtrt framing rear large penetrations.

Lacking information from such a series of tests, it wab recommended (Ref. 27) that, for the short term, the 25 percent reouction in buckling capacity, as defined in the interim criteria currently being used by the NPC for licens-ing decisions, continue to be used. Results of the proposed tests should give the information needed to determine whether the 25 percent reduction in 6 12

buckling capacity is needed or should be changed. The tests would also provide information for evoluation of individual framing ciesigns and should provice inferrration for the incorporation of framing rules into NRC licensing criteria and formai ASME design codes.

In steel containments with shallow c' emes, compressive stresses are developed in the region of sharpest curvature, known as the " knuckle region." The pur-pose of another Los Alamos National Laboratory investigation (Ref. 28) was to detail the investigation of the knuckle region buckling ard failure for 4:1 (ratio of cylinder diarneter to head height) torispherical shallow dctre geo-metries typical of nuclear plant containments under internal pressure. Because the main interest of the NRC is in the reserve margin aveilable before loss of containment would occur, all experiments conducted in this progran were carried to rupture or loss of pressure-applying ability. In contrast, most experi-mental studies found in the literature oealing with knuckle buckling do not continue beyond or extensively discuss post-buckling behavior to rupture.

This program conducted several small-scale experiments on nominal 24-inch diorneter heads in preparation for larger experiments. The initial experiments featured two heeds that were joinec at their cylindrical skirt regions to form small pressure vessels. This configuration was acequate for buckling initia-tion, but because the failures that occurred at thc ring interface joining the l

two heads appeared to be related to the constraint there ano to the method of joining the head to the ring, a new configuration was tested that featured the head welded cn to an identical thickness cylindrical skirt that tapered into a standerd 24-inch diameter pipe segment. Finally, the prograrr culminated in two jointly funded tests carried out at the Chicago Bridge and Iron (CBI) research f acilities in Plainfinld, Illinois. One-half the funding for the large tests was provided by the NRC through Los Alamos with the rerraining one-half provided by the A5ME, CBI Industries, EPRI, EXXON, the Pressure Ves-sel Research Council of the United States ano Japan, and the Tennessee Valley Authority.

The following trajor conclusions were drawn:

1. If the vessels were cesigned in accordance with the ASME Code, the design allowable pressure is below the buckling pressure by a factor of 1.6 to 2.4. The ASME design pressure is conservative with respect to knuckle buckling even for fabricateo shells.

2. Clearly, the " margin-to-failure" in terms of loss of fluid for these shells is large. The ratio of failure pressure to design-allowable sure for the 1/8-scale fabricated containment heads (the CBI tests) pres- is 7 to 7.5. There is every reason to expect a similar ratio for a full-scale containment, which is an indication of much reserve margin in terms of loss of containment for these heads.

4-13

5. RESEARCH PLANS AND SCHEDULES This section presents summaries of activities planned in the four Irajor areas devoted to examining failure modes of containment structures subjected to pressure and temperature histories associated with scenarios. A milestone chart is shown Figure 2. These efforts are aimed at the major objective of providing a bas.s for improved estimates of containment performance in severe accidents. Finally, an outline is provided for the research planned on the effects of very large earthquakes on containment performance. This effort is motivated by ongoing programs aimed at including the effects of external events in probabilistic risk assessments. The basic ovestion concerning containm it performance is: If a severe accident were to develop as a result of a large earthquake, rather than from an internal event, would containment performance be significantly different?

5.1 Containment Model Tests A 1/6-scale reinforced concrete containment was designed b and Constructors to a design pressure of 46 psig (317 kPa)iny accordance United Engineers with the ASPE Code. The dicmeter of the rrodel is 22 feet (6.7 m), with a total height of 37 feet (11.3 m). The c and the deme is 7 inches (18 cm)thick. ylinder wall is 9-3/4 This model inches has #4 (12-mm (25 cm) thick, diameter) reinforcing bar for the eight layers of primary reinforcing (two layers of meridional, four layers of circumferential, and two layers of diagonal rein-forcing steel). The containtrent also has a steel liner, which is 1/16-inch (1.6 mm) thick for the base and cylinder wall and 1/12-inch (2.1 nm) thick in the dome. Two equipment hatches, two airlocks, constrained penetrations, and several piping penetrations are also included in the containment. All ma-terials used for constructing the model were selected such that their struc-tural characteristics are the same or as close as possible to actual con-tainment building materials.

