ML20206J270

Draft Executive Summary from Analysis of Station Blackout Accidents for Bellefonte Pwr

ML20206J270
Person / Time
Site:	Bellefonte, 05000000
Issue date:	01/24/1986
From:	Bieniarz P, Gasser R, Tills J JACK TILLS & ASSOCIATES, INC., RISK MANAGEMENT ASSOCIATES, SANDIA NATIONAL LABORATORIES
To:	NRC
Shared Package
ML20204G644	List: ML20206C066 ML20206E142 ML20206F071 ML20206F086 ML20206F182 ML20206F313 ML20206G821 ML20206G928 ML20206H081 ML20206H113 ML20206H144 ML20206J252 ML20206J270 ML20206J368 ML20206J403 ML20206J600
References
RTR-NUREG-1150 NUDOCS 8704160056
Download: ML20206J270 (12)

Text

-

m 1 1

Analycio of St0ticn Eleckgut Accid nto for tho Bellofonto Proccurizcd Wator Rscctsr l

d R. D. Gasser ,

Sandia National Laboratories P. P. Bioniarz  ;

Risk Management Associates J. A. Tills Tills and Associates 1.0 EXECUTIVE

SUMMARY

A first attempt at an integrated "best estimate" of containment loading and radiological releases for specific accident sequences was carried out at Battelle Columbus Laboratories and disseminated in the now familiar BMI-2104 report. In that study five reactors were selected in order to characterize different major reactor types, and analyses were performed for a few of the accident sequences that were thought to dominate the risk or present unique challenges to the containment. The BMI-2104 analysis employed computer codes that were run independently and in large part were in an early stage of development. A number of issues have been raised in the interim, some of which are related to phenomenology not modeled in the BMI-2104 suite of codes. It is the objective of this study to address some of these issues with a view to improving the calculated estimate of containment loading and tission product releases to the environment. The present study l

has been limited to large dry containments in general and to the '

TVA Bellefonte Unit 1 in particular.

I

! CURRENT ISSUES An issue of considerable importance for PWRs, since it l supplies a mechanism for early containment failure, is that of j direct containment heating due to the high pressure ejection of molten core debris into the upper containment atmosphere.

Because the calculated loads on the containment structure due to

- direct heating may challenge its integrity simultaneously with near maximum concentrations of radiological sources in the i

containment atmosphere, this scenario probably has the potential j

for the largest releases and the most severe effects on public 1 health.

There are, however, a number of uncertainties attached to the direct heating question which range from the magnitude of its effect to whether it will, in fact, occur. Large I

uncertainties in the core meltdown phase of the, accident, for I

i

! DRAFT i 8704160056 B70408

PDR NUREG

! 1150 C PDR

\ .

' ' t' V

e

cxtsplo, yiold a wido rcngo of poociblo initial cenditiens for coro d0 brio et tho tito of voccol brocch. Much of thio uncertainty may be removed when the more cdvanc d colt-i progression codes such as MELPROG become available. win view of the current lack of knowledge regarding direct heating phenomenology, the present study has taken a parametric approach

! to direct heating.

The BMI-2104 calculations for primary system fission j product transport employed a set of codes (MARCH, CORSOR, MERGE, i and TRAP / MELT) which were run successively (without feedback) l and which did not contain models for fission product heatup of

! the primary system *, nor did they contain models for treatment j of natural convective processes. Transport of volatile fission

! products.through the primary system and releases of these

! materials into the containment are extremely sensitive to the temperature history of the primary system, so that these two processes cannot be neglected. As a consequence of the absence j

i of models for the aforementioned processes, the BMI-2104 calculations for the TMLB' sequence may underestimate primary system releases during the meltdown phase of the accident.

j Further, it was not possible to calculate long term reevolution (rtvaporization) subsequent to vessel failure of those volatile fission products that had been retained on primary system surfaces during the meltdown phase. Certain containment i

conditions, such as a gradual depressurization due to leakage l

over the time period in which the volatiles are being reevolved, ,

i can supply a mechanism for moving these sources out of the primary system and ultimately into the environment.

