ML20126E654

Draft Containment Event Analysis & Estimation of Source Term Frequencies,App to NUREG-0956

ML20126E654
Person / Time
Site:	Peach Bottom, Grand Gulf, Surry, Zion, 05000000
Issue date:	07/31/1984
From:	Behr V, Benjamin A, Haskin F, Murfin W SANDIA NATIONAL LABORATORIES
To:	NRC
Shared Package
ML20126E618	List: ML20126E616 ML20126E654
References
FOIA-85-73, RTR-NUREG-0956, RTR-NUREG-956 NUDOCS 8506170166
Download: ML20126E654 (56)

Text

..

c CONTAINMENT EVENT ANALYSIS AND ESTIMATION OF SOURCE TERM FREQUENCIES APPENDIX TO NUREG-0956 by V. L. BEHR, A. S. BENJAMIN, F. E. HASKIN, and W. B. MURFIN with Contributions by S. E. DINGMAN and R. D. GASSER CANDIA NATIONAL LABORATORIES ALBUQUERQUE, NEW MEXICO 87185 JULY 31, 1984 i

i f

i 8506170166 850311 PDR FOIA SHOLLY85-73 PDR DRAFT INFORMAL AND PREllMINARY AND AS SUCH :.1AY CONTAIN ERRORS NOT YET CORREC1ED.

FOR IN -00SE RkATE DISTRIBUTION AND NOT FOR EXTFJtNAL REEASE WITHOUT CONSENT OF

Amons,

/

+--%. -

,m.,,

e

,wc.w,,,,

t.-

e

-- - i,

,,yg--g-

--,p

--.-.,.-,,----7.--

1 TABLE OF CONTENTS I.

INTRODUCTION 2.

METHOD 2.1 General Description and Observations 2.2 Specific Example - Surry S D 2

2.3 Treatment of Verbal Descriptors 2.4 Treatment of Other Se,quences and Other Plants 3.

RESULTS 3.1 Results for Surry Sequences 3.2 Results for Zion Sequences 3.3 Results for Peach Botton Sequences 3.4 Results for Grand Gulf Sequen'ces 3.5 Sensitivities 4.

OVERALL

SUMMARY

AND RECOMMENDATIONS i

l DRIFT - INFORMAL AND PREllMINARY AND AS SJM.T/ C HTA! ! E Ur*.:. OT iET CORR:CTED.

FOR..!

!J:E ;:i A.E 0.ainiduil0N AND NOT 2-FOR LXTdRNAL RELEASE WITHOUT CONSENT OF MITHORL

.,,,--,.r.

-m--

4 1.

TNTRODUCTION The intent of this document.is to provide a limited risk perspective for the fission product source terms reported in the BMI-2104 documents.

One might ask, "Why is there a need for a risk perspective?"

i To answer the question, let us consider as an example, the BNI-2104 analysis of the sequence S D at Surry.

Two calcula-2 tions were made.

In the first, a break of a specified size (2-inch diameter) was presumed to occur in the cold leg through i

the reactor coolant pump seals, and the primary system coolant was presumed to end up in the containment sump.

The containment was assumed to fail from rapid pressurization due to events occurring just after meltthrough of the reactor vessel, and the containment sprays were assumed to fail at that time.

The release from containment was assumed to bypass the auxiliary building.

In the second, a break of the same size was assumed to occur

,in the hot leg piping.

Both the containment and the containment sprays were assumed to survive the events following vessel meltthrough, but the water flow to the reactor cavity was assumed not to prevent the core-concrete interaction from occurring.

Containment remained intact until basemat melt-through, at which time the containment partly depressurized.

While the events assumed in the BMI-2104 calculations are plausible, one might ask whether they are the most likely series of events that could occur following an S D, or whether they 2

produce the highest source terms that could occur within the realm of reasonable probability.

Is it more likely, for example, i

that the size and location of the break would be different from what was assumed in BMI-2104; that the containment sprays would fail prior to containment failure, because of debris in the containment sump plugging the pump intakes; that containment might fail by some means other than early overpressurization or basemat seltthrough; that the release pathway would be through the auxiliary building, where further reduction of the source i

term would take place?

These questions raise the need for a systematic identifi-cation of the various pathways that the accident can take and j

assessment of the likelihood, or probability, of each.

Such an analysis provides a basis for evaluating whether the source i

terms developed for a particular accident sequence cover the range of risk-significant source terms for that accident sequence.

In addition, it is important to recognize that the BMI-2104 analyses, by virtue of limited time and funding, did not consider all the accident sequences that are thought to be potentially important to risk.

They g

f a1

'IY FOR IN. ]J.lc.

1S h0T YETCORRECTED

,,,,:. e duidlBUTION AND N0f i

ERNAL RELEASE WITHOUT CON

s S

loss of feedwater transients (TMLU) or anticipated transients without scram (TKMU) for either of the two PWR dry containments, Surry and Zion, whereas the Accident Sequence Evaluation Program (ASEP) identified these accidents as being potentially important.

To determine whether the BMI-2104 source terms cover tL:

ange of significance in risk space, it is necessary als: A +.'d r a s G accident sequences not treated in BMI-2104.

The overall objective of this study are (1) to ic <Atfy accident pathways (i.e., combinations o'f accident seqe ences and containment events) that delineate source terms which may be important to risk, (2) to estimate the frequencies of those pathways and hence the frequencies of the source terms they attend, (3) to ascertain how well the BMI-2104 source terms cover the accident pathways that are important to risk, and (4) to identify accident pathways for which additional source term calculations are needed.

These objectives extend to the six reference plants addressed in BMI-2104.

In Section 2, we wi?1 describe the method we used to achieve the objectives stated just above.

First we will provide a general description of our containment event trees and the procedure we use for quantifying the branches.

Then we will provide a detailed example, illustrating the application of the method for a particular accident sequence in one of the i

reference plants.

The example will show how we used information from recently developed sources to quantify the containment event tree.

Then we will discuss some of the special consider-ations we made for the other accident sequences and reference plants.

In Section 3, we will present the results for all the sequences and plants we analyzed.

In Section 4, we will summarize the results and indicate accident pathways for which additional source term calculations are needed.

(By agreement with NRC, this version of the document.

submitted in July 1984, is somewhat narrower than the version to i

be submitted in November 1984.

This version treats only 4 plants - Surry, Zion, Peach Botton, and Grand Gulf - and for i

those four plants, only the accident sequences treated in BMI-2104 are considered, a total of 12.

At this time, we are not including phenomenology that is outside the capability of the BMI-2104 code methodology - most notably steam explosions and direct heating of the containment atmosphere.

Accident sequence frequencies account for some but not all of the plant modifications that have occurred as a result of the TMI accident.

The November version will address each of these items and remove many of the limitations.)

As a final comment, it is important to differentiate between the objectives stated above and those of a risk assessment.

We do not calculate risk here, only the frequencies of source terms.

I Evaluation of risk requires two additional steps:

(1) estima-tion of source terms for important sequences and accident pathways not treated in BMI-2104, and (2) determination of the mean consequences associated with each g r,c y g.

T gs R E ' M0T YETCORRLCIED.

_4_

79 ' '

'4-iTRiduil0N AND NOT FOR UT:.4:e m aaiAS nlTHOUT CONSENT OF f

M

. 2

4 objectives are part of the Severe Accident Risk Reduction Program (SARRP), which will use the results reported here to calculate risks for the 6 references plants.

That analycis is scheduled to be completed and documented by summer 1985, 2.

METHOD 2.1 General Descriotion and observations In traditional probabilistic risk assessments (PRAs), the accident pathways that contribute to risk are described by two types of event trees.

System event trees" are used to define the spectrum of accident sequences (i.e., the combinations of accident initiators and subsequent system failures) that can Containment event trees" are used to lead to core melting.

define the containment failure modes which lead to fission product releases beyond the containment boundary.

In our analysts, we take the accident sequences to have been previously defined by the existing PRAs, and we obtain estimates i

of their frequencies of occurrence from the Accident Sequence Evaluation Program (ASEP).

The sequence frequencies are provided in the form of central estimates and upper / lower bounds.

Our primary focus is upon the containment event trees.

We have developed a containment event tree for each accident j

sequence analyzed in this study..Because many of them are i

similar, we intend to combine them into a single containment event tree for each type of plant.

i Our containment event trees are considerably expanded beyond those considered in many previous PRAs.

We ask the following types of questions:

i (1)

Reactor coolant system failure modes.

What is the size l

and location of the reactor coolant system breach and l

the pressure in the system at the time of breach?

(2)

Containment system survivability.

Do the containment i

sprays, fan coolers, and suppression syntens survive I

the conditions occurring during severe accidents that l

exceed their design bases?

1 l

(3)

Containment failure modes.

What are the loads that i

challenge containment, does containment survive these I

loads, what is the nature of the failure (approximate i

size and location), and what is the subsequent pathway i

for fission product release to the environment?

The questions on the containment event trees are posed in ways that require the answers to be expressed in terms of DRAFT INF0WAl AND PREUMINARY AND AS

" ~0i YET CORRECTED.

33

IM10N AND NOT FO!:

/

FOR dT_M.. mast. WlIHOUT CONSENT OF

/

. AUTHORS.

---

r-c-----e----e--w-

-w--v-v se - -

-e-e--w-w-ww

- - -, - - + - - + - -

---

4----,-

r-mer-%

=m-w-w-r--e--wet----w----m----sr--u-e-wm--+------

ww

1 i

likelihoods.

For a loss-of-coolant accident, for example, we i

might ask how likely it is that the reactor coolant system breach will be in the cold leg piping as opposed to the hot leg piping; or how likely that containment will fail due to a hydrogen burn following reactor vessel failure.

Answers to such i

questions require information about the reactor design, the

+

phenomenology of reactor accidents, and the capabilities of containment.

For example, to answer the two likelihood questions i

just posed, one would need to know about the characteristics of the cold leg versus hot leg piping, the amount of hydrogen generated prior to vessel breach, the availability of ignition l

sources, and the failure pressure of the containment.

l l

Some of the issues addressed by the containment event trees are listed in Table 2.1.

We point out that these are not the events themselves, but rather the issues that must be addressed in order for the event trees to be quantified.

Also shown are l

the subsets of these issues that have been considered in some of

)

the recent PRAs as a basis for defining fission product release l

categories.

Observe that none of the PRAs account for all the issues we consider for binning source terms, but the most recent one (Seabrook) accounts for more than the others.

We have utilized a large number of sources to obtain the I

needed information, including the following:

(1)

Containment Loads Working Group (CLWG), References 1-3.

(2)

Containment Performance Working Group (CPNG),

Reference 4.

(3)

Battelle calculations for Accident Source Term Project i

Office (BMI-2104), Reference 5.

j (4)

Quantitative Uncertainty Estimate for the Source Tera j

j (QUEST), Reference 6.

(5)

Industry Degraded Core (IDCOR) program, Reference 7.

i j

(6)

Severe Accident Sequence Analysis (SASA) program, r

References 8-12.

l (7)

Severe Accident Uncertainty Analysis (SAUNA),

Reference 13.

(8)

Accident Sequence Evaluation Program (ASEP),

References 14-15.

5 l

(9)

SARRP Phenomena Assessment Task Force (PATF),

Reference 16.

(10)

Available probabilistic risk assessments (PRA),

References 17-22.

DRAFT - INFORMAL AND PREUMINARY AND AS SUH / A" O FAIN ERRORS NOT YET CORRECIED.

FOR 1 N E il/ ATE DISTRIBUTION AND NOT FOR ixTERNAL WELEASE WITHOUT CONSENT OF i

NM05, 4

l

.- -.-m..~

._..___.-..-_-_c-,v-.-

_m,-

-s.-

.... ~... _... _.. _

1 (11)

Final Safety Analysis Reports (FSAR), Reference 23.

(12)

Architect-engineer (AE) and other estimates of containment failure pressure References 24-25.

(13)

Filtered-Vented Containment System (FVCS) reports, References 26-27.

(14)

Others, References 28-30.

We encountered several problems in attempting to utilize l

information from these various sources.

One of the biggest was incomplete coverage.

Table 2.1 illustrates the relationship l

between some of the issues addressed by the event trees and the information provided by the two containment working groups, the 1

two ASTPO studies, and the IDCOR program.

It is clear from the table that the results from these studies address only a j

fraction of the questions asked.

When a question was addressed by one of the studies, the j

information provided often required us to make extrapolations.

)

For example, the Containment Loads Working Group provided estimates of the size of steam spikes for only the PWR reference l

plants, and then with preconditions appropriate for only one i

l accident sequence.

We had to extrapolate this information to i

other plants and other sequences.

The same was true for the analyses of global hydrogen burns, diffusion flames, and l

containment temperatures achieved from core-concrete inter-actions.

Similar statements apply to other studies.

i Furthermore, the information provided to us often did not specify a single best estimate but rather a range of possible i

values.

In particular, the CLNG and CPNG generally declined to 4

l provide best estimates of containment loading and performance l

and instead provided low, medium, and high estimates.

In the l

CLWG, concensus was generally reached more often on the low and high estimates than on the medium estimates, whereas the CPWG j

stated that all three of their estimates were highly conjectural and subject to chango.

Neither group felt that it was appropriate to specify weighting factors or probability distributions for their results.

I When we quantified our containment event trees, therefore, we also propagated three separate estimates -- optimistic, I

central, and pessimistic.

Thus, we derived three sets of l

accident outcome probabilities for each sequence.

The one l

labeled " pessimistic" tends to provide higher probabilities for the pathways that lead to higher source terms and lower probabilities for the lower source tera pathways.

The ones l

labeled " central" and " optimistic" are analogously inter-preted.

Like CLWG and CPNG, we do not propose weighting factors l

I or distributions for these estimates, nor purport that one is i

DRAFT. INFORMAL AND PREUMINARY AND AS

SU2, 1

M E.4R33,07 YET CORRECTED. FOR W

.!!VATE DISTRIBUIl0N AND NOT

.~

F0d i.XTah dLEASE WITHOUT 00NENT OF

/

M

... _ _ _. 2..-.... _ _...... _...._ __ _ _._ _ __ _ _ _ _ _. _ _,

~ - -

u.; _

_m..

I l

better than another.

Rather we present them as a reflection of the information that is available.

l The general method is depicted schematically in Figure 2.1.

It is worth noting again that we have calculated the frequencies We of accident pathways that can lead to distinct source terms.