lhe design of the containment model began in January 1985 at:d was completed in June 1985. To ensure that a high-quality rrodel could be constructed, the design of the model and a series of preconstruction tests were conducted in parellel. Due to the containment size and schedule constraints, particuler attention was peid to lir,er welding, splices for the reinforcing steel, con-crete placement, and other construction end sequencing problems. Ne problems were found to preclude the containment model's construction.

Construction of the containment model began in August 1985 after comments and suggestions made by a peer review panel were incorporated into the design.

The model's construction was completed in June 1986. During construction, over 250 strain gauges were applied to the reinforcing steel. In addition, several embedrrent gauges and thermoccuples were also cast in the concrete wall.

Postconstruction instrumentation includes over 100 displacement transducers and 350 edditional strain gauges, and other instrumentation is planned to be applied to the containment model. At test time, there will be approximately 5-1

1 nco p e s eT i

Y AmC F s s

t d Rs t t r s oe ie ic et nee yw pT e e S r o Osimml p al l Rv ieem r me^B lla a

Spo m inV ExC un So FnO o dt se sT es t rt dteK el os pe t r e 8 oU Gtsd eT sd ee t 8 Mn eo Rk Tt e

r ots 9 r c 1

eei PM y r

sl ps Y tei n s

alio^

r ip ee RT h F r g ce c n t

mA to m- o ye c r

nB ie n m n leC r v a o

Cno mmig S uo B al a e

si ir e mV s di et eeB mn e s cGSp s r u x tr ts nd g e SueO R et E o s e s tr s pT e is e te e sn eo dn s

Rs Dlp t r e.

rd s d os un t o

yi la ei PrCOtee g r r ae t

s e o [ets m

e sa s

e e, s

e Tde M s e lpB mg mta ^ *tsd T C RT y r r nt e t

n r on mM rA Put oe r e lude Ct i u n ee aP daOi tl o u 7 Tt t 8

fimd iat s S o e n s mE er ne am l 9

1 or e Fl e pGioT ne k lp c

e w mn ip m.

r w ia s o- uo oC F Y n ple et o t e td bm F gx p t t o m tar n s u ds e n et lomE r

iC o er le u B l

S ^" m e oT f

e t n

isE m s es a Te A c C D e e eC o Di co f

T mer t lp- r o ae l

mC uP ekm " P m is t r n e Ono swSsoaC G d ia S i L Ate t n

Pe Ep l

o Xm C n

s Ao^" r o NC f o

it cd Ots ie dt d Ce s e

6 ee e T 8 t it 9 Pp.

rl

" e g e

i v

1 t m lp n s in d Y so m gi ut e it c

TeC os F o Age e

- Ce nB he gT t

le A r l e o ie n d P d n t t as itPA ^ "m d e o s o io ie^" eE C e n M t c dT a

t n

R W le Cl ur dp n

la

/

R t

s am P ns e o on rC 2 GCig u e

leB e t a

r s sO t

r A e u pTe d Pde o

s n E i Fig 5

8 n g M nf e me r 9 yi r g is C ep eu Ts 1 e Y

F i ae nB D

/

R s t

im*

r B o s

e r

s mi."lde dsed O' C P lis n yt e .t ey o a Gse rl l t a Pan M a ei t .T CO D A /

ekin" Ss e I R G t

s n

e o it s s

T a t e t r s i s e t s et e i

t l e

t s k ns T v d e s t e e e e t s

PTy i

t o T s e g c G e T e A M s n T lal e g o d s b k

a e s y n it n k icr m e t

a a c w e la it lo t e L r

c n s e

t r

e l

a lor l cs e n e i e les v o A T n S A B EA le e

C P V

[N l

i I

l 1

l l 1,000 channels of instrumentation. In early 1987, final checking of the data acquisition system and the overpressure testing of the containment will be performed.