)

4 In the present study an initial attempt has been made to l treat fission product heating and natural convection in the i

primary system heat and mass transport calculations. A spin off l

of this modeling capability has been a somewhat improved j characterization of the primary system temperature history, and j as a result, the ability to gain some insight into another i important issue: that of potential temperature induced failure

! of the hot-leg piping. Enhanced heat transfer due to natural

! circulation processes could generate high enough temperatures at

! the hot-leg nozzles or in downstream piping to weaken the structure and cause a breach. This may be particularly likely j

for the TMLB' sequence due to the high system pressure j (2000-2400 psia). If primary system failure in the hot leg l I occurred before core slump, the direct heating phenomenon might be effectively precluded. The potential for setting up a i

countercurrent flow regime in the pipes leading from the vessel l to the pressurizer has not been established, and the present

! analysis probably yields heat transfer rates that are somewhat j high. Multidimensional codes may be required to achieve a

! definitive answer regarding natural convection in the l

complicated geometry of the primary system.

1

• The current Battelle code set does include fission product

heating of RCS surfaces but only up to the time of vessel breach.

l DRAFT

CODES AND METHODOLOGY I

The primary strategy in this analysis was to use the best calculational tools available to treat each phase of the accident. The RELAP5 code.was used to calculate the thermo-hydraulic sources during the pre-core-damage phase, and the l

MARCON code was employed during the period from incipient core meltdown out to and including the core / concrete interaction phase (the CORCON code which treats the core / concrete inter-i action has been combined with the MARCH code in the MARCON

! code). A "hard linked" version of the CORSOR, MERCE, and l TRAP / MELT codes was utilized to calculate primary system fission product releases. This code, called the MCT code, contains new l

j models for fission product heating and natural circulation. It must be noted that these models were developed by P. P. Bioniarz and have not been extensively evaluated. The natural circula-tion flows are one-dimensional and driven by temperature  :

l differences between connecting volumes. Radiological releases  !

from core / concrete interactions were calculated using the VANESA  !

code. The thermohydraulic loads due to the direct heating event i were calculated using a small stand-alone code backed up by hand I

calculations. Both the code and the hand calculations used an I adiabatic model in which the core debris was brought into i thermal equilibrium with the atmosphere. The aerosol sources I for direct heating were obtained from results of the HIPS i experiments conducted at Sandia National Laboratories. Heatup

! and degassing of the concrete floors in the containment after '

l deposition of the debris from the direct heating event were also calculated using a stand-alone code. The various sources, both thermohydraulic and radiological, were supplied to the CONTAIN code which calculated the containment loading, the transport and j deposition of fission products in containment, and the release i of radiological sources to the environment. 6 4

! ACCIDENT SEQUENCES l 4

Two basic scenarios were examined in the study. The first l l

l scenario consisted of a TMLB' sequence in which temperature ,

induced failure of the primary system did not occur. The primary system eventually failed at instrumentation penetrations in the lower vessel head due to the thermal attack of molten core debris. Since a previous breach in the primary system '

boundary did not occur (except for normal operation of the

SRVs), the system was at high pressure (SRV set point, about j 2400 psia) and the potential existed for a direct heating l event. For this scenario a matrix of cases was analyzed in i

which selected direct heating parameters were varied. The i primary system radiological releases were also varied by using ,

l MCT code predicted releases and exercising selected combinations  ;

of the models in that code (i.e. fission product heating, natural circulation, reevolution of volatiles). Some cases were also analyzed using BMI-2104 zion releases from the primary ,

system ratioed up according to core power.

j l PW FT l

9

- m Tho occcnd cc0nnrio occum:d a TMLB' with failuro of tho seals in the primary system coolant pumps (pump seal LOCA). In this scenario all four pump seals were failed at the outset of the accident, and the system was at a somewhat lower pressure at the time of vessel breach. Thus, the direct heating scenario, if it occurred, would be of a smaller magnitude. For this analysis it was assumed that a direct' heating event did not occur in the pump seal ICCA scenario.

RESULTS AND CONCLUSIONS DIRECT HEATING  :

For the direct heating calculations a number of parameters were varied including the fraction of core injected into the containment, the quantities of hydrogen and molten steel burned during the event (all the injected zirconium was burned), and

• the amount of reactor cavity water involved in the process.

For the case in which 90% of the core debris was involved, the peak containment pressures were in excess of the estimated failure pressure range for the Bellefonte containment (144.7 to ,

153.7 psia).

With 50% core debris involvement the peak pressures were marginally close to the estimated containment failure criterion. Without the involvement of hydrogen and steel combustion, pressures were somewhat below the failure criterion while inclusion of hydrogen and steel combustion resulted in i pressures at or slightly above the failure criterion.