However.

r have not calculated the magnitudes of the source terms.

j one can often qualitatively judge that certain pathways are j

similar enough in character to permit them to be binned to the i

i same source term.

2.2 Snacific Examnle Surry Ssg We will illustrate the application of the method by providing the details for a particular accident sequence - Surry This is a sequence initiated by a small-break LOCA with S D.

2subsequent failure of the emergency core cooling system in the injection mode.

Containment sprays are operative as the accident develops toward a core meltdown.

The various questions for S D are listed in Table 2.2 2

together with the answers we assigned.

It should first be noticed that some of the answers are expressed numerically and others are supressed verbally.

This distinguishes the fact that for some questions the available information is sufficient to make quantitative estimates of likelihood, while for others the data supports only qualitative likelihood descriptors.

Ultimately, we will assign numerical values to the qualitative l

descriptors in order to evaluate source term frequencies.

However, we will recognize that this' assignment of numbers is highly subjective and will accordingly evaluate the sensitivity of the results to the numerical choices.

For the present, we need not be concerned about this aspect of the work.

The remainder of this subsection provides the rationale for our assignment of values in Table 2.2:

Question 1:

Likelihood of RCS Break Size in the Larcer Rance type LOCAs were taken to represent In previous PRAs, S2 break areas of 1/2 to 2 inches (RSS, Ref. 17).

More recently, LOCAs have been subdivided into an 52 category (~1 the S2 to 2 inches) and an 53 category (~1/2 to 1 inch), with the letter representing a class of LOCAs initiated by reactor coolant pump seal failures.

ASEP (Ref. 14) estimates the to be approximately lo times that of 52-probability of 83 Hence, the likelihood of the break size being in the larger range is 0.1.

Ouestion 2.

Likelihood of the Break Beina in the Hot Lea Given the RCS Break Size is in the Larcer Rance If the initiating event is a pipe break rather than a DRAFT INFORMAL AND PRELIMINARY AND AS SUCH MAi 1 JAW E ORS NOT YET CORRECTED.

-s FOR IN -

%A'd JGTRIBUTION AND NOT

/

FOR EXT rt.h altiASE MTHOUT CONSENT OF

/

AllTHORS.

i 1

I reactor coolant pump seal failure, the likelihood that the break occurs in the hot leg versus the cold leg is taken to be governed by the respective lengths of piping.

From the FSAR, l

the total length of cold leg and hot leg piping are comparable.

Hence the likelihood of the break being in the. hot leg was taken i

to be 0.5.

Question 3.

Likelihood of Preexistine Containment Leakace or Isolation Failure Sufficient to Preclude Containment Over-nressurization. Given Delayed Failure of Containment ScraYs The containment leakage area required to preclude over-pressurization in the event of a loss of containment cooling is

'1 j

generally figured to be at least 4 square inches (BMI, Ref. 5).

The likelihood of such an opening in subatmospheric containments is extremely small because the leak would be almost immediately detected.

This observation is borne out by precursor data (NRR,

[

i

]

Ref. 28), which shows that even very small preexisting leaks are i

rare for subatmospheric containments.

Generally, preexisting j

leakage is limited by the capacity of the vacuum pumps, i

2 (CPWG, Ref. 4).

corresponding to a leak area of 0.07 in The RSS estimate of containment isolation failure for Surry l

.was 2 x 10-3, which we used for all cases (optimistic / central /

pessimistic).

Ouestion 4.

Likelihood that the RCS Pressure Falls Below the j

Accumulator Discharco Pressure f

We evaluated two subcases for this question:

(a)

RCS break size in the larger range (1-2 inch-diameter), and (b) RCS break size in the smaller range (0.5-1 inch-diameter).

First consider the larger size range.

BMI-2104 calculations for Surry 5 D assuming a 2-inch diameter break size indicated 2

that the primary system pressure declines to about 130 psia at i

the time of core slump.

This value is well below the accumulator l

setpoint of 600 psia.

IDCOR results (Ref. 7) corroborate the i

finding that the accumulators discharge before core slump.

Other calculations performed with the MARCH and RELAP codes i

indicate that the accumulator setpoint will be reached prior to vessel breach for break diameters at least as small as 1-inch, but with a decreasing margin for the smaller break size (SASA, l

Ref 8: BNL, Ref. 29).

Thus, we took the answer for this question to range from "likely" to "almost certain", as shown in Table 2.2.

j Now consider the smaller size range.

Based on an extra-pelation of the results mentioned above, we assessed that it was unlikely for the accumulator setpoint to be reached for a 0.5-inch-diameter LOCA but likely for a 1.0-inch-diameter LOCA.

We assigned answers to this question accordingly.

l DRAFT INFORMAL AND PREllMINARY AND AS i

SUCH MAY CO'iTAIN E R9S NOT YET CORRECTED.

l FOR IN. HOUSE rR!iAsc JJir! dull 0N AND NOT l FOR EXTERNAL RELEASE WITHOUT CONSENT OF l

AUTHORL i

I.-..-.

.,. -,,. -., - - - - - -, - - - -, - - - - - - - - - - - ~ - -

~ ' ' " ' " ' ' ' " ' ' ' ~ ~ " ~ ' '

" ' ' ' ' " ' ' " " ~ ' ' ' ~ ' '

~

~~

1 v

Question 5.

Likelihood of a Steam Spike Just Followina Reactor Vessel Meltthrouch The CLNG and IDCOR program both agreed that a quenching of some of the core debris is likely if water is available.

In this sequence, a large amount of water will exist hoth in the l

reactor cavity and on the containment floor because of the continuous operation of the cont,ainment sprays up to the time of vessel breach.

(If this were not enough, the portion of accumulator water not discharged prior to vessel breach would discharge into the reactor cavity after vessel breach.)

Hence, we took a steam spike to be "likely."

l The primary issue between CLWG and IDCOR regards the size of the steam spike rather than its likelihood of occurrence.

IDCOR calculated a steam spike of 1.0 bar (15 psi) for the S D 2

sequence, assuming that 50% of the core debris quenched..CLWG (Ref. 2) reported that the spike from core debris quenching could be nil or as high as 1.9 bar (27 psi) corresponding to quenching of 100% of the core debris.

To this must be added the pressure increment from primary system blowdown, about 0.5 bar (7 psi) according to BMI-2104.

The CLWG figures were obtained without consideration of containment cooling: however, BMI-2104 i

calculations indicated that the effect of sprays on the steam spike would be small if one assumed, for the pessimistic case, that the debris was highly fragmented.

Thus, we took the size of the steam spike to be 0.5 bar (optimistic), 1.5 bar l

i (central), and 2.4 bar (pessimistic).

Question 6.

Likelihood of a Global Hydrocen Burn Prior to or f

l Just Followina Reactor Vessel Meltthrouch For S D, calculations in BMI-2104 indicate that the 2

hydrogen concentration in containment is sufficient for a global burn to occur any time after core slumping into the lower plenum of the reactor vessel.

If a burn does not occur prior to reactor vessel meltthrough, many experts (CLNG, BMI-2104, SASA) l consider a global hydrogen burn to be a likely occurrence when l

the core debris is first discharged from the reactor vessel.

The ignition source is the hot core debris itself.

Others (IDCOR) contest this supposition on the basis that the i

interaction would produce such large amounts of steam as to i

inert the atmosphere locally.

We therefore took the likelihood l

l of a global hydrogen burn prior to or just following the vessel breach to be "unlikely" for the optimistic case and "likely" for the pessimistic case.

For the central estimate, we took the likeliacod to be " indeterminate".

l The size of the burn is also an issue.

For a global burn occurring during the time frame of interest, the amount of hydrogen participating in the burn is limited by that which can be produced in-vessel (i.e., during the core heatup and slumping portions of the accident).

CLNG and SASA calculations for a different Westinghouse reactor (Sequoyah) ranged in in-vessel DRAFT. INFORMAL AND PREllMliiARY AND AS SUCH Mb ;..: AN E MS h0T YET CORRLCIED. FOR INeTJSE %

1STRIBUT10N AND NOT FOR EXTERNAL RELEASE WITHOUT CONSENT OF f

nilTHORS.

hydrogen generation from 35% to 100% Zircaloy oxidation (Ref. 2 A separate CLWG submittal (Ref. 3) used a lower bound and 9).

of 25% and upper bound of 100% for large, dry and subatmospheric PWR containments.

BMI-2104 calculations for Surry S D showed 2

about 50% oxidation occurring during core heatup and another 10%

to 40% occurring during core slumping, amounting to a total of 60% to 90%.

The SARRP Phenomena Assessment Task Force (Ref. 16) set the total lower and upper bounds at 10% and 100%,

respectively, for a TMLB' accident in Surry.

IDCOR calculated about 25% Zircaloy oxidation in-vessel for Surry S D (Ref.

2 7).

Given this variety of possible choices, we selected 25%

Zircaloy oxidation (400 lb. hydrogen) as our optimistic estimate, 50% (800 lb.) as our central estimate, and 100% (1600 lb.) as our pessimistic estimate.

According to the aforementioned CLWG submittal (Ref. 3), the containment pressure increments corresponding to these amounts of Zircaloy oxidation would be 20 psi (optimistic), 38 psi (central), and 71 psi (pessimistic).* The same submittal showed that the effect of sprays on the pressure increment would be small if the burn time were equal to that which occurred during the TMI-2 accident (i.e., about 8 seconds).

Question 7.

Likelihood of Containment Structural Failure Just Followina Reactor Vessel Meltthrouch The RSS (Ref. 17) estimated the mean failure pressure of the Surry containment structure to be 85 psig, based on an assess-ment that the most probable failure mechanism was tearing of the liner.

A standard deviation of 15 psi was assigned to this estimate.

More recently, Stone and Webster (Ref. 24) calculated a failure pressure of 119 psig, corresponding to general yielding of the reinforcement.

They gave no estimate of uncertainty; however, an analogous estimate for the Zion containment (Ref.

produced a standard deviation of about 2.5 psi.

This latter 19) standard deviation, accounted for material property uncertain-ties, but not for uncertainties in the modeling of the structural response or for possible structural deviations from design.

For our pessimistic estimate, we used the RSS failure pressure of 85 psig and standard deviation of 15 psi and assumed a normal distribution.

For the optimistic estimate we used the Stone and Webster failure pressure of 119 psig and the Zion standard deviation of 2.5 psi.

For the central estimate, we combined the Stone and Webster failure pressure of 119 psig with the RSS standard deviation of 15 psi.

DRtFT - INFORMAL AND PREllMINARY AND AS SE ! ' ' 7 i.i.u, E R1RS NOT YET CORREC1ED.

FOR IN oeE n.

DGTRIBUTION AND NOT.,

FOR EXTERNAL RELEASE WITHOUT CONSDIT OF AUTH0as.

/

=-.

To obtain the failure likelihoods in Table 2.2, we evaluated the containment pressure just following reactor vessel meltthrough by adding the pressure increments for the steam spike and the hydrogen burn discussed above under Questions 6 and 7 to the pressure existing prior to vessel breach.

Based on BMI-2104 calculations, we took the containment pressure prior to vessel breach to be 5 psig if containment was not leaking, or o psig if it was.

Table 2.3 summarizes the pressure estimates leading to the likelihoods in Table 2.2.

Question 8.

Likelihood of Laroe Inducp*. Containment Leakace Lust Followina Reactor Vessel Meltthrouch. Given No Structural Failure Occurs The Containment Performance Working Group (Ref. 4) developed two types of models to evaluate containment leakage before failure.

One was based on the degradation of penetration seals i

caused by exposure to high temperatures for a sustained period of time.

The other was based on pressure-induced yielding of the penetration stiffeners, valves, or seals.

j The model for Surry was temperature-based, with leakage occurring only for seal temperatures exceeding 350*F.

To test the impact of the model on containment response, the CPNG performed calculations for sequence TMLB' (station blackout) with the MARCH code, modified to include the leakage model.

The results indicated that containment atmospheric temperatures would rise to about 450*F, that significant leakage would occur, and that the leakage would preclude containment overpressuri-zation.

These results for Surry are now believed to be extremely pessimistic for a variety of reasons.

Two of the principal reasons are:

(1) the presence of the outboard penetration valves was ignored, and (2) the Mod 1.1 version of MARCH greatly overestimates the containment temperature.

Regarding the second point, BMI-2104 calculations for the same TMLB' sequence predicted maximum atmospheric temperatures of only 280*F, far below the threshold required to initiate seal de' gradation.

The temperatures were even lower for the S D 2

sequence due to spray operation.

A likely reason for the difference is that MARCH 1.1 uses the subroutine INTER to i

calculate the core-concrete interaction, whereas the BMI-2104 i

calculations utilized CORCON, an improved core-concrete analysis code.

For these reasons, we concluded that there was no basis to asssume that the Surry containment would leak due to temperature-induced failures of the penetration seals.

We did feel, however, that there was a potential for presssure-induced leakage to occur.

To investigate this point, we looked at the CPWG leakage model for Zion, which was based on pressure loadings, and assumed that it applied to Surry.

DRAFT INFORMAL AND PREUMINARY AND AS I

SUCH tw ~4TAIN ERRORS NOT YET CORE:D.

FOR N aJJdE PRIVATE DISTRIBUTION ANL NOT FOR EXTERNAL RELEASE WITHOUT CONSENT O

~

AllTHORS.

Figure 2.2 shows the pressure dependent leakage model developed by the CPNG.

The area of leak which would preclude 2

overpressure failure is very large (20-26 in ) if one does not consider internal heat sinks and condensation on containment walls.

With consideration of heat sinks and condensation, MARCH runs have sometimes shown failure precluded with leaks as small as 4-6 in2 (SASA, Ref. 8: BNL, Ref. 29).

If a minimum of 6 in2 is considered necessary to preclude overpressure failure, the CPWG medium leak model would never preclude failure: the high leak model would a1 ways preclude failure for pressures exceeding 90 psig, and would never preclude failure for pressure less than 90 psig.

i There are numerous uncertainties as to the pressures at l

which leaks develop, and in fact we understand that a present consensus is that the model in Figure 2.2 probably overstates the expected amount of leakage.

For this reason, we used the data in Figure 2.2 semi-qualitatively.

For the optimistic case, we interpreted Figure 2.2 to imply that no leakage could occur given there was no structural failure.

For the central case, our interpretation was that no leak could develop that was large enough to preclude later overpressurization, but a 2

lower-capacity leak (<1 inch ) was 50% probable.