Effort related to the test to failure of the 1/f;-size containment model will be completed in FY 1987. Final attachment of the strain rosettes, displacement transducers, thermocouple, ano PTDs (resistance temperature detectors) will also be performeo before testing. Displacement transducers will be attached to measure displacements around penetrations, including deformations of the sealing surfaces, as well as general displacements of the containment shell.

Temperatures about the containment wall will be measured using thermocouple.

Results from these measuremer.ts will be used to correct strain gauge readings.

Air temperature inside the containment will be measured using RTDs. The results from these measurements will be used for leak rate measurements.

Optical measuring and recording systems will also be installeo prior to the testing of the containment model. A suite of video cameros, recorders, and monitors and still cameras will be relied upon for qualitative measurements as well as feedback for the conduct of the test.

When all data acquisition systems have been installed and checked, the mocel will be tested at low-pressure levels, much as a full-size containment build-irp would be tested. While collecting structural data, a Structural Integrity Test (SIT) will be conducted--pressurizing the containment building in steps to 1.15 tires its design pressure. The model will be visually inspected at its design pressure, including the mapping of substantial crecks in predeter-mined regions.

Following the SIT, a series of integrated ieak rate tests will be conducted.

The leakage of the containment acdel will be brought into acceptable standards (19/ day), if necessary, before further pressure testing is conducted.

Instrumentation will be added to the cracks developed during the SIT. After this has been accomplished and the data from the SIT has been digested, the contaircent model will be testec to failure. Foilure will need to be defineo during the test but will be tither structural failure of a compcnent or the containment, ur excessive leakege beyend the capabilities of the pressurt control system.

Pretest analyses of the 1/f-scale reinforced concrete conteinnent model will be completed. Results will be obtained using other axisymraetric and three-diroensional mcdels. Investigation of the possibility of local failures oue to steel liner-concrete interaction will also continue. Comparisons of prttest predictions and experimental results and a r, assessment of capabilities for predicting the beFavior of reinforced concrete containments will be completed.

Because of anticipated shortcomings in the state of the art, exterisive posttest analysis and code modifications roay be necessary to understand differences between predictions and the experir* ental results.

5-3

Nine institutions (three from the U.S., three from the United Kingdom, and one each from the Federal Fepublic of Germany, France, and Italy) are porticipat-ing in the model test.

Efforts to coordinate the activities cf those participating in the analysis of the 1/0-scale model will continue. Two topical reports will be issued; the first will include a discussion of each participant's pretest predictions for the perfonnance of the model, and the second will document comparisons with experimental data as well as any posttest analyses. Participtrts will be responsible for the sections in these reports that describe their findings.

In addition, there will be a posttest meeting at Sandia to "eview the outcome of the test and e special session at the 9th Conference on Structural Mechanics in Reactor Technology that will be devoted to discussion of the testing and c analysis of the model.

Efforts to extend the results of the reinforced concrete model test to permit applications to prestressed concrete containments will continue into FY 1988.

The centerpiece of this effort will be a cooperative test to be performed in the United Kingdom. A 1/10-size rrcdel of the Sizewell B containment is planned for cons truction in 1987. The Sizewell B containment is very similar to i containroents of the SNUPPS design used in the U.S. A test to failure is planned for 19EC. This model will provide a vehicle for testing the applic-ability of methods verified by the reinforced concrete model test.

The hRC research effort has proceeded with cognizance of and cooperation with an EPRI-sponsored effort on containment integrity. EPRI has sponsored one of the pretest calculations performed for the concrete containment model test and will participate in the posttest analyses of the results. In addition, EPRI has sponsored research complementary to the containment model test.

EPRI's concrete containment integrity research (Ref. 29) is airred at estab-lishing the true failure modes and load-carrying capabilities of reinforced and prestressed concrete containments unaer internal pressures beyond those for which they were designed. The immediate goal is to develop a test-verified analytic tool for evaluoting containment integrity. The ultimate goal, in conjunction with overall risk studies, is to characterize leak rates and radioactive releases as functions of pressure and time under various postulated severe accident scenarios.