However, for the more probable scenarios in which cavity water was involved, complete combustion of the steel and the hydrogen (from the in-vessel oxidation of the cladding) was required to yield a peak pressur,n high enough to challenge the containment.  ;

The results of the direct heating calculations were used to postulate the con *:ainment failure mode for the radiological  !

source calculations. The 90% injection case was assumed to fail containment (a 7 sq. ft. hole) at the time of the direct heating event. The 50% injection case, since it was marginally close to i

failure, was assumed to stress the containment sufficiently to induce a relatively small leakage pathway.

PRIMARY SYSTEM RADIOLOGICAL SOURCES For the high pressure cases using only the fission product heating model the meltdown phase releases (up to and including the releases at vessel failure, Table I-type 'A' releases) of CsI and CsOH were slightly higher (by a factor of 1.5) than those predicted by BMI-2104. The Te releases were higher by a factor of 6 primarily due to differences in the March code-options that affected the oxidation of zirconium in the meltdown phase and resulted in higher Te releases from the core.

Although the fission product heat-up model diddhot produce significant changes in predicted releases during the meltdown i

(  :

I _ __ _ _ .- _ . - . _ _ _ . _ _ _ . _ - _ . _ _ _ _ . _ _ _ _ - _ - - . _ _.

an s s

, \

pheco, tho long torm h0ct-up offecto cubsequsnt to vaccol failure produced significant late reevolution of volatile fission products (Table I, type 'C' releases). For the leakage cases involving a gradual containment depressurization, a mechanism existed for moving revolatilized fission products out of the breached vessel, and with these additional releases included, the total releases of CsI and Cs0H from the RCS exceeded the BMI-2104 estimates by a factor of 10.

Implementation of the natural convection model for the meltdown phase releases (without reevolution, type 'C' releases) produced RCS releases higher than BMI-2104 by a factor of 15 for CsI and CsoH and a factor of 4.5 for Te (approx. 30% for CsI and Cs0H and 17% for Te of the original inventories of these species). 3 Although performance of the rather expensive computer i calculations for long term reevolution using the natural l circulation models after vessel breach was beyond the scope of this analysis, the releases could be still higher than those quoted above.

There were no comparable BMI-2104 calculations against which to compare the pump seal LOCA calculated releases.

Without natural convection the primary system volatile releases for this scenario were on the order of lot of the original inventories (Table I, type 'A' releases). With natural convection the predicted RCS releases of CsI and CsoH were very high, on the order of 80 or 90% and To was about 13% (type 'B' releases). However, since direct heating was not involved in I

this sequence, there was no early containment failure and very ,

small containment releases.

Also of significanca was the predicted primary system temperature history. Hot-leg piping temperatures on the order of 1700 dag. F were calculated prior to core slump. It should l be emphasized that the natural circulation calculations employed a rather crude one-dimensional model which represents a first l cut at qualitatively accounting for this phenomenon. The r complicated geometry of the primary system may not be amenable l to so simplistic a treatment, and it seems likely that the  !

primary system temperatures predicted by this.model are somewhat l However, the calculated temperature at the outlet nozzle high. '

would be within the regime in which failure could be induced, even if the actual temperatures were several hundred degrees lower. If the hot leg piping failed in this mode soon enough before core slump to depressurize the primary system, the direct ,

heating scenario could be precluded. i CONTAINMENT RETVASES A general description of the containment release calculations are given in Tables II and III and theRadiological releases forreleases these cases from are summarized in Tables IV and V.

the containment are heavily dependent on the containment failure mode. Direct heating calculations indicate that for core debris involvement greater than about 504, gross containment failure is DRAFT 1

likely in the Bellefonte containment. The 904 injection cases were, therefore, assumed to rupture the containment (7 sq. ft.

hole), and the releases of the volatiles, CsI and Cs0H were on the order of 1.5% of the initial inventories (Table IV, Cases 000 to 011A). Since the primary system releases of these materials were a factor of 10 higher when natural convection was modeled, the releases of these materials from the containment were also proportionally higher for this case, about 15%

(Case-0115). Te releases Releases were on the order of 20 to 30% while for the latter two fission product Ru was about 40%.

groups (Te, Ru) were strongly enhanced by an oxidation reaction ,

that occurred during the direct heating event. Specifically, l the species Te, Ru, and Mo formed volatile oxides during the  ;

direct heating event and were recondensed as aerosols.

Containment failure at the precise point of maximum fission

. product concentration in the containment made these cases the -

most severe in terms of releases to the environment.