For the pessimistic case, we took the pressure at which large leaks can develop to be 90 psig, from Figure 2.2, with a standard deviation of 15 psi from the RSS.

We calculated leakage likelihoods for each of the cases considered in Table 2.3, and i

then reduced the calculated likelihoods by 50% to reflect the downside uncertainty.

We also took Figure 2.2 to imply that a 2

~

lower-capacity leak (<1 inch ) was 100% probable for the pessimistic case.

Question 9.

Likelihood of Containment Soray Failure within 30 Minutes After Vessel Breach This question asks whether the containment sprays operate long enough to remove most of the airborne fission products I

released from the fuel during the melt phase.

The RSS considered containment spray failure to be inevitable after containment rupture due to pump cavitation.

No other cause of

~

l' spray failure was considered.

However, it is believed that sprays might also fail because of debris in the sump clogging the screens and causing cavitation or passing through the screens and damaging the pumps.

One of the sourcas of debris might be an energetic fuel-water interaction that sweeps core debris, tubing, ductwork, and insulation out of the cavity.

Even without a steam spike, some debris could be expected.

It may be observed, for example, that the sump water at TMI-2 was laden with particulate matter.

DRAFT INFORMAL AND PREllMINARY AND AS SU;" w l i iTAN ERRMS NOT YET CORNcCkD.

. FORiN a GivA;E UGTRIBUTION AND N0T FOR EXTdNAL RELEASE WITHOUT CONSENT OF AUTHORS.

.-=r==--===

==

i i

It is very difficult to obtain specific information about I

the operability of pumps under conditions that exceed their j

design basis.

The manufacturers, of course, do not guarantee l

the success or failure of their pumps if operated beyond j

specifications.

The prevailing opinion regarding centrifugal i

i pumps, however, is that they are capable of operating in i

cavitating environments for hours before failing due to i

mechanical damage.

This opinion is supported by tests conducted at Sandia under Task Action Plan A-43 (Ref. 30).

Based on these observations, the likelihood of containment

)

spray failure is considered somewhere between "unlikely" and

" remotely possible" if there is no steam spike and no containment failure.

If containment fails, the optimistic judgment is that spray failure is "unlikely" (consistent with an IDCOR observation, Ref. 7), whereas the pessimistic judgment is that spray failure is "likely" (consistent with the RSS).

The likelihood descriptors in Table 2.2 reflect these judgments.

Question 10.

Likelihood of Containment Soray Failure Given turvival within the First 30 Minutes After Vessel Breach.

j' This question asks whether the containment sprays operate long enough to remove most of the fission products released from the fuel during the vaporization (core-concrete) phase.

If the sprays do not fail in the first 30 minutes following vessel breach, it is considered possible that earlier pump damage could cause failure during continued operation.

The descriptors in Table 2.2 represent our judgments consistent with the observations made under Question 9.

Question 11.

Likelihood of Core-Concrete Interaction Producina g Vanorization Release.

d j

The vaporization release may be precluded by either of two j

occurrences.

First, if sprays continue to operate throughout l

the accident, there is a possibility that a permanently coolable debris bed will form in the reactor cavity.

Second, if there is a strong fuel-water interaction (steam spike), the i

core debris may be scattered so sparsely through containment i

l that significant core-concrete interaction is precluded.

l l

Conversely, if there are no sprays and no steam spike, the core I

l debris will mostly remain in the reactor cavity, the debris bed will dry out, and a core-concrete interaction will occur with l

high certainty.

For the case where the sprays continue to operate, the IDCOM analysis (Ref. 7) takes the outcome of the accident to be a permanently coolable debris bed.

The BMI-2104 analysis, to I

the contrary, allows the core-concrete interaction to occur i

with the water in the reactor cavity boiling off faster t.han it can reinfiltrate.

Thus, we took th occurrence of the j

vaporization release to range from "unlikely" (consistent with I

DRAFT INFORMAL AND PREUMINARY AND AS SUCEU' C VAIN ERR 3S NOT YET CORRECIED. FOR IN L 4 WE XiTRl5UTION AND NOT FOR EXTERN 4. REl. EASE WITHOUT CONSENT OF f

AUTHORL

i the IDCOR analysis) to "likely" (consistent with the BMI-2104 i

l analysis) if sprays are operating.

Ouestion 12.

Likelihood of a Global Hydrocen Burn Durina or Lgst Fo11ovino the Vanorization Release.

Because ignition sources are available during this sequence and the atmosphere is not steam-inerted, we took a late hydrogen burn to be a "likely" occurrence provided there is a sufficient amount of hydrogen to support the propagation of the burn.

The maximum quantity of hydrogen or other flammable gases 3

l could exceed 100% of that due to zirconium oxidation, if-hydrogen due to steel oxidation and core-concrete interaction is included.

However, the oxygen available could only burn about 150% zirconium equivalent.

The amount available depends on whether core-concrete interaction releases flammables and j

whether a prior burn has occurred.

1 l

We estimated that a late hydrogen burn could involve as such as 800 lb of hydrogen (optimistic), 1200 lb (central), or "1600 lb (pessimistic) if core-concrete interactions occurred.

These figures correspond to hydrogen ignition thresholds of about 6%, 9%, and 12%, respectively, if sprays are operating.

These are consistent with the range of ignition thresholds l

considered in previous analyses (CLWG, Ref. 2: SASA, Ref. 9).

Question 13.

Likelih'ood of Containment Structural Failure from a Global Hydrocen Burn Durina or Just Followina the l

Vanorization Release.

We calculated the likelihood of containment structural failure from a late hydrogen burn as described under Question 7.

Table 2.4 summarizes the pressure estimates leading to the likelihoods in Table 2.2.

We obtained containment pressures i

just prior to the burn from BMI-21*04 assuming that the burn occurred just after the peak of the vaporization release.

t Question 14.

Likelihood of Late Containment Soray Failure.

Given Containment Failure Occurs i

As described under Question 9, we based our optimistic and

[

pessimistic descriptors for the likelihood of spray failure given containment structural failure on information from IDCOR and the RSS.

DRAFT. INFORMAL AND PREUMINARY AND AS

~" t !DRS NOT YET CORRECTED.

ST,

- FOR.

E DISTRIBuil0N AND NOT

./

F0d EM.4M. siLEASE WITHOUT CONSENT OF AUTHORS.

-- : ~: : ~:

Question 15.

Likelihood of Containnent Leakace from a Global Hydroaen Burn Durina or Just Followana the Vaporization Release Given No Structural Failnre Occurs we calculated the likelihood of containment leakage from a late hydrogen burn as described under Question 8, using containment pressure loadings from Table 2.4.

Question 16.

Likelihood of Basemat Meltthrouch. Given the Occurrence of a Core-Concrete Interaction If the core-concrete interaction occurs (viz., Question 11), there is some uncertainty as to whether the core debris I

will penetrate completely through the basemat.

The models in.

core-concrete interaction codes such as CORCON are not considered to be as valid when the core debris freezes and

{

starts to attack the concrete as a heat-producing solid.

We thus have to rely on the limited experimental evidence that exists.

Experiments at Sandia appear to indicate that considerable erosion of concrete continues to occur after the melt solidifies.

If water is supplied to a core debris layer which is already attacking concrete, the penetration continues but the debris layer cools down more quickly.

Based on these observations, we took the occurrence of meltthrough to be "likely" (optimistic) if the sprays have failed.

If the sprays continue to operate, we took the occurrence of meltthrough to be less assured, as shown in Table 2.2.

Question 17.

Likelihood of Late Containment Overoressurization Late overpressure failure is considered to be " impossible" if containment sprays continue to operate, consistent with all analyses performed te date (BMI-2104, SASA, IDCOR).

It is also considered to be " impossible" if there has been an isolation 2

failure or a pressure-induced leak exceeding about 4 inch,

Late overpressure failure is estimated to be "certain" if there is no core-concrete interaction and no sprays, because all the core decay energy is transmitted to the containment atmosphere.

j If there is a core-concrete interaction but no meltthrough, and sprays have failed, some of the energy could be transmitted through the basemat to the substrate underlying containment.

Overpressure failure is therefore estimated to be "likely" f

(optimistic), "almost certain" (central), or "certain" (pessimistic) for this case.

If the core debris melts through the basemat, late over-pressure failure is not necessarily precluded, because pressure relief through the ground might be too slow to prevent over-pressurization.

Calculations in BMI-2104 presume that basemat seltthrough leads to a depressurization of containment as a result of venting of the gases through the ground; however, the DRAFT - INFORMAL AND PRE.lMINARY AND AS ST" W CT4TAIN ERRORS NOT YET CORRECED.

6-FOR 4

c. 'Rl/ ATE DISTRIBUTION AND NOT

/

F0a JTcel. RELEASE WITHOUT CONSENT OF

/

AUTHORS,

.. =....

m authors of that document recognized this te be an area of high uncertainty.

We therefore took the occurr.nce of late contain-ment overpressurization to range from "unlikely", consistent with the above discussion, to "likely", consiFtent with the BM1-2104 analysis, if basemat seltthrough were to occur.

Question 18.

Likelihood that the Release Bynasses the Auxiliary Buildina. Given Containment Leakace or Structural E111E1 Most containment penetrations lead into the auxiliary building.

An exception is the equipment hatch: however, CPWG results seem to indicate that pressure-induced leakage through

,)

I the equipment hatch is not the dominant pathway for leakages l

that are large enough to preclude containment overpressuri-l zation.

We therefore estimated that it was "unlikely" that most of the leakage would bypass the auxiliary building.

On the other hand, the auxiliary building subtends only a small portion of the containment structural surface area, and in particular, does not subtend the upper springline.

We therefore estimated that it was "likely" that a structural failure of containment would result in bypass of the auxiliary building.

For the pessimistic case, we took bypass to be "almost certain".

2.3 Treatment of Verbal Descrintors Interpretation of words such as "likely", " indeterminate",

"unlikely", or "almost impossible" is subjective.

In cases

{

where we have used these words, we did so because there was no clearcut way to quantify the likelihoods of the questions being asked.

Still, some assignment of numerical values is necessary if the frequencies of the outcomes are to be estimated.

Table 2.5 shows 4 plausible assignments of values for the l

verbal descriptors we have used.

In most cases, we used Alternative 1 to quantify the outcome frequencies; however, we also investigated the sensitivity of some of the results to the choice of quantification alternatives.

The results of the sensitivity study are described in Section 3.5.

2.4 Treatment of Other Secuences and Other Plants The questions asked on the containment event tree and the utilization of information to quantify them vary from sequence to sequence and from plant to plant.

Below we shall provide a brief description of some of the important differences.

DRAFT. INFORMAL AND PRELIMINARY AND AS Si

' ' 1 'TAIN ETR'1RS NOT YET CORECKD..

AiE Da1Rl30T10N ANu NOT F0 M..tts nELEASE WITHOUT CONSENT OF

/

AUTHORS.

1

/

i I

Sequences evaluated for Surry in addition to S D are 2

TMLB' (station blackout) and AB (large LOCA with station blackout).

We did not perform a containment event tree analysis for sequence V (interfacing systems LOCA) because it is a sequence defined by a unique containment fai)'te, mode (i.e., the bypass of containment).

TMLB' and AB are somewhat easier to analyze than S D 2

because the containment sprays do not operate thence the question of delayed spray failure is moot).

For this first iteration, we assumed that power was not recovered prior to containment failure: the second iteration will include the possibility of power recovery after core melting and bsfore containment failure.

For TMLB', a key question is whether the primary system fails from high temperatures before the core melts through the reactor vessel, and if so, where the failure occurs.

Possible locations are the reactor coolant pump seals (cold leg), the steam generation tubes, or the reactor vessel nozzle welds (hot leg).

Temperature-induced failure could cause the primary system to depressurize prior to vessel breach, allowing the accumulators to discharge while the core is in the vessel, and reducing the size of the steam spike following vessel breach.

(It would also preclude the occurrence of direct atmospheric heating resulting from high pressure ejection of core debris from the vessel however direct heating was not analyzed in this iteration.)

We took the temperature-induced LOCA for TMLB' to range from "unlikely" for the optimistic estimate (consistent with previous PRAs and with BMI-2104 analyses) to-

"likely" for the pessimistic estimate (consiste'nt with the majority opinion of the CLWG).

For the central estimate, we took the likelihood of the induced LOCA to be " indeterminate".

Sequences evaluated for Zion are S D and TMLB'.

There 2

are several differences between the Zion and Surry analyses.

The Zion containment has fan coolers; hence, one must ask about survivability of the fan coolers (for S D) as well as 2

survivability of the sprays.

The containment is atmospheric rather than subatmospheric; hence the likelihood of preexisting leakage is somewhat higher.

The containment failure pressure l

is higher than at Surry; hence induced leakage becomes relatively more important as a containment failure mode.

The BWR plants evaluated in this study were Peach Bottom (Mark I containment) and Grand Gulf (Mark III containment).

The sequences analyzed for Peach Bottom were TW (transient event with loss of containment cooling), TC (transient event r

with failure to scram), and AE (large LOCA with failure of emergency core cooling).

Sequences analyzed for Grand. Gulf were TC, TPI (transient event with stuck-open safety / relief valve and loss of suppression pool cooling), and TQUV (transient event with loss of feedwater and emergency core U*

DRAFT INFORMAL AND PRELIMINARY SUCH w CT4TAIN ERRORS NOT YET C FOR IN. ORE al AE DISTRIBUTION AND FOR EXTERNAL RELEASE WITHOUT CO,

Alml0BL

/

m+

l Many of the questions we posed for Peach Bottom and Grand Gulf were quite different from those we posed for Surry and Zion.

This is to be expected, since the BWR containment designs are very different from the PWR designs.

Below is a list of some of the questions that are specific to Peach Botton, together with some of the observations we used for quantifying the likelihoods:

(1)

Will containment fail before the core melts?

For the TW and TC sequences, it is usually assumed that containment fails before the emergency core cooling system fails (RSS, SASA, IDCOR).

The likely causes of ECCS failure, given con-tainment failure, are cavitation of the pumps or deformation of the cooling lines.

There is some likelihood, however, that the emergency core cooling pumps will fail before the containment fails.

Possible causes of early ECCS failure are insufficient cooling of the lube oil or underventilation of the pump room.

For the AE sequence, containment could fail as a result of overpressurization from steam and hydrogen after the core has become severely degraded but not yet completely molten.