The approach of EPRI's containment integrity research has been to conduct experimental and analytic work in parallel. To avoid the uncertainties os-sociated with small-scale modeling of detailed structural response and leak-age characterization, EPRI's experimental program features tests of large- and full-scale segments of concrete containments. Work sponsored by EPRI at the Construction Technology Laboratory of the Portland Cement Association began with simple tests on structural elements to define material behavior and progressed systematically to rnore prototypical tests of containment segments with penetrations and structural discontinuities to provide data on realistic liner failure mechanisms and leakage.

5o

The analytic effort, conducted by Anatech International Corporation, has focused on verifying and applying the nonlinear finite-element code ABAQUS-EPGEN for predicting the pattern of concrete cracking and its resultant inter-action with rebars and liner that could produce local failure of the liner plate.

5.2 Penetration and Materials Tests Analytical models have been developed for both the scale representation of an equipment hatch in the 1/6-scale concrete contain_ ment and the full-size per-sonnel airlock. Testing of both pressure-seating and pressure-unseating equipment hatches, at 1/6-scale, will be a part of the concrete containment model tests planned for mid-FY 1987. In addition, a full-scale personnel airlock, with pressure-seating seals, will be tested to failure in FY 1987 under pressure and temperature? conditions. This test will be a key element in assessing the applicability of the seal and gasket materials test data for predictions of performance in actual penetrations.

Inflatable seals, aged and unaged, will be tested to pressure and temperature profiles for both PWR and BWR Mark III severe accident environments. These tests, planned for late FY 1987, seek to determine the influence of the severe accident ten:perature (and seal degradation) and the severe accident pressure on leakage through intlatable seals. To obtain the necessary test temper-atures, steam will be used to pressurize the test chamber. Leakage will be measured by condensing the steam that leaks past the seals.

Finally, a search of literature and operatina data concerning bellows con-nections is being conducted. Plans for testing bellows will be formulated pending completion of the literature search and collection of operating cata.

It is anticipated that bellows could be evaluated for the effects of terrper-ature and loading such as torsion, internal pressure, external pressure, axial deformation, and lateral motion on stability and leakage (rupture).

5.3 Electrical Penetration Assembly Tests Testing was completed in FY ME6. A final report will be published in FY 1987 documenting the results of the tests.

5.4 Purge and Isolation Valve Tests Tests of containment isolction systeci valves (butterfly valves, small globe v61ves, and gate valves) under combined pressure, temperature, and loaaings due to containment and displacements will be ccrrpleted in FY 1987. Typical piping supports ana penetratiert will also be used. The purpose of these tests is to determine whether the loads transmitted to the valves through the piping will prevent the valves from isolating under ccnditions caused by severe accidents.

5-5

5.5 Seismic Lffects on Containment Performa,n,ce Containment structures correctly designed and fabricated in accordance with Section III of the ASME Boiler and Pressure Vessel Code can be expected to perform adcquately even if subjected to seismic loads for in excess of their cesign basis. The fundamental reason fcr this expectation is that the structures were proportioned srd detailed so that they would respond elas-tictlly when subjected to the combined effects of a LOCA and the safe shutdown earthquake. Although different parts of a containment are most affected by seismic loads and internal pressures, the net outcome of including the effects of both loadings is to significantly increase the capacity of the containment to withstand either loadino, considered separately. In addition, the design criteria specified essentially elastic response. Since significant yielding must occur before the cnset of failure, that aspect of design practice adds to the margin above design levels that can be expected in any containment.

Estimates of structural capacities for a number of containment shells have, Two studies indeed, are indicated especially worth large expectea noting. One marg (inssunmarizeo Ref. 30) in earthquake resistance.

estimates of failure levels for concrete containments made in connection with seismic PRAs. The other (Ref. 31) focused on providing estimates of the resistance of steel containment shells that could be used in probabilistic evaluations. The plants studied were all located in the Eastern U.S. (i .e. , east of the Rocky Moun-tains) in regions of low-to-moderate seismic activity and had safe shutdown earthquakes ranging from 0.109 to 0.259 . The concrete containment study estimated median capacities ranging f rom 2.59 to 9.79 . The steel containment study provided mear. estimates of capacity ranging from 1.45g to 16.259 .