This could occur if the amount of hydrogen produced in-vessel exceeds about 70% of the Zircaloy equivalent.

The possibility of early containment failure caused by hydrogen generation during AE is treated in BMI-2104.

(2)

Will the nrimary system still be pressurized at vessel breach?

This question applies only to the TW and TC sequences, since primary' system depressurization is guaranteed for AE.

In most cases, the automatic depressurization system (ADS) would actuate automatically during these sequences to reduce the primary system pressure before the core melts.

(For TW, operation of the steam-driven turbine of the high pressure coolant injection system also reduces the primary system pressure.)

However, the ADS would fail to operate auto-natically if the containment rupture were such that the drywell pressure stayed above ~75 psig.

In that case, the i

safety / relief valve pilot air pressure would be insufficient to actuate the ADS, and the primary system would remain pressurized unless the operator took some special actions.

j (3)

Will the containment breach be in the drywell?

The RSS originally predicted that containment failure would occur just above the midplane of the toroidal suppression chamber (i.e., in the wetwell).

A more recent analysis (Ames, Ref. 25) predicted that the failure point would be in the drywell.

In BMI-2104, a drywell failure was assumed, but the authors discussed the possibility that the lo' cation of failure could be different.

For TC, dynamic loads in the suppression pool could increase the likelihood of a wetwell breach.

DRAFT. INFORMAL AND PREl.IMINARY AN SUO4 ' AY CPTAIN ERRORS NOT YET CORR;C dD.

cuj. RisATE OtSTRIBuil0N AND Ndi FOR u FOR EXTatNAl. REl. EASE WITHOUT CONS

/ MITH0RS.

.f

. -~~~~-.-~~~u.~ 2.~-

~-.- :..~ - :

7-4 (4)

Will containment failure lead to failure of the guncression nool function?

While no one has published an analysis of whether the suppression pool in a Mark I containment would survive an overpressurization failure of the i

containment, many structural experts feel that pool survival would be very questionable (FVCS, Ref. 27).

Because the containment is a free-standing steel shell structure with a high failure pressure, the forces associated with the failure could be violent.

Failure of the suppression pool would be a moot question if the containment failed in the drywell.

(5)

Will leak naths develon in the drywell that could gause the sunoression nool to be bvoassed?

This question primarily concerns the AE sequence.

The containment penetra-tion seals at Peach Bottom are elastomeric, like those at Surry.

The potential for overheating of these seals is greater at Peach Bottom than at Surry, however, because the small size of the drywell compartment makes it more susceptible to thermal loading from the core-concrete interaction (CLWG, SASA).

If the seals survive the thermal loading, there is a chance that a leak path could develop as a result of a direct core debris attack on the drywell structure, causing failure of the shall at a location where it is not directly backed by concrete (CLWG).

(6)

Will secondary containment be bynassed?

The secondary containment could be bypassed if the primary containment failed in the wetwell at a location where there is a direct pathway to the outside environment.

The RSS considered this possibility.

(7)

Will the standby aas treatment system (SGTS) fail to remove fission nroducts from the secondary containment g_tmosohere?

It is likely that the blowers in the SGTS will operate throughout the accident, but it is also likely that the filters will become ineffective due to one or more of the following occurrences:

(a) steam overloading, (b) particulate overloading, and (c) overheating (BMI-2104, FVCS).

In this first iteration, we did not address the change in procedures which allows for the venting of containment.

Further, for TC, we neglected the possibility that the primary system might be overloaded by the pressure transient occurring just after containment isolation.

The implementation of automatic trip of the reactor coolant pumps should cause the probability of this event to be very low.

The questions we posed for Grand Gulf were similar to Peach Botton, particularly for the sequences TPI and TC.

For these sequences, the primary differences were in the quantifi-cation.

Grand Gulf is not preinerted as is Peach Bottom: hence pre-existing leakage is somewhat more likely to go undetected.

The dryvell is contained within the wetwell: thus, drywell leakage results only in supprGasion pool bypass, not release from containment.

The penetration seals are steel welded:

DR4FT. INFORMAL AND PRELIMINARY AND AS S

S? JAW ESR9RS NOT YET CORRECMD.

. F.'

!.. ; TIM Dill?WJi!0N AN. N)I fl.

af..h nELEASE WiiHOUT CONSENT OF AJM3RS.

[*

e hence temperature-induced leakage is relatively less credible (excepting the possibility of diffusion flames, see below).

The containment is a fairly low-pressure concrete structure whose likely point of failure is at the upper springline: hence the suppression pool function is much more likely to survive the failure of containment.

The reactor vessel gedestal supports the vessel at the nozzles; hence an attack on the pedestal by core debris could lead to destruction of the vessel and bypass of the suppression pool (CLWG).

For the Grand Gulf TQUV sequence a number of questions were posed to address issues associated with hydrogen burning.

Burning above the suppression pool could cause the following_

significant events to occur:

(1)

The pressure increase in the wetwell could cause water to flow over the weir wall onto the drywell floor and into the pedestal area, thus increasing the likelihood of a steam spike.

(2)

The high temperatures produced by diffusion flames could induce leakage through drywell or containment penetrations (CLWG).

(3)

The occurrence of a steam spike that rapidly forces hydrogen into the wetwell could lead to a global deflagration or local detonation that could threaten the containment structure.

Finally, hydrogen burning in the drywell during the core-concrete attack could lead to overheating of the drywell.

For this first iteration, our quantification of the BWR ctentainment event trees was based far more heavily on the use of verbal descriptors than for the PWRs.

We pursued this approach partly because the information base for the BWRs was less complete than for the PWRs, and partly because we had insufficient time to do otherwise.

We plan to provide a more quantitative assessment of the BWR event likelihoods during the second iteration.

DRAFT INFORMAL AND PRELIMINARY AND AS ST ' "W C1 ;i Al1 ERRORS NOT YET CORR; CIED.

FT

1) E RiiATE DISTRIBUTION ANs NOT F i d m dELEASE WITHOUT CONSENT OF hHdRS.

)

7

' ' T 2

l 1_________,_-______-_-____.._.--

J

I 3.

RESULTS 3.1 Results for Surry 3.1.1 Secuence AB Table 3.1 summarizes the optimistic, central, and pessimistic containment failure modes for sequence AB at Surry.

In all three quantifications significant fractions of the containment failures are attributed to late overpressurization due to the' accumulation i

of gases from core-concrete interactions.

In the pessimistic quantification, leakage in excess of design leakage but insufficient to preclude late overpressurization is postulated to be induced before the late overpressurization.

Basemat melt-through is another containment failure mode which is significant in all three quantifications: 79% of the optimistic containment failures, 49% of the central containment failures, and 5% of the pessimistic containment failures.

Induced leakage in excess of design leakage is postulated in half of the central and all of the pessimistic basemat meltthroughs.

BMI-2104 analyzed basemat seltthrough for sequence AB at Surry: however, pressure-temperature-induced leakage in excess of design leakage was not considered.

Possimistically, a significant fraction (38%) of containment failures is attributed to late hydrogen burns.

Such burns are i

precluded early in the accident due to high steam concentrations; however, eventually such steam inerting could be negated'due to condensation on passive heat sinks and aerosols thereby permitting the combustion of accumulated hydrogen and carbon monoxide.

This possibility is consistent with recent, unpublished MARCH and CONTAIN analyses performed at Sandia National Laboratories.

Containment failure due to a late hydrogen burn is one of the containment failure modes analyzed in BMI-2104.

l The pessimistic quantification also indicates a significant fraction (15%) of containment failures due to induced leakage sufficient to preclude gradual overpressurization.

Such leakages are postulated early enough to result in higher releases than j

would be obtained from late overpressurization.

3.1.2 Secuences S,D and S R 3

Table 3.2 summarizes the optimistic, central, and pessimistic containment failure modes for the sequences S D and S D at 2

3 Surry.

I Optimistically, the most likely outcome (95%) of S23D core melt is no containment failure with core-concrete attack being l

prevented or arreste.d before meltthrough or gradual overpressur-ization can occur.

Only a 5% chance of basemat seltthrough and a 0.1% chance of late, gradual overpressurization result from the optimistic quantification.

Basemat a g ht g sgogggy g

SUCPW GTAIN ERRORS NOT )ET C0W10.

i

. FOR.N.

O DE.'RivATE DdTitiSull0. A.. iUT l

f F0i dTbb. REl. EASE WITHOUT CONSENT OF

..r MTHORS.

e..

.. a.

• * * ~

taan gradual overpressure because of continued containment heat removal by the containment sprays.

Containment failures due to hydrogen burning do not occur in the optimistic quantification because of optimistic assumptions regarding hydrogen production, core-concrete termination, and ignition thresholds.

The principle containment pathways in the central quantifi-cation are the same as for the optimistic quantification:

however, in the central quantification, the likelihood of arresting core-concrete interactions before basemat meltthrough is deemed "indeterminant" resulting in a nearly equal split between basemat meltthrough and no containment failure.

The fraction of containment failures due to gradual overpressuri-zation also increases in the central quantification due to higher -

probabilities of delayed containment spray failure.

Half of these late, gradual overpressures are postulated to be preceeded by induced leakages in the central quantification.

The higher probabilities attached to in-vessel hydrogen production and sustained core-concrete interactions result in a small fraction (0.1%) of the central containment failure being attributed to late hydrogen burns.

Possimistically, containment failure at vessel breach due to a coincident 27 psi steam spike and combustion of hydrogen from 100% in-vessel Zr oxidation accounts for 79% of the containment failures.

Early hydrogen burns contribute 4% and late hydrogen burns contribute 6% of containment failures in the pessimistic quantification.

The chances of induced leakage, either due to the containment loadings following vessel breach or later also appear in the pessimistic quantification.

4 3.1.3 Sequence TMLB' Table 3.3 summarizes the optimistic, central, and pessimistic containment failure modes for the TMLB' sequence at Surry.

In all three quantifications, there is a significant fraction of containment failures attributed to basemat meltthrough.

Basemat neltthrough was one of the containment failure modes analyzed in BMI-2104.

In the pessimistic quantification, the largest fraction of containment failures is attributed to late hydrogen burns, postulating that inerting by high steam concentrations would be negated due to condensation on passive heat sinks and aerosols after the buildup of significant hydrogen (and possibly carbon monoxide) concentrations.

This result is consistent with recent J

unpublished results of a MARCH sensitivity study being performed at Sandia National Laboratories.

The pessimistic quantification also indicates significant fractions of containment failure due to late, gradual over-i pressurization and late, pressure-temperature-induced leakage.

These two containment failure modes are mutually exclusive in DRAFT INFORMAL AND PREUMINARY AND AS SJF V.Av GTAIN ERRORS NOT YET C' A D. FOR N

; E.)Rl/ ATE DGTRl3Ull0.. A, NOT F04 XTA u. RELEASE WITHOUT C0aSENT Of f.'

MHORS.

_.T ~ d..

~'

~

'l 4

that the extent of the induced leakage must be sufficient to preclude late, gradual overpressurization.

However, in all of the pessimistic gradual overpressures (and in 50% of the central) i we postulate that high containment loadings would result in l

leakage in excess of the design leakage postulated in BMI-2104.

Since induced leakage would likely occur long before gradual overpressure, the releases for either the late-leakage or late-overpressure containment failure modes would depend on the magnitude of the induced leakage.

The central quantification is similar to the pessimistic quantification except the likelihood of late leakage is greatly reduced.

In the optimistic quantification, no containment failures are attributed to late hydrogen burns.

This results from optimistic

• assumptions regarding the extent of combustible gas production, the ignition threshold, and combustion completeness.

3.2 Results for Zion 9

3.2.1 Secuence S D 7

Table 3.4 summarizes the optimistic, central, and pessimistic containment failure modes for the sequences S 2D and S D at 2

3 Zion.

The optimistic and central estimates for Zion 5 D are very 2

similar to Surry S D, Section 3.1.2.

The higher probability of 2

preexisting leakage for Zion results from the fact that Zion is an atmospheric containment whereas Surry is subatmospheric.

Possimistically, containment failure at vessel breach due to a coincident 27 psi steam spike and combustion of hydrogen from 100% in-vessel Zr oxidation accounts for 27% of the containment failures.

The threat from hydrogen burning is generally lower at Zion than at Surry because of the higher containment failure pressure.

The chances of induced leakage, either due to the containment loadings following vessel breach or later also appear in the pessimistic quantification.

Possimistically, all late, i

gradual overpressurizations are assumed to be proceeded by 1eakage in excess of design leakage due to high pressure-I temperature loadings.

3.2.2 Sequence TMLB' l

Table 3.5 summarizes the optimistic, central, and pessimistic l

l containment failure modes for the TMLB' sequence at Zion.

The results for Zion TMLB' are similar to Surry TMLB', Section 3.1.3, except for the estimates for late hydrogen burning.

As opposed to Surry, late hydrogen burns were found not to be a threat for the Zion containment because the atmospheric conditions for flammability were not attained.

It was judged that late burns i

large enough to threaten containment could occur only if contain-ment cooling were restored, but as mentioned in Section 3.4, DRW. INFORMAL AND PREUMINARv mD AS SJ

' D dAl1E3RJRSt.0i'.E7

.. ED.

FOR

. UE RI AiE Dalkl30Wr 4 NOT F0..J 6.ht. HELEASE WlIHOUT C0dni 0F f

AUTHORS.

=

e ees en

a.. e 9 e +4emp < *- 9 m W '

~

l restoration of power after core degradation was not considered as

~

a pathway in this iteration.

3.3 Results for Peach Botton 3.3.1 Secuence AE The conditional containment failure mode probabilities for the AE accident sequence are summarized in Table 3.6 for the optimistic, central, and pessimistic cases.

In the pessimistic case, containment failure probabilities are divided approximately equally between early overpressuri-

~

zation and late overpressurization.

Early overpressurization occurs either before or at vessel breach as a result of a buildup of steam and hydrogen in containment.

As mentioned in Section 2.4, about 70% cledding oxidation must be assumed to attain sufficient amounts of hydrogen to threaten containment.

(Of course, the hydrogen does not burn because the containment is inerted.)

If early failure does not occur, containment eventually overpressurizes from the noncondensibles produced by concrete ablation in the drywell.

In the pessimistic case, structural failure is estimated to be a more likely outcome than leakage because the outboard containment penetrations are assumed to be sufficiently protected from overheating.

The failure most often occurs in the drywell, which causes the suppression pool to be bypassed.