Considerable uncertainty is associatcd with these estimates. For example, in the concrete containment study, an attempt was made to include the effects of both randomness and uncertainty and derive capacity estimates that were felt to have a high confidence of low probability of f ailure (HCPLF). The net result of reflecting the uncertainties in this was to reduce the estimated capacities by a factor of about three--with HCPLF estimates of shell capacity ranging from 0.89 to 2.79 . Notwithstanding the large uncertainties, it is still apparent that structural failure of the containment shell is not a significant potential failure mode under earthquake loading.

It is more likely that, for soil sites, any containment failure would be associated with a foundation failure, leading to failure of piping. This potential failure mode has not been studied extensively. But, in one case where it has been considered, it has been found to be dominant. In the pre-viously cited review of seismic PRA studies (Rcf. 30), the median estimates for failure of the shell and for soil failure beneath its foundation slab were 2.99 and 0.99, respectively. When uncertainties were factored into the esti-mating procedure, HCPLF estimates were 0.8g and 0.3 9 , respectively.

The fundamental question is: Will containment performance in a severe accident be significantly degraded if the accident were initiated by a large earthouake?

5-6

A soil f ailure would amplify motions at piping penetrations and could create a leak path. Another possible failure mode is local damage at operable pene-trations, which might reduce pressure-retaining capability.

Research on containment failure modes under seismic loads will be carried out during FY 1986-1988 with the intention of assessing the ability of analytical methods to predict behavior near failure. It is important to remember that containment design technology has focused on assurance of performance under design conditions--not margins to failure. The first effort, currently in progress, is a detailed examination of six plants (three PWRs, three BWRs) to identify possible failure modes under large seismic loads. This scoping effort will be completed in the summer of 1987. The extent to which predictions of failure at the most likely locations rest on calculations that have been k verified by experiment is of particular significance. There is a distinct possibility that, notwithstanding the wide differences in details for existing containment designs, it will be possible to show with confidence that im-plausible earthquake levels would be required to cause failure at most loca-tions. But it is aisc likely that, for some designs, some failure modes will be predicted at levels of interest. Necessary experiments will be identified <

and planned during FY 1987-1989 and carried out in FY 1988 ana 1989. It is likely that the outcome of this research will, in the context of PRAs, improve confidence in raean or median estimates of failure level.

Discussions are currently under way exploring the possibilities for cooperation with the Japan Atoric Energy Researca Institute and the Ministry for Inter-national Trade and Industry on the question of containment performance under severe earthquakes. It is possible that exchanges of information about ex-periments and analyses conducted on this topic may be found to be mutually beneficial ana will be irrplemented.

,i 5-7 0

REFERENCES

1. U.S. Nuclear Regulatory Commission (USNRC), "Reacter Safety Study--An Assessment of Accident Risks in U.S. Commercial fluclear Pcwer Plants,"

WASH-1400 (NUREG-75/014), October 1975.

2. M. Silberberg et al., " Reassessment of the Technical Bases for Estimating Source Terms," NUREG-0956, July 1985.

3. USNRC, " Reactor Risk Reference Document," NUREG-1150 Vol. 3, Draft for Comment, February 1987.

4. A. R. Perkins et al., " Containment Loading for Severe Accic'ents in BWRs with a Mark I Containment," Brookhaven National Laboratory Report, Novem- q ber 1984.*

5. P. J. Pelto et al., " Reliability Analysis of Containment Isolaticn Systems," Battelle Pacific f:orthwest Labora tories , NUREG/CR-4220, PNL-5432, June 1985.

6. USNRC, " Policy Statement on Severe Reactor Accidents," Federal Register, Vol. 50, p. 32138, August 8, 1985.

7. USNRC, " Implementation Plan for the Severe Accident Policy Statement and the Regulatory Uses of New Source-Term Information," December 1986.*

8. J. F. Kiefner et al., " Failure Stress Levels of Flaws in Pressurized Cylinoers," Progress in Flaw Growth and Fracture Toughness Testing, ASTM STP536, American Society for Testing and Materials, pp. 461-481, 1973.

9. D. B. Clauss, D. S. Horschel, and T. E. Blejwas, " Insights into the Behavior of LWR Steel Containment Buildings During Severe Accidents,"

Nuclear Engineering and Design, Vol.100, No. 2, March 1987.