(Structural failures in the wetwell also lead to pool bypass in the pessimistic case.)

In the optimistic case, the majority of releases are associated with temperature-induced leakage caused by overheating of the penetration seals during the core-concrete interaction.

l Since the leakages occur in the drywell, the suppression pool is bypassed.

In the central case, most of the containment failures are late but a significant fraction (~104) are early.

The late containment failures are about equally divided between structural failure and temperature-induced leakage.

3.3.2 Secuence TC Table 3.7 summarizes the optimistic, central and pessimistic containment failure mode fractions for TC sequences leading to core melt.

The most likely containment failure scenario for the 1

t TC sequence is one in which the reactor stays at elevated power (20 to 30 percent) leading to rapid heatup of the suppression pool, steam break-through and buildup in containment, containment failure followed by suppression pool boiling (and/or draining),

loss of reactor coolant makeup, and core melt.

Possimistically, all containment failures are taken to result in this manner (i.e.

before core melt) and are assumed to occur in the drywell as DRAFT. INFORMAL AND PREllMINARY AND AS SJ'

:. TAM ETRS NOT YET C0:

ED.

F02 i i M. AiE Du1RIEUIl0a AA OT f F0

.i_mht dELEASE WITHOUT CONSENT OF f

AUTHORS, r L : 7

~ -

J indicated by.the structural analyses performed at Ames Laboratory (1).

In the central quantification, the possibility of early containment failure in the wetwell is permitted but deemed unlikely, and, if failure occurs in the wetwell, draining of the suppression pool is considered likely, so that only a 1% chance of containment effluent passing through suppression pool water results in the central quantification.

Further, fraction of this i

1% is assigned to account for the possibility of a small bypass of the suppression pool (for example backloakage through vacuum breakers or temperature induced drywell leakage).

Optimistically, the probability of wetwell failure is taken to be equal to the probability of drywell failure, and the probability of retaining water in a failed suppression pool is taken to be 0.5.

This increases the fraction of early containment failures in which effluent would pass through water in the suppression pool to 234.

Of course the high temperature of the suppression pool water would reduce the effectiveness of fission product scrubbing.

In the optimistic quantification, we also permitted (as "unlikely") the possibility that ECC injection would fail early, leading to core melt with containment intact (effectively an accelerated TQUV sequence).

The possibility of l

vessel breach resulting from the initial pressure spike was not considered in our quantification although this could conceivably result if the recirculation pumps failed to trip.

)

The detailed event tree used to quantify containment failure modes for the BWR sequences included secondary containment effects.

However, for the TC sequences involving containment failure before core melt, we judged that if the secondary con-tainment was not failed or bypassed, the secondary containment blowout panels would relieve and the standby gas treatment system would not be extremely effective in removing fission products.

I 3.3.3 Sequence TW j

The conditional probabilities for containment failure modes for TW are summarized in Table 3.8 for optimistic, central, and l

pessimistic sets of assumptions.

The results for TW are similar to TC, Section 3.3.2, in that the most significant pathway is pre-core melt overpressurization with an unscrubbed release.

However, the likelihood of the accident degenerating into a TQUV was judged more likely for TW than for TC, as was the likelihood of early induced leakage.

These differences reflect the fact that the TW accident develops l

auch more slowly than TC, with high temperatures and pressures i

persisting for a much longer period of time before containment fails.

D3W dFORMAL AND PREUMinARY AND AS SW: 1 0) W : E 3'm01)ET CORRICIED. FOR M TJ:.E 3..n.E udlill3Ull0N AND NOT

/

FOR EXTcdM. dELEASE WITHOUT CONSENT OF

/

AUTHORS.

L h "r!Y r r r*"

r:-::LL

~

1 3.4 Results for Grand Gulf 3.4.1 Sequence TC Table 3.9 summarizes the optimistic, central, and pessimistic containment failure mode fractions for TC core-melt accidents at Grand Gulf.

The most likely containment failure mode in all three quantifications is failure before core melt due to overpressure following suppression pool overheating.

This i

containment failure mode is the one analyzed in BMI-2104 for the Grand Gulf TC sequence.

The suppression pool survives but is at saturation temperature.

Hence, subsequent fission product releases are scrubbed, but the efficiency of scrubbing is lower than if the pool were subcooled.

In the optimistic quantification the possibility that ECC is lost early is recognized although considered "unlikely."

Early loss of ECC renders the core subcritical and essentially converts the TC sequence into an accelerated TQUV sequence (see Section 3.4.3).

The result, in the optimistic quantification, is a significant fraction (9%) for late containment failure and a small fraction (1%) of no containment failure.

Possimistically, the possibility of some bypass of the suppression pool (due i

primarily to induced leakage) is recognized although deemed "unlikely."

{

3.4.2 Secuence TPI Table 3.10 summarizes the conditional containment failure mode probabilities for TPI based on optimistic, central., and pessimistic sets of assumptions.

The results for TPI are similar to TC, Section 3.4.2.

However, the likelihood of the accident degenerating into a i

TQUV-type accident with a depressurized primary system was taken to be higher for TPI, as was the likelihood of early induced leakage.

As mentioned in Section 3.3.,

these differences reflect the much longer period of time leading to the buildup of temperature and pressure in containment.

I l

3.4.3 Secuence TOUV Table 3.11 summarizes the optimistic, central, and pessimistic containment failure mode fractions for TQUV core-melt accidents.

i In all three quantifications, there is a significant fraction of I

containment failures which occur " late" -- more than one-half l

bour after vessel breach.

These late containment failures are due primarily to the accumulation of non-condensible gases from l

concrete ablation with a smaller contribution from late combustion events.

In the optimistic and central quantifi-cations, such late containment failures are most likely.

j Possimistically, the fraction of early containment failures due DRAFT. INFORMAL AND PRttlMINA 1

SU ' W C01TAl.1E;RUS OT W COCTED.

F0R 10 OUSE. RI.AIE INmidUliSa A FOR EXTERNAL RELEASE WITHO g l

/

~

u

_-__-__--,,---___._--__t___-.---..

to hydrogen burning (especially hydrogen burning at vessel breach) approaches the fraction of late containment failures.

Possimistically, drywell leakage which would result in a small bypass of the suppression pool is considered unlikely early but indeterminant late.

This results in a relatively significant pessimistic fraction of containment failures which are accompanied by small bypass of the suppression pool.

In the optimistic and central quantifications, the possibility of leakage sufficient to prevent overpressurization of containment is recognized, albeit unlikely; so that, some containment failures occur by early and late leakages.

Finally, in the optimistic case, there is a significant fraction (10%) attributed to no containment failure.

The containment would not fail if hydrogen releases were small and the concrete ablation was limited due to spreading of the debris within the drywell.

\\

l DRAFT INFORMAL AND PREUM!w AND Al S'E ' ".'.' CTiT AM E.'n "' '

MED.

FOR.., 00iE. RI. Aid u.a.a.,ui.v.. n... NOT F0d EXicRNAL RELEASE WITHOUT CONSENT of AUTHORS.

f i

3.5 sensitivities As mentioned in Section 2.3, the numerical values assigned to verbal descriptors such as "unlikely" or " remotely possible" are somewhat arbitrary, and the results could be sensitive to i

these choices.

Accordingly, we performed a sensitivity study for Surry S D using the four alternative numerical sets in 2

Table 2.5.

The results, depicted in Table 3.12, indicate that the variation of conditional probability within each class (optimistic, central, pessimistic) is small compared to the difference in results between classes, i

Although the results are not very sensitive to the choice i

of numerical values, they are sensitive to the choice of verbal descriptors, i.e., whether phenomena are considered "likely" or "unlikely" to occur.

This sensitivity has been covered by the i

choice of optimistic, central, and pessimistic walk-throughs.

I i

i l

I i

DRAFT. INFORMAL. AND PREllMINARY At'D A Sr

.A'I C'.; ; TAI.l ERR 3R5 NOT YET C0.m.u :.D.

, FbR d. O' 3E PRI.A1E 0131xl8Ull0N AND NOT J

FO.t EXTERNAL RELEASE WITHOUT CONSEN AUTHORS.

-. ~. - ~.

i 4.

OVERALL

SUMMARY

AND RECOIGEENDATION (We have not yet prepared an overall summary of our results nor completed our consideration of recommendations for additional source term calculations.

Based on our work to date, however.

we are able to make some preliminary observations on the latter subject.

The following paragraphs provide these observations.)

l 4.1 surry and Zion The principal containment pathways identified in this appendix for Surry AB, S23D, and TMLB' are, for the most part, treated in existing BMI source term calculations.

In many cases, however, it would be necessary to combine or extrapolate the results of existing BMI calculations to achieve i,

complete coverage.

For example, a calculation exists for the Surry AB sequence in which late containment failure occurs due to hydrogen burning, but this possimistically significant containment failure mode is not specifically addressed in the i

BMI calculations for the Surry TMLB' sequence.

By coupling-primary system results from the BMI TMLB' calculation with containment results from the BMI AB-gamma calculation, one i

could achieve a surrogate for the TMLB' -late gamma scenario.

l~

In other cases, existing BMI calculations for rne scenario may i

serve as adequate surrogates for other scenari~s.

For example, j

BMI calculations performed for a hot-leg S D could be used as 2

j a conservative surrogate for a cold-leg S D at the same plant.

2 We have, bewever, identified two areas in which additional calculations may be warranted.

First, in Surry sequences, there is a significant occurrence of late leakages in our j

pessimistic results.

If the final CPNG model for pressure-induced leakage is consistent with the assumptions cited in Section 2 of this report, then additional leakage calculations for Surry would appear warranted.

BMI has already performed i

such leakage calculations for Zion.

Second, in TMLB' accidents l

it has been postulated that the reactor coolant system pressure i

boundary could fail due to high temperatures relatively early I

in the accident (near the beginning of the melt release).

If such failure were.to occur in the hot leg, existing BMI calcu-l 1ations (e.g., AB-hot leg) might well serve as adequate j

surrogates for the primary system retention.

However, if temperature-induced steam generator tube ruptures prove feasible in TMLB' and TMLU accidents, the resulting source term j

would be unique in that fission products would be relieved j

directly to the atmosphere through the main steam relief valve.

i 4.2 Peach Bottom Me have identified two areas in which the existing BMI l

calculations for Peach Bottom appear insufficient.

First, considering the frequency of TQUV sequences and the possibility i

of ECC pump failure before containment failure in TC and TM DRAFT INFORMAL AND PREllMINARY AND AS ST ' "AY CT4TAIN ERR 3RS tiOT YET CORh.C1ED.

i FOR..i. OUSE PRIVATE Di3TRl30il0N AND NOT

,- F04 EXTERNAL RELEASE WITHOUT CONSENT OF f

AUTH0RS.

l l.

..r l

i l

sequences, we recommend source term calculations be performed for the TQW sequence at Peach Bottoa.

Second, scenarios i

involving late leaks or late containment failures must currently be conservatively binned with scenarios involving i

early containment failure.

This seems overly conservative, and we recommend late leakage be addressed as one of the possible TQW containment f ailure modes.

4 4.3 Grand Gulf i

Both the TC and TW sequences lack a calculation which includes a small bypass of the suppression pool.

It would not necessarily take a large fraction of the flow bypassing the l

pool to significantly change the magnitude of the fission product release.

Thus, we recommend that a calculation be performed for either TC or TW with leakage through the drywell wall developing soon after vessel breach.

The TC sequence is i

probably the better choice since it has a higher source ters.

j A similar situation exists for the TQW sequence, in that none of the calculations have included suppression pool bypass.

The recommended scenario would be an early suppression pool bypass followed by late containment failure.

In addition, our i

pessimistic quantification showed a high likelihood of contain-ment failure at vessel breach.

Since the calculation performed l

in BMI-2104 had containment failing very late, we recommend i

that an early failure scenario be calculated.

From our results, j

it would be advisable to include a small suppression pool bypass I

in this calculation.

I l

j 4

i i

i i

1 i

1 i

l 1

l DRAFT. INFORMAL AND PREllMINARY AND Af i

S'M ! 'tAY C01TAIN E3R')RS NOT ' E' NICTED.

E ON.G!n. AE NOT l F0t cXT.iRNAL RELEASE WITHOUT CONSEN AUTHORS.

i

. f.

2

_--_,. ~_ __ - ---_,_... __--. _.. _ __ _

_~,.r__--_,____--.--,_...---.

4 l

REFERENCES l

1.

(CLNG).

Report of the Containment Loads Working Group, to I

be published.

2.

(CLNG).

Concensus Summaries for Standard Problems 1 through 6.

Letter Reports Addressed to J. Telford, USNRC, i

i May-June, 1984.

3.

(CLNG).

F. E. Haskin, et al., Combustion-Induced Loads in Large Dry PWR Containments. Containment Intearity j

Workshon, June 1984.

Also, letter to M. Silberburg, USNRC, March 9, 1984.

4.

(CPNG).

Containment Performance Working Group, I

Containment Leak Rate Estimates. NUREG-1037, Fourth Draft, i

April 1984.

j 5.

(BMI).

J. A. Giesecke, et al., Radionuclide Release Under

-l Specific Accident Conditions," BMI-2104, Volumes 2, 3,

5, j

and 6. Drafts, January-June, 1984.

l 6.

(QUEST).

R. J. Lipinski, et al., Uncertainty in l

Radionuclide Release Under Specific LWR Accident i

Conditions, SAND 84-0410, Draft Report, March 1984.

l 7.

(IDCOR).

IDCOR Program Reports. IDCOR Task 23.1, j

Integrated Containment Analysis, Preliminary Drafts, May 1984.

8.

(SASA).

F. E. Haskin, et al., Analysis of Hypothetical l

l Severe Core Damage Accidents for the Zion Pressurized l

i Water Reactor NUREG/CR-1989, SAND 81-0504, October 1982.

9.

(SASA).

A. L. Camp, et al., MARCH-HECTR Analysis of l

Selected Accidents in an Ice-Condenser Containment.

l NUREG/CR-3912. SAND 83-0501. Draft Report, November 1983.

10.

(SASA).

D. H. Cook, et al., Loss of Decay Heat Removal Sequences at Browns Ferry Unit One - Accident Sequence Analysis NUREG/CR-2973, ORNL/TM-8532, May 1983.

i l

11.

(SASA).