10. L. Greimann, F. Fanous, and D. Bluhm, " Crack Propagation in High Strain Regions of Sequoyah Containment," Ames Laboratory, NUREG/CR-4273, Ames, Iowa, Draft, July 1985.*

11. L.  ?! . Koenig, " Leakage Potential Through Mechanical Penetrations in a Severe Accident Environment," Proceedings of the Third Workshop on Con-tainment Integrity (Vashington, DC), NUREG/CP-0076, August 1986.

12. M. H. Shackelford et al., " Characterization of Nuclear Reactor Contain-ment Penetration - Final Report," Sandia National Laboratories, NUREG/

CR-?855, SAND 84-7139, April 1985.

• Available in the NRC Public Document Room,1717 H Street NW., Washington, DC.

R-1

13. C. V. Subramanian, " Integrity of Containment Penetrations Under Severe Accident Conditions," Proceedings of 8th SMiRT Conference, (Brussels, Belgium), Paper J 4/2, August 19-23, 1985.

14. D. H. Cook et al., " Station Blackout at Browns Ferry l' nit One - Accident Sequence Analysis," Oak Ridge National Laboratory, .JREG/CR-2182, Vol. 1, ORNL/NUREG/TM-4, November 1981.

15. W. A. Sebrell, "The Potential for Containment Leak Paths Through Electri-cal Penetration Assemblies Under Severe Accident Conditions," Sa ridia National Laboratories, NUPEG/CR-3234, SAND 83-0538, August 1983.

16. R. C. Tennyson, "The Effects of Unreinforced Circular Cutouts on the Buck- i ling of Circular Cylindrical Shells Under Axial Compression," Journal of 1 Engineering for Industry, Vol. 90, pp. 541-546, November 1968.

17. J. W. Starnes, "Effect of a Circular Hole on the Buckling of Cylindrical Shells loaded by Axial Compression," AIAA Journal, Vol.10, No. 11, pp.

1466-1472, November 1972.

18. ASME Boiler and Pressure Vessel Code,Section III, Subsection NE, Class MC Components, 1980 Ed.

l

19. ASME Boiler and Pressure Vessel Code, Code Case N-284, " Metal Containment Shell Buckling Design Methods,"Section III, Division 1, Class MC,1980.

20. C. D. Miller, " Experimental Study of the Buckling of Cylindrical Shells  ;

with Reinforced Openings," Advances in Containment Design and Analysis, M. D. Bernstein et al., Eds. (ASME), pp. 7-18, 1982.

21. R. F. Snell, " Experimental Eveluation of the Effects of Cutouts on the Stress State and Buckling Stability of the Isogrid Structures of the Thor Delta Interstage and Fairing," MDC Report G2873 April 1972,

22. J. A. Cervantes and A. N. Pala20tto, " Cutout Reinforcement of Stiffened Cylindrical Shells," Journal of Aircraft, Vol. 17, No. 3, pp. 203-208, March 1979.

23. B. O. Almroth and A. M. Holmes, " Buckling of Shells with Cutouts," Inter-natier,al Journal cf Solids and Structures, Vol. 8, pp. 1057-1071, M .

24. R. C. Dove, J. G. Bennett, and T. A. Butler, " Buckling of Steel Cylinders Containing Circular Cutcuts Reinforced According to the Area Replacement Method," Proceedings of the 198? Joint Conference on Experimental Me-chanics (0ahu-Maui, Haweli), May 23-28, 1982.

R-2

25. W. E. Baker and J. G. Dennett, " Buckling Investigation of Ring-Stiffened Cylindrical Shells with Reinforced Openings Under Unsymmetrical Axial Loads," Los Alamos Notional Laboratory, NUREG/CR-3135, LA-9646-MS, March 1983,

26. E. Meller and D. Bushnell, " Buckling of Steel Containment Shells," 1-4, Lockheed Palo Alto Research Laboratory report, LMSC D812950, September 1982.

27. W. E. Baker and T. A. Butler, "A Study of the Effects of Penetration Framing on Steel Containment Buckling Capacity," Los Alamos National Laboratory, NUREG/CR-4892, LA-10977-MS, May 1987.