R. M. Harrington, et al., ATNS at Browns Ferry i

l Unit One - Accident Sequence Analysis, NUREG/CR-3470,

)

ORNL/TM-8902, to be published.

f 12.

(SASA).

S. Z. Bruske, et al., severe Accident Sequence i

Analysis (SASA) Program Sequence Event Tree:

Boiling l

~

l Water Reactor Anticipated Transient Without Scran, Preliminary Draft, December 1983.

l l

i l

,, fyRMAL ANO pyguygygy,MD J iTAli Etim,%0T F

~32-

{d.

M:3EPRo,ygo p.78

?' CKD.

l tJ. itTdlm gggE %I U iini of s.. II]T l

MTHORs, n

~

.. ___ _ I __

13.

(SAUNA).

J. B. Rivard, et al., Identification of Severe Accident Uncertainties, NUREG/CR-3440 SAND 83-1689 Draft Report, March 1984.

14.

(ASEP).

A. M. Kolaczkowski, et al.,

Interim Report on Accident Sequence Likelihood Reassessment, Draft Report, August 1983.

15.

(ASEP).

F. T. Harper, et al., Appendix to NUREG-0956, to be published.

t 16.

(PATF).

A. S. Benjamin, et al., SARRP - Risk Rebaselining and Risk Reduction Analysis, lith Water Reactor Safety Besearch Information Meetina, NUREG/CP-0048, Vol.

3, January 1984.

17.

(RSS).

Reactor Safety Study, NASH-1400, NUREG-75/014 October 1975.

I 18.

(RSSMAP).

Reactor Safety Study, Methodology Applications Program, NUREG/CR-1659. Volumes 1-4, 1981-1982.

I 19.

(ZPSS).

Zion Probabilistic Safety Study Commonwealth Edison Co., 1981.

20.

(SPSS).

Sizewell B Probabilistic Safety Study, NCAP-9991, Westinghouse Electric Co.,

1982.

.i 21.

(SPSA).

Seabrook Plant Probabilistic Safety Analysis, Yankee Atomic Electric Co., December 1983.

I 22.

(GESSAR).

GESSAR-II, BWR/6 Standard Plant Probabilistic Risk Assessment, General Electric Co.,

1982.

I 23.

(FSAR).

Final Safety Analysis Reports for Surry Unit 1.

Zion Units 1,2, Peach Botton Unit 2, and Grand Gulf Unit 1.

24.

(S&W).

A. Drozd, et al., Parametric Study of Aerosol Behavior Following AB and TMLB Accidents. ANS 1984 Annual Meetina, TP84-54, Stone and Webster Engineering Corp.,

1984.

25.

(AMES).

L. Greimann, et al., Final Report, Containment Analysis Techniques, A State-of-the-Art Summary, NUREG/CR-3653, SAND 83-7463, March 1984.

26.

(FVCS).

F. T. Harper, et al., The Effect of Filtered-Vented Containment Systems on Severe Accident Frequencies j

and Consequences for a Mark III Boiling Water Reactor, Sandia Draft Report, July 1984.

i i

DR4FT. INFORMAL AND PREllMINAPY #,'D A$

S' T C3: tai'iE.m Sto u;D.

l F;.1.1. O'J,il JRI. A.E 0. i.u. -

,. NOT F0.. EMAL RElf.ASE WiiH0ul was Of

.../

AUTHORS.

1 27.

(FVCS).

A. S. Benjamin, et al., Value Impact Assessment of' Filtered-Vented Containment Systems and Other Safety Options for a BWR Mark I Containment, Sandia Draft Report.

August 1984.

28.

(NRR).

M. B. Weinstein, Primary containment Leakage Integrity:

Availability and Review of Failure Experience, Nuclear Safety, 21, 1980.

29.

(BNL).

W. T. Pratt, et al., Containment Response During Degraded Core Accidents Initiated by Transients and Small Break LOCA in the Zion / Indian Point Reactor Plants, NUREG/CR-2228 BNL-NUREG-51415, July 1981.

30.

(A-43) A. W. Serkiz, Containment Emergency Sump Performance. Technical Findings Related to Unresolved Safety Issue A-43, NUREG-0897 April 1983.

l i

i j

i t

DRAFT INFORMAL AND PREllMlvaeV AND AS SU: i "V/ C3NTAl'1 EE.T193 ' 0*

CIED.

FOR i 4 OUSE /Rl!AiE Li...

a.u NOT F0,1 LXTERNAL RELIASE WITH0Ji C0iiSENT OF

/

AUTHORL

/

---

,,----,-----,v-

_. = -

I e

e

,h TASLE 3.1.18803S TO DE ASSSESSES AND RSL8vamCE OF BBC3p? ItercenATION SOURCES *

.N BASES FOR BIWutuG S00eCE TsanS:

INF0peATION SOURCES:

't IIDs SISEtsELL SEAge00E CIAfG CPuG Bel-2154 QUEST IDCOR s

I 4

1.

Sise of Proestettag Contatsomet Emekage.

I E

E

==

3.

stee and Lacettee of the primary systee steek surtog the melt senesee (e.g., set Leg, Cold Lege PORT Steen Generator Tone, tris Chect valvel.

I 3

I 2.

Tielag of Accomelater elecharge selettee to T!steg of Beactor veseen Sceech.

4.

Occorreece of to-Teeeel stees Bepleeten 4

Large sneegh to Fett the peector vessel E

I S.

Occertence of In-Weesel Steam gepleeten E

E E

Large sneegh to Fe!! the Caetatement, 6.

Tietag and Itegnitade of Berly pydrogee I

E I

E I

E

.=

Sarae.

E I

E 3

3 7.

megettede of the to-Vessel Steam Spthe.

I W

8.

Beteet of Street Beetteg of the Atesophore telleeleg Teees! Breech.

I g.

Caetatement structural tellere pressere E

E I

I I

18. Sise of Centelmeest Leenage ledeced g g y en g;;p Dy Temperatore et Pressere.

-2

% x ^.. '**

St. servivahtlity of Centelemoet spreys X

o {g M

and Fan Coolere at Varless Times dortog Jg $ [.-

ch. Aecteest.

2 c32

13. servivahtitty of suppreselee Poele 3

a. ~

and Ice Ceedessers at Vertees Times F '" _7 O' Dettag the Accident.

= > -

se I

" 3_ 2 y

13. sateet of Core-Concrete Interacties.

E I

E 5M I

I E

~

'D

14. Tietog and Itogettede of Late Nydrogen mW2 g,gne, m C lb

• 5 i, I

15. poteettel for street Core Detrie Attack y{

en contatement Structures.

I I

h s'y"*

O '~

16. Peteettet for Core Dobria to flett I

E I

i C

4g Throegh the Saeeeet.

C E

17. Peteettet for Seseest neltthreegh to O

g Caese a sepressertsattee of Centelement.

mm E

18.' Peteettel for Ef fleest to Fame Throegh d4ay c,

Adjacent structures, such as the QQ%g Austitary setteleg.

t

=ameed on teformetten made avellable prior to August 1984. I marks where issue was used for a baste er where

[

teforsettee eserce to app!! cable.

I E'

~

~

i

)

Sta44 3.3. SUWF SWER3pften ass LISE3seems ye sumet 8 8 3

&!stLlecos EUWT pet 0R STWPs SPT!a!STIC CesTe&& passteistig 4.

SCS etese same 1e the Legget meege

.2

.1

.1 3.

SCS Steet Se the set tog 3Cs Steet stee le the larger anage

.5

.3

.3 3.

Freestaties Centateneet seeeege et

.883

.003

.002 Seelettee Pentate sufflesent to Ltestues steeme! OverPressurteettee 4.

BCS Proseste Pelle teles the SCS Steet slee to the 8afget anage Aleset

&&tely Lately I

asesensaatet 94echetge Ptteente Cettete SCS Steet stee le the smaller Seege

&&tely gedetet.

geltaely i

steete S.

stees spite Fellosteg Beester Centelemoet Spters Operating Likely Lisely Ligely Veseed Seit4&teste i

4.

Saese! Byeregee tute Ptiet to et testateneet Steye W esetteg Belitely fedetet.

Mtely Jeet Felleutee Soestet Wessel a&sete meatt&tauch 1.

Costeleaset structural Fallate Just eyetages ante ette Freesto'. leg & set 4

4

.33 pellawaag aseeter vessel meltteteuga sydrogee Sure w/o Presetetleg & set 4

4

.45 i

steen spite e sydrogee Sure with Pte.

4 4

.91 estettag & set steen spite e sydtegen sets e/o Pte.

-4 4

.95 4

estettag & set 8.

Cente&seest neceseo suffseteet to sydrogee Oute and se 8ttettste!

4 4

.14 preclude stadual overPreoeuratettee Pe61ste l

3eestes Jeet Felievang Soestet l

Vessel heltteteuge stees spite e sydrogee Sure med 4

ee.

.45 me attuttute! Pasiste i

9.

Caetatemmet Spray Peilute Ge enttu e as asentes aftot vese.estileg to steen Spite med me Caste &ammet Beestely peestely gelately 1 aneue Petaste twesni.

Pen sale stees spite and me Casteisenet Soestely Belitely fedeter.

}

Pe11ste resettle steete Costateneet Pe11ste gelitely gelitely Litely As. Centeleeeet spray Fellure Geeuttleg stees spite end se Centalmeest Petlere Soestely gelitely

! ster.

IB6tue De maestee Mter Wessel steest med me Ptiet spesy realute posesele eteate 4

1 i

Costatement Pettete end be priot selteely gelitely Likely spray Pailure

22. C.f Ca

,.t.

ist.t-u-esie.

St.es epite.e4,t.re seep.t.u ee sed..t.

ut.ly u se.t ptesmees a Weyet&setise Selease aasete certate spreye spetettag ge116ely fedeter.

Likely manete

13. Clemet systegee sure curing et Just ateseptete le Paesettle utely

&ltely Lately Pellessag tae vapotasettee selease

33. Cantatesset structural retaste se Pilot Gute, We Cote-Centrete, 4

et.

.37 free a saamel sydrogee sure aptere operettag et &eet setete Butleg M Jeet Felleslag the 9eportsettee meAceae De Frier Sure, Cote Centrete Geeste, 4*

=4

.37 aptere Operetteg et Amet Selete Prior Sute, Cote.Ceestete Geeste, 4

4 4

speare spetelleg et &see gesete se Pflet Oste, se este-Censtete, 4*

=0 e77 es Speer et less se Priet Gute, Cete.cometete escute, 4

.006 77 as seter a ames i

Friet sets, Coto.cometete essets, ee.

.405

-4 se seter et nees

34. sete Casteimeest opger Pettere Centateneet Pe11ste esemes solitely es!!aely utely

48. nate Centeteneet beseege sete sydrogen este, aptore opetetteg,

-e.

.e.

.e9 se Pttet heet e

se Pt408 Guise note sydrogen este, spreys seapotestee.

4 4

.3 Se Pflet Gute, se Pilot imot

.3 so,sete e,etegen eues, a,tey.

see w eelee umri. mruna. mu rraumaru ar:u m 3.........,...g.3......,,.

geseQie.43

...si

c....

1

~ 3 6 Y "' I ' M I*

' ' G W

f.

F.S d.TiR.M. FE.:ki 'C' a...,,,t.:... f 0F i - -

y,,,,p...

_w...

.,,,_,__c

_._.,__,m

_.m..__.-.___.____.,_,_m._,_

... ~.

=

s TaaLa 3.2. grarr BASCRIPTIONS Ame L2EaL2300es paa sumar age CONT'D LIBELie000 Sveut pe10e BVSNTS OPT!alSTIC CBNTRAL PESS1m!STIC

~

le. essenet aalttarenge Sprays Isoperative, Core-cenerete Geeers Lately aloest Certale certale i

Spreye Sporettag, Core-Concrete Geeers ledeter-Lately Aleost steate certata

17. Late Centstaneet Overpressurisettee me sesseet melttarough, we spray er Leak Lately Aleest Certata Certata asee. met meltthreegh Occure, no Umlately e.d.eter.

Lately In r..r m.

.te

18. askease typesees the ass 111ery aseve Greemd streeteral Pe11ere Lately Lately Aleoet as11eaeg Certa 1a Ateve tremed Caetatammet Leek Dalikely Dalitely De18tely DRAFT - INFORMAL AND PREllMINARY AND AS S W 'Y'CTJTAl'4 E3MS.' 0T '.ET CT. Ci;D.

F0R,.'

.;'E RI. A ti sai.tidull01 Allu NOT F03 cXi tdil itELEASE WITHOUT C0iiSENT OF AUTHORS.

4....._,.

l I

=

T?

TABLE 2.3.

ESTIMATES OF CONTAINMENT PRESSURE.

LOADING AND CAPACITY FOR SURRY S D 2

Optimistic Central Pessimistic Containment Pressure Just Following Vessel Breach (psig):

H2 Burn With 27 45 78 Preexisting Leak H2 Burn Without 32 50 83 Preexisting Leak Steam Spike + H2 Burn 27 60 105 With Preexisting Leak Steam Spike + H2 Burn 32 65 510 Without Preexisting Leak i

Containment Pressure 119 119 85 (1

5)

(1 15)

(i 15)

Causing Structural Failure 2

(1 Standard Deviation)(psig) 4 DRAFT. INFORMAL AND PRELIMINARY AND AS SJ' ; & 03 iTAIN ERR 1RS i;0i YET CORECliD.

FOR IN. 0J'E. RI. AiE 3.i.ileUlla 41. NJT F0.1 EXTERN.4L dELEASE WITHOUT CO..GI 0F AUTHORS.

...f..

s.a,

...-._- ?.-..

Table 2.4.

ESTIMATES OF CONTAINMENT PRESSURE CAPACITY AND LOADING FROM A LATE l

HYDitOGEN BURN Ootimistic Central Possimisti Containment Pressure Following a Late Hydrogen Burn (psig):

No prior burn, 5*

43 76 No core-concrete, Spray or leak exists No prior burn, 43 59 76 Core-concrete occurs, Spray or leak exists Prior burn occurs, 43 59 43**

Core-concrete occurs, Spray or leak exists No prior burn, 25*

63 96 No core-concrete, No spray or leak No prior burn, 63 79 96 Core-concrete occurs, No spray or leak Prior burn occurs, 63 79 63**

Core-concrete occurs, No spray or leak Containment Pressure 119 119 85 Causing Structural Failure (1 5)

(115)

(115) 2 (1 Standard Deviation)(psig)

• No burning occurs because of-insufficient hydrogen

• • Oxygen depletion occurs before burn completion.