28. J. G. Bennett, "An Assessment of Loss-of-Containment Potential Because of Knuckle Buckling for 4: 1 Steel Containment Heads," Los Alamos National Laboratory, NUREG/CR-4926, LA-10972-MS, Advance Copy.*

29. H. T. Tang and S. W. Tagart, " Containment and Piping Pesearch,"

Proceedings of the Fourteenth Water Reactor Safety Information Meeting, NUREG/CP-0082, Vol. 5, pp. 523-529, February 1987.

\

30. R. J. Budnitz et al., "An Approach to the Quantification of Seismic Margins in Nuclear Power Plants," Appendix C Lawrence Livermore National Laboratory, NUREG/CR-4334, UCID 20444, August 1985.

31. L. Greimann et al. , "Probabilistic Seismic Resistance of Steel Contain-ments," lowa State University, NUREG/CR-3127, January 1984.

Available in the NRC Public Document Room, 1717 H Street NW., Washington, DC.

R-3

Nac comu aan n am a u s. NucLEAn aEcutAToa v couurssiog g RE PQH T NvM8E R f 4mpnad Oy TfDC. Add Vd/ 40, #f

'l"',*3,' $ BIBLIOGRAPHIC DATA SHEET NUREG-1264 sti msravCTias oN Tai vEasE a Tira AND suer.Ta 3 LE AVE 8L ANK Containment Ir grity Research Program Plan 4 D A T E R E POR QMPLETED MONT-  % EAR

, AuT oais, July .

1987

. ofE .EPosT issvED vtAn MoNT f l Augustf 1987 7 PERFOAMsNG ORGANI2 ATION NAME ANO MA ~ NG ADDRE $$ U4clupe l@ Codef 6 PRO.f ECT/T WORK UNif NUMBER Division of Engineerr /

Office of Nuclear Regu atory Research * ***'*""'"

U.S. Nuclear Regulator fommission Washington, DC 20555 10 5PON$QA#NG OHGAN62 ATION N AME AND MAILING AOD Itaciude /W Curse.' i b its TYPE OF REPORT Division of Engineering Office of Nuclear Regulatory esearch Technical U.S. Nuclear Regulatory Commi ion * *E a'o o co *

• a ' o "o ** *'-d Washington, DC 20555 o suPPuMiN r AR, NoTis 3 Ae s r a A C r ,m -,. ., e.,,

(g Thisreportpresentsaplanforresekch n the question of containment performance in postulated severe accident scenari' It focuses on the research being performed by the Structural and Seismic Engineer 19 Branch, Division of Engineering, Office of Nuclear Regulatory Research. Summarisgftheplansforthisworkhavepreviously beenpublishedinthe"NuclearPower/Plad Severe Accident Research Plan" (NUREG-0900).

This report provides an update to flect urrent status. This plan provides a summary of results to date as wel as an o 'line of planned activities and milestones to the contemplated completion o the progr- in FY 1989.

h a

Id DOCUMt NT AN Aiv$ 5 e a f vvv0 ADS DE 5' ;PTO"3

\ 16 A V A.L A91 Li r v

$TATEMENT Containments Severe Accidents R Unlimited

'6 SECumiTV CL A55'FICATICA trnan oePoi n ,oi ~~ a as oPi N E Not o VE aM- ~

Unclassified tre.s reoom Unclassified 5 7 NUMetR OF P AGE 5 18 PRIC&

4

UNITED STATES snaat rount*ctass aArs NUCLEAR REGULATORY COMMISSION " S3jstn'S "^*

WASHINGTON, D.C. 20555 ,g*SAf Ej OFFICIAL BUSINESS ,

PENALTY FOR PRIVATE USE, $300

'1 0 0 5 5 3 0 7 'l 4 7 7 1 1ANIRD14Ml?1 LE NRC-OA:e* AJP DIV 0F ' U t_s SJCS c:0LICY ^ PUP MGT RR-PDP NUREG

..:- ? ! 1

~ Ass;Nu'rN DC 00555

- _ - - _ _ - - - _ _ _ _ _ _ _ - _ - _ -