DRAFT. INFORMAL AND PREUMINARY A.ND AS S.U. 'A : HTA!N ERR 3RS NOT YEiCORCC ED.

FOR A 0J:E ?RNATE D31816U110.~. Ak NOT FOR NERiu REM MOM W.T W,-

AMHORS.

'.,....... /- _,,,

,_-_-..---_,--..,------_.-...-.-,-._----s w.

TABLE 2.5.

ALTERNATIVE ASSIGNMI!UiT OF VALUES TO VERBAL DESCRIPTORS LIKELIHOOD VERBAL ALT. 1 ALT. 2 ALT. 3 ALT. 4 DESCRIPTOR (BASE CASE)

Certain

)

tr 1.0 1.0 1.0 1.0 Almost Certain s

Likely 0.9 0.9 0.9 0.9 Indeterminate 0.5 0.5 0.5 0.5 Unlikely 0.1 0.01 0.01 0.1 Remotely Possible 0.001 0.001 0.0001 0.01 Impossible 0

0 0

0 4

ors i. INFORMAL AND PRELIMINARY AND c

S

.i : TAl:4 E7R1R;;.0T YET L3n...CiiD.

al ATE OdlRioUil0N AND NOT F0.1 1

. F0., IXTu..

.(ELEASE WlIHOUT CONSENT OF AUTHORS.

i TABLE 3.1 RESULTS FOR SURRY AB I.

SEQUENCE FREQUENCIES OPT.

CENTRAL PESS.

AB (Hot Leg)

<1x10-9 AB (Cold Leg)

<lx10-9

' II. CONTAINMENT FAILURE MODE PROBABILITIES l

OPT.

CENTRAL PESS.

Lo Containment Failure

.01 Basemat Meltthrough

.79

.49

.05 Late Overpressurization

.20

.50

.41

.15 Late Induced Leakage

.002

.38 Late Hydrogen Burn Early Induced Leakage Carly Steam Spike Early Hydrogen Burn Early Steam Spike + N2 Burn Isolation Failure or

.002

.002

.002 Preexisting Leak III. PRINCIPAL CONTAINMENT PATHNAYS BMI-2104 CALC.

(1) Basemat Meltthrouch.

Vaporization Yes, but without release occurs.

A low-capacity leak leak may develop (central and pessimistic).

(2) Late overnressurization.

No vaporization release may or may not occur (optimistic) or occurs (central and pessimistic).

A low-capacity leak may develop (central and pessimistic).

(3) Late Induced Leakace.

No Vaporization release occurs.

(4) Late Hydrocen Burn.

Vaporization Yes release occurs.

l (5) Isolation Failure or Preexistina Yes Lggi.

Vaporization release occurs (central and pessimistic) or may occur (optimistic).

DRAFT. INFORMAL AND PREllMINARY AND S ;: ' A ' O rlTAl:i ERRORS i 0n ET CORRsciED.

RI A.E ).Misull0,i A:4D NOT F9.1.:4 F0.

Jh....,.tELEASE WITHOUT COl4SENT OF /

AUTHORS.

/

~

l TABLE 3.2 RESULTS FOR SURRY S2,3D f I.

SEQUENCE FREQUENCIES 3

OPT.

CENTRAL PESS.

i 8 D Bot Leg 5x10-6 2

S D Cold Leg 5x10-6 2

5 D Cold Leg 9x10-5 3

i II. CONTAIIGEENT' FAILURE MODE PROBABILITIES OPT.

CENTRAL PESS.

4 No Containment failure

.95

.46

.006' Basemat Meltthrough

.05

.45

.06 i

Late Overpressurization

.001

.09

.03

)

Late Induced Leakage

.01 j:

Late Hydrogen Burn

.001

.06 Early Induced Leakage

.03 l

Early Steam Spike Early Hydrogen Burn

.04 Early Steam Spike + H2 Burn

.77 l

Isolation Failure or

.002

.002 Preexisting Leak i

III. PRINCIPAL CONTAINNENT PATHNAYS BMI-2104 CALC.

i (1) No Containment Failure.

Sprays survive.

No, but source Core is coolable in reactor term similar to cavity, hence no vaporization release.

(2) 4 (2) Basemat Meltthrouch.

Sprays survive.

Yes, for 5 D 2

Core attacks basemat, and there is a hot leg vaporization release.

(3) Late Overnressurization.

Sprays fail No either before or after vaporization release.

A low capacity leak may develop in central and pessimistic cases.

j (4) Late Hydronen Burn.

Sprays fail either No q

before or after vaporization release.

L 1

l j

(5) Early Steam Snike & h,,,3313 Sprays Yes. for 8 D l

2 l

tail at containment failure.

Vapori-cold leg

{

sation release occurs.

1 (6) Isolation Failure or Freexistina Leak.

No Sprays survive (central) or fail atter

(

)

vessel breach (pessimistic).

Vaporization i

j release occurs (pessimistic) or may occur (central).

i DRAN. Wrnomt un wp u'u T A3 SUM W e,. ;iA:*i EP C ' OP E : CZ~.*MD.

i FOR S h A c s ih o' na. A'n.'ai J

, F01 dia....

..aiA3E WliliOUT C0&M 0F

/

AUiliOK<

. _= -

i I

i i

I TABLE 3.3 l

l RESULTS FOR SURRY TMLB'

> 1.

SEQUENCE FREQUENCIES 4

j OFT.

CENTRAL PESS.

'ISELB Short Term 2x10-5 l

j TMLB Long Tera 1x10-4 i

1

)II.CONTAIIREENTFAILUREIIODEPROBABILITIES f

OPT.

CENTRAL PESS.

}

-[

No Containment Failure

.004 i

i Basemat IIeltthrough

.48

.47

.04

]-

Late Overpressurisation

.52

.52

.38

.14 Late Induced Leakage

.005

.43 J

Late Ilydrogen Burn j

Early Induced Leakage Early Steam Spike 4

.001 Early Ilydrogen Burn Early Steam Spike + N2 Burn Isolation Failure or

.002

.002

.002 Freexisting Leak III. PRIIICIPAL C00f7AIISERNT PATNtfAYS

[

j 3361-2104 CALC.

i i

(1) Me containment Failure.

No, but source Vaporisation Release occurs, term is similar j

to (2)

(2) Ras mat Meltthrouah.

Vapori-Yes. but without I

sat on release occurs.

A low-leak capacity leak may develop j

(central and pessimistic).

l (3) Lat9 Overnressurisation.

Vapori-No sation release may er may not l

occur (optimistic) or does occur l

3 l

(central and pessimistic).

A l

low capacity leak may develop (central and possimistie).

i (4) Lat, Induced Leakana.

Vapori-No.

i sation release escurs.

l 1

(5) { dip Hydronen Burn.

Vapori-No i

j sathen release escurs.

)

(4) ET Myerosen turn.

Vaperi-No.

sathen release escurs.

(7) Isolation Fahlure er Freamistina No Isaag.

Vaperasation release ecours (eentral and possimistie) er OR4FT. INFORMAL AND PHillMINARY AND5 may escur (optimistie).

S'D !

'Y' C31TAl1 E1R)RS fiOT YET CORR:Ci :D.

l run IJat ifl. A6L Wi46MlJUlIUit Ad. " )I

/ F0 s...i..thL.siLEASE WITHOUT CONS.. 1r l

AUTH3RS.

/

i u,. - -

O TABLE 3.4 RESULTS FOR ZION S D 2

I.

S~QUENCE FREQUENCIES OPT.

CENTRAL PESS.

SEFC (Seal LOCA due to loss 2x10-4 of component cooling water)

II. CONTAINMENT FAILURE MODE PROBABILITIES OPT.

CENTRAL PESS.

.s No Containment Failure

.95

.49

.02 2asemat Meltthrough

.05

.45

.17 Late overpressurization

.05

.10 Late Induced Leakage

.13 Late Hydrogen Burn Early Induced Leakage

.30 Early Steam Spike Early Hydrogen Burn Early Steam Spike + H2 Burn

.27 Isolation Failure or

.003

.009

.009 Preexisting Leak III. PRINCIPAL CONTAINMENT PATHWAYS l

BM1-2104 CALC.

(1) No Containment Failure.

Sprays' survive.

Yes Debris is cooled in reactor cavity.

and there is no vaporization release.

(2) Basemat Meltthrouch.

Sprays survive.

A No, but source term l

vaporization release occurs, is similar to (1)

(3) Late Overnrossurization.

Sprays fail No either before or after the vaporization release.

A low-capacity leak may be induced (central and pessimistic).

(4) Late Induced Leakane.

Sprays fail after No the vaporization release.

(5) Early Induced Leakaae. A leak is induced No shortly after vessel breach by a pressure spike (HB+SS).

Sprays may or may not survive.

Vaporization release occurs.

(6) Early Steam Spike + Ho Burn.

Sprays fail No at containment breach-Vaporization release occurs.

(7) Isolation Failure or Freexistina Leak.

No Sprays survive (optimistic and central) or fail after vessel breach (pessimistic).

DRAFT. INFORMAL AND PREllMINARY AN ) AS Vaporization release does not occur (optimistic) or does occur (pessimistic).3.), MY CTITAM E?R300 01 YET C0P.P.3 TED.

FOR 'i CE Til AiE Sa.sl3tJT10ti AN NOT

/

F0.4.AI..t.tw.tiLF.A5h WiltiUUl WN5tNI 0F 44

/

AUTH0RS.

.....a

~.

i TABLE 3.5 RESULTS FOR ZION TMLB' I.

SEQUENCE FREQUENCIES OPT.

CENTRAL PESS.

TE (Seismic Induced; 6x10-6 leads to seal LOCA)

II. CONTAINMENT FAILURE MODE PROBABILITIES OPT.

CENTRAL PESS.

{

No Containment Failure

.006 Basemat Meltthrough

.48

.45

.04 Late Overpressurization

.52

.54

.44 Late Induced Leakage

.50 Late Hydrogen Burn Early Induced Leakage Early Steam Spike Early Hydrogen Burn Early Steam Spike + H2 Burn Isolation Failure or

.003

.009

.009 Preexisting Leak III. PRINCIPAL CONTAINMENT PATHNAYS BMI-2104 CALC.

(1) No Containment Failure.

Yes i

Vaporization release occurs.

(2) Basemat Meltthrouch.

Vaporization No release occurs.

A low capacity leak may develop (central and pessimistic).

(3) Late Overnressurization.

Vapori-No sation release does not occur (optimistic) or does occur (central and pessimistic).

A low capacity leak may develop (central No and pessimistic).

(4) Late Induced Leakace. Vaporization No release occurs.

(5) Early Steam Snike + H, Burn.

No Vaporization release occurs.

(6) Isolation Failure or Preexistina Leak.

No Vaporization release may or may not occur (optimistic and central) or does occur (pessimistic).

DRVT - INFORMAL AND PRELIM F$"'

" > TAM E'R]RS LOT YET C0R.C

-4s-

'ji'A E 0011d710:, Ax. ;oi r0 ggfH5b~.

H0dT COMEar or

.._7._.._....

_ m _.

TABLE 3.6 PEACH BOTTON AE

.I.

SEQUENCE FREQUENCIES OPT.

, CENTRAL PESS.

AE 2x10-7 II. CONTAINMENT FAILURE NODE PROBABILITIES

• OPT.

CENTRAL PESS.

No Containment Failure Late Induced Leakage Scrubbed Partly Scrubbed Not Scrubbed

.90

.45

.05 Late Overpressurization Scrubbed

.04

.02

.005 Partly Scrubbed Not Scrubbed

.05

.43

.44 Leakage Induced Before or at Vessel Breach Scrubbed Partly Scrubbed Not Scrubbed

.005 Overpres. Before or at Vessel Breach Scrubbed

.005

.005 Partly Scrubbed

.005

.04 Not Scrubbed

.09

.45 Pre-Core-Melt Overpressurization Scrubbed Partly Scrubbed Not Scrubbed III. PRINCIPAL CONTAINMENT PATHNAYS BMI-2104 CALC.

(1) Late Induced Leakace/Unscrubbed Release.

No No early containment leakage or failure.

A temperature-induced leak develops in the drywell during the core-concrete interaction.

bypassing the suppression pool but preventing gross containment failure.

The secondary containment remains intact (blowout panels relieve).

The standby gas treatment system does not filter the vaporization release (central &

pessimistic)

(2) Late OverDressurization/Unscrubbed Release.

No No containment leakage, no early containment failure.

Containment fails either in the drywell, causing bypass of the suppression No pool, or in the wetwell (optimistic and central), causing the suppression pool to drain.

The standby gas treatment system does not filter the vaporization release.

(3) OverDressurization Before or at Vessel Yes Breach /Unscrubbed Release.

Containment failure occurs in the drywell early due to buildup of steam and hydrogen.

Suppression

_g_

pool is bypassed, and the secondary containment and the standby gas treatment system fail such R

DRAF-that the release is not filtered.

b*

__ ___ _i.I_ l_.

TABLE 3.7 i

PEACH BOTTON TC I.

SEQUENCE FREQUENCIES OPT.

CENTRAL PESS.

TC 7x10-6 l

II. CONTAINMENT FAILURE MODE PROBABILITIES OPT.

CENTRAL PESS.

No Containment Failure Late Induced Leakage Scrubbed l

Partly Scrubbed Not scrubbed Late Overpressurization Scrubbed Partly Scrubbed Not Scrubbed Leakage Induced Before or at Vessel Breach Scrubbed Partly Scrubbed Not Scrubbed

~

Overpres. Before or at Vessel Breach Scrubbed Partly Scrubbed Not Scrubbed Pre-Core-Melt overpressurization Scrubbed

.23

.009

.005 Partly Scrubbed

.001

.005 Not Scrubbed

.67

.99

.99 Equivalent to TQUV

.10

.001 III. PRINCIPAL CONTAINMENT PATHNAYS BMI-2104 CALC.

(1) Pre-Core Melt OverDressurization.

Containment No fails in the wetwell before core-melt due to steam overpressurization.

The primary system has depressurized before core-melt.

The suppression pool remains filled and is not by-passed.

A vaporization release occurs and secondary containment is not bypassed.

(2) Pre-Core-Melt OverDressurization/Unscrubbed Yes except BMI Release.

Containment fails either in the calculation drywell, causing bypass of the suppression assumes primary pool, or in the wetwell, causing the pool to system remains drain.

The primary system has depressurized pressurized before vessel breach.

A vaporization release occurs, and secondary containment is not bypassed.

(3) Ecuivalent to TOUV.

The ECCS pumps fail before No l

containment fails.

The core melts before contain-ment failure, and the accident progresses as an accelerated TQUV sequence.

(TQUV has not yet been analyzed for Peach Botton).

/

iDRAFT

(

TABLE 3.8 PEACH BOTTOM TW l

I.

SEQUENCE FREQUENCIES OPT.

CENTRAL PESS.

1 TW 8xlO-6 II. CONTAINMENT FAILURE MODE PROBABILITIES OPT.

CENTRAL PESS.

No Containment Failure Late Induced Leakage Scrubbed Partly Scrubbed Not Scrubbed Late Overpressurization Scrubbed Partly Scrubbed Not Scrubbed Leakage Induced Before or at Vessel Breach Scrubbed

.01 Partly Scrubbed Not Scrubbed

.01 Overpres. Before or at Vessel Breach Scrubbed

.006 Partly Scrubbed Not Scrubbed

.02 Pre-Core-Melt Ove,rpressurization Scrubbed

.11

.008

.005 Partly Scrubbed

.001

.005 Not Scrubbed

.34

.89

.99 Equivalent to TOUV

.50

.10

.001 III. PRINCIPAL CONTAINMENT PATHNAYS BMI-2104 CALC.

(1) Pre-Core Melt OverDressurization/ Scrubbed No Release.

Containment fails in the wetwell before core-melt due to steam overpressurization.

The primary system has depressurized before core-melt.

The suppression pool remains filled and is not by-passed.

A vaporization release occurs and secondary containment is not bypassed.

(2) Pre-Core Melt OverDressurization/Unscrubbed Yes Release.

Containment fails in the drywell causing bypass of the suppression pool, or in the wetwell, causing the pool to drain.

The primary system has depressurized before vessel breach.

A vaporization release occurs, and secondary containment is not bypassed.

(3) Eauivalent to TQUV.

The ECCS pumps fail before No containment fails.

The core melts before contain-ment failure, and the accident progresses as an accelerated TQUV sequence.

(TQUV has not yet been analyzed for Peach Botton).

./

DRAFT

.u___._,_._----...-.-.;

--_.---.-r l

'~

TABLE 3.9 GRAND GULF TC

'I.

SEQUENCE FREQUENCIES OPT.

CENTRAL PESS.

T23C 5x10-6 II. CONTAINMENT FAILURE MODE PROBABILITIES OPT.

CENTRAL PESS.

No Containment Failure

.009 Late Induced Leakage scrubbed

.004 Partly Scrubbed Not Scrubbed i

Late overpressurization Scrubbed

.09

.001 Partly Scrubbed Not Scrubbed Leakage Induced Before or at Vessel Breach Scrubbed

.001 Partly Scrubbed Not Scrubbed overpres. Before or at Vessel Breach Scrubbed

.001 Partly Scrubbed Not Scrubbed Pre-Core-Melt overpressurization Scrubbed

.90 1.00

.90 Partly Scrubbed

.001

.002

.10 Not Scrubbed l

III. PRINCIPAL CONTAINMENT PATHWAYS BMI-2104 CALC.

(1) Late Overnressurization/ Scrubbed Release No The ECCS pumps fail due to high temperature before containment failure.

The scenario progresses as an accelerated TQUV scenario.

The primary system i

depressurizes.

Containment does not fail or leak I

early.

Core-concrete interactions occur yielding a i

vaporization release and late containment failure due to accumulation of noncondensible gases.

(2) Pre-Core-Melt Overnressurization/ Scrubbed Release.

Yes, except BMI Containment fails before core-melt due to steam calculation overpressurization.

The primary system has assumes primary I,

depressurized.

There is no bypass of the system remains suppression pool.

A vaporization release occurs.

pressurized.

(3) Pre-Core-Melt Overnrossdrization/Saall Evna33, No Containment fails before core-melt due to steam overpressurization.

The primary system has depressurized.

A leak through the drywell wall develops after vessel. breach which allows a small bypass of the suppression pool.

A vapori-zation release occurs.

1 k.

,. ~ /

^ ~ ^

l TABLE 3.10 CRAND GULF TPI l

I.

SEQUENCE FREQUENCIES OPT.

CENTRAL PESS.

23 QI 2x10-7 T

P II. CONTAINMENT FAILURE MODE PROBABILITIES OPT.

CENTRAL PESS.

No Containment Failure

.05 Late Induced Leakage Scrubbed

.02

.004 Partly Scrubbed Not Scrubbed Late Overpressurization Scrubbed

.42

.08 Partly Scrubbed

.009 Not Scrubbed Leakage Induced Before or at Vessel Breach Scrubbed

.03

.001 l

Partly Scruht,ed Not Scrubbed Overpres. Before or at Vessel Breach Scrubbed

.03

.007 Partly Scrubbed

.001 Not Scrubbed Pre-Core-Melt Overpressurization Scrubbed

.45

.90

.90 Partly Scrubbed

.002

.10 Not Scrubbed

!III. PRINCIPAL CONTAINMENT PATHNAYS BMI-2104 CALC.

(1) No Containment Failure.

The ECCS pumps fail No, but

• i due to high temperature before containment failure.

similar to The scenario progresses as a retarded TQUV scenario.

TQUV.

The primary system depressurizes.

Containment does not fail or leak early.

Core-concrete interactions are arrested and steady state is achieved before l

containment failure.

(2) Late overnrossurization/ Scrubbed Release.

No Same as (1) except as follows.

Core-concrete interactions occur yielding a vaporization l

release and late containment failure due to accumulation of noncondensible gases.

(3) Pre-Core-Melt Overnressurization/ Scrubbed Release.

Yes Containment fails before core-melt due to steam overpressurization.

The primary system has depressurized.

There is no bypass of the suppression pool, and a vaporization release occurs.

(4) Pre-Core-Melt OverDressurization/Saall Bvoass No Same as (3) except as follows.

A leak through the drywell wall develops after vessel breach which allows a small bypass of the suppression pool, and a vaporization release occurs.

){j /

a

.-.-.a.u.-.-

--..n o-1 TABLE 3.11 i

GRAND GULF TQUV I.

SEQUENCE FREQUENCIES l

OFT.

CENTRAL PESS.

7 00V 4x10-6 1

II. CONTAIISIENT FAILURE NODE PROBABILITIES OFT.

CENTRAL FESS.

No Containment Failure

.09

.001 Late Induced Leakage Scrubbed

.04

.04 Partly Scrubbed

.004 Not Scrubbed Late overpressurization Scrubbed

.85

.79

.23 Partly Scrubbed

.001

.09

.25 1

Not Scrubbed Leakage Induced Before or at Vessel Breach Scrubbed

.003

.004

.002 Partly Scrubbed

.004

.001 Not Scrubbed Overpres. Before or at Vessel Breach Scrubbed

.W6

.07

.22 l

l Partly Scrubbed

.007

.27 l

Not Scrubbed l

Pre-Core-Melt Overpressurisatica Scrubbed Partly Scrubbed l

Not Scrubbed III. PRINCIPAL CONTAIISIENT PATIBIAYS I

l BNI-2104 CALC.

(1) No cantainment Failure. The primary system No is depressurised before vessel breach. Core-concrete interactions are arrested and steady state is achieved.

(2) Late overnressurination/ scrubbed Release Yes, except The primary system is depressurised before vessel no modeling breach. Nydrogen burns in the wetwell may cause of steam spikes.

water to overflow into the drywell. If so, a steam diffusion flames.

spike occurs at vessel breach, and diffusion flames or acatainment la the wetwell may cause a small containment leak to leaks.

develop. Core-concrete interactions occur. giving rise to a vaporisation release and late containment failure due to the buildup of noncondensible gases.

(3) Late overnressurination/small avnans. Same as (2)

No except a leak is induced la the drywell wall after vessel breach such that a small bypass of the suppression peel occurs.

(4) overnressurization Before or at vessel areach/

No Scrubbed Release. The primary system depressurises before vessel breach. A steam spike occurs forcing hydrogen into the wetwell which burns at about the same time. The containment fails. Core-concrete interactions take place yielding a vaporization release. All releases from the primary system pass through the suppression pool.

(5) Overnroscurination Refere or at Vessel areach/

No small aveams. Same as (4) except a leak is induced la the drywell wall after vessel bretch allowing a small bypass of the suppression pool.

DRAFT INFORMAL AND PREllMINARY AND AS SUCH MAY CONTAIN ERRORS NOT YET CORRECTED. FOR IN. ~00SE PRIVATE DislRIBUTION ANu N]T FOR EXTERNAL RELEASE MTHOUT CONSENT W

/

WINDE, y...

-.=,...v..-

'o

's l

f; t

..h

!I 4

l e

e P

i TABLE 3.12 SENSITIVITY OF SURRY S D RESULTS

['

2 J

TO ALTERNATIVE ASSIGNMENTS OF NUMERICAL VALUES j,i i

l'

.i OPTIMISTIC CENTRAL-PESSIMISTIC Unlikely

.1

.01

.01

.1

.1

.01

.01

.1

.1

.01

.01

.1 g

Remote Poss.

.091

.001

.0001

.01

.001

.001

.0001

.01

.001

.001

.0001

.01 No Failure

.95

.99

.99

.93

.46

.54

.54

.45

.006

.007

.007

.006

.l i

1 i

,i Meltthrough

.05

.006

.005

.06

.45

.45

.45

.45

.06

.06

.06

.06 l

ut hJ Late Overpressure

.001

.001

.01

.09

.01

.01

.09

.03

.02

.02

.03 e

i j

Late Leak

.01

.01

.01

.01 f

Late M Burn

.001

.001

.06

.05

.05

.06 l}

2 Early Leak

.03

.03

.03

.03 23 23 !$

Steam Spike 30 30 C7 DC"%

Early H Burn

.04

.04

.04

.04 e

2 1st 5-i

• gg j, SS + P 8"'"

'77

• 77

'77

• 77 2

gjc3 Isol. Failure

.002

.002

.002

.002

.002

.002

.002

.002 j

M4 25 90 22

.l.

2D 3E i

• M1 3$ 30 E l

ni j' 3 mE 52E5

=a sc se g_

EC 22 53 _t hkc$-eE

ec3 CD c3 3E c) t 1

3 30 II EP ?? ji

-4 E3

,. l sR!k93Et

,l.

i t-I.

i t

l j.

s, I.

3 N

t

.'t i

i.

TO BE PROVIDED f

I ui Wl 6

n, o o, en a

u. m N26 p pE~-7

-. - 3;.

rs 4

2* o n5 E "f. v 3 r ar, g,ut 5..>& 2 ;E b h Q m !3 =

wa aa E Y$

mmo_

o-Ek#=

g_*nE a _ z o4 mg g ;g,

FIGURE 2.1.

Illustration of Containment Event Methodology M c O

=

m m.o a 5

0

12 e

?

I i

e

!'\\

~

10 i

i g

t l

s t

ii' 8

i' N5s t'i' i

un a,

i

i 8o

• O PESSIMISTIC i

i y w,o '- x 2

opi 1

l g

h et

• i x v

~

l f

M CENTRAL

< 2. '_' e

.J ma g..,mg

.a.... >

0 i

e i

e i

I

- < r-m m fg g 0

20 40 60 80 100 120 140 i

E 9 3 o g $. t-5$ k PRESSURE, PSIG 5 c' < E FIGURE 2.2.

CPNG Pressure Dependent Leak Model for Zion.

Q 8 = o :r

=l8a5

.i B> 2 2:D :>.

r

%y 9"' o C

2 l

M N.b @

f.

1 I

i

s.

AUG 15 E64 4

Sandia National Laboratories date:

August 9, 1984 Albuquerque. New Mexico 87185 to:

Distribution 4

-,7--

-w S.Benjkmin,6411 Im" A.

Risk Perspective for NUREG-0956 Last April, Bob Bernero asked the Severe Accident Risk Reduction Program (SARRP) to apply its resources toward development of a risk perspective for NUREG-0956, the final report of the Accident Source Term Project Office (ASTPO).

The primary objectives are as follows:

(1)

For each of the six reference plants treated by ASTPO, identify the accident pathways (i.e., the combinations of accident sequences and contain-ment events) that are important to risk.

(2)

Estimate the frequencies of those accident pathways, utilizing to the maximum extent the results of the Containment Loads Working Group, Containment Performance Working Group, Accident Sequence Evaluation Program, and other NRC and Industry Studies.

(3)

Ascertain how well the BMI-2104 source term calculations cover the risk-significant accident pathways, and identify pathways for which additional source term calculations are needed.

(4 )

Provide a letter report suitable as an appendix to NUREG-0956.

The study will be accomplished in two iterations.

The first, completed July 31, 1984, provided preliminary estimates for four of the reference plants -- Surry, Zion, Peach Bottom, and Grand Gulf.

The second, to be completed by November 30, 1984, will complete the analyses for these four plants and will cover the two remaining reference plants, Sequoyah and Limerick.

The attached draft document entitled " Containment Event Analysis and Estimation of' Source Term Frequencies," fulfills our obligation for the July 31 iteration.

It is the result of an intensive effort by members of the SARRP team conducted over E41813 i

/

l

.__..______m._,

. a period of about 2 months.

We have sent an advance copy to Walt Pasadag at NRC, at his reque,st, and are providing copies to those within Sandia who are interested in this work.

We would appreciate comments that will help us to optimize our product for the second iteration.

ASB:6411:cgt Attachment Distribution:

w/ attachments 6414 D.M.

Ericson USNRC J. C. Glynn 6415 D.

C. Aldrich USNRC M. A. Cunningham 6415 J. M. Griesmeyer 6400 A. W. Snyder 6417 D.

D. Carlson 6410 J. W. Hickman 6420 J. V. Walker 6411 A.

S. Benjamin 6422 D. A. Powers 6411 V. L. Behr 6427 M. Berman 6411 S. E. Dingman 6440 D.

A. Dahlgren 6411 R.

D.

Gasser 6442 W. A. von Riesemann 6411 F. E. Haskin 6445 J.

H. Linebarger 6411 D. M.

Kunsman 6449 K.

D.

Bergeron 6412 A. L. Camp 6449 D.

C. Williams 6412 F. T. Harper l

i l

l

~

j

/