For the 50% direct heating cases (Table IV, Cases 002 to 112A), it was assumed that containment leakage pathways open at the time of vessel failure. Areas of 5 sq. in. and 12 sq. in, were used to estimate releases for these scenarios. For both of the leak areas used the containment was gradually depressurized, and since much more time was available for aerosol removal the releases were significantly lower. For the worst case (Case-112A), in which reevolution of volatiles was calculated and a 12 sq. in. leak area was used, releases of the volatiles and Ru were on the order of 2% of inventory. These releases were about a factor of 12 higher than those calculated by assuming BMI-2104 releases from the primary system (Case-102A).

For the TMLB' scenarios without direct heating (Cases 000,100) the induced leakage area was varied between 1 sq. in, and 2.4 sq. in., and the releases for these cases were very small, on the order of 1.0E-4 of the original inventories of CsI  ;

and Cs0H.

The pump seal LOCA scenarios, since they do not involve  :

direct heating, also yielded relatively small releases, on the order of 1.0E-4 for CsI and Cs0H (Table V, Cases 030,130). Primary system releases calculated with natural convection, however, increased the containment releases to about 0.2% for CsI and Cs0H and to about 0.8% for Te (Table V, Cases 031,131).

There are several conclusions that can be drawn from this study. One of the most significant is that, if direct heating  ;

occurs in which more than about 50% of theIncore debris is involved, containment failure is likely. addition, this l scenario fails the containment at a time when the fission ,

product concentration in the containment atmosphere is Improved maximized and leads to high releases outside the containment. t modeling in the primary system fission product transport codes ,

has significantly increased the predicted rolesbes into the .

containment due to the added releases from reevolved volatiles and to the enhancement of fission product transport due to  ;

DRAFT natural convection heating of the primary system. However, the same modeling improvements have lead to an indication that primary system structural temperatures may be high enough to induce failure and depressurization of the primary system before core slump, thus eliminating the direct heating event and with it the primary mechanism for early containment failure for this l accident sequence.

l l

4

.I\

TABLE I PRIMARY SYSTEM RELEASES +

MCT WITHOUT

- NATURAL CONVECTION

- - m MCT MELTDOWN REEVOLUTION W/ NATURAL FISSION RMI-2104 PHASE ** PHASE *** TOTAL CONVECTION SCENARIO PRODUCI TYPE A TYPE C TYPE B

.02* .028 .21 .24 .35 TMLB'-HPE - CSI

.02* .030 .18 .21 .28 CSOH

.039' .24 -- .24 .17 TE t

l

~

-- .084 --

.084 .93 TMLB'-PSL CSI CSOH --

.073 --

.073 .77

.089 -- .089 .13 TE

+ FRACTIONS OF INITIAL INVENTORIES

• ZION STUDY

• • UP TO AND INCLUDING VESSEL FAILURE

• • RELEASES SUBSEQUENT TO VESSEL FAILURE - CALCULATED USING THE DEPRESSURIZATION HISTORY FOR CONTAINMENT LEAKAGE SCENARIOS i AFT i

I

TABLE II CASE MATRIX.

l i TMLB' HIGH PRESSURE EJECTION l

CONTAIN. (VANESA) 1 CAVITY BREAK P.S. F.P. NAT. CORE / CONC AREA RELEASES HEATING CONV. REEVOLUTION RELEASES CASE # INJECTION WATER j

IIN BMI-2104 NO NO NO YES 000 0 --

2.4IN BMI-2104 NO N0 N0 YES 100 0 --

2 BMI-2104 NO NO NO NO 001 90 NO 7FT 2 MCT YES NO NO NO I 011 90 NO 7FT 7FT BMI-2104 NO NO N0 NO

' 001A 90 YES 2 MCT YES NO NO N0 011A 90 YES 7FT -

l 2 NCT YES YES NO NO i 0118 90 YES 7FT 1

NO N 002 50 50 NO NO SIN 12IN BMI-2104 BMI-2104 NO NO NO NO NO NO NO q I 102 2 YES NO YES N0 012 50 NO SIN MCT 12IN MCT YES NO YES NO 112 50 N0 SIN BMI-2104 NO NO NO NO

002A 50 YES 12IN BMI-2104 NO NO NO NO 102A 50 YES 012A 50 YES sit;2 MCT YES NO YES YES(LATE) 112A 50 YES 12IN2 MCT YES NO YES YES(LATE) i

~

TABLE III .

CASE MATRIX f.

TMLB' PUMP SEAL LOCA CONTAIN. (VANESA)

I CAVITY BREAK P.S. F.P. NAT. CORE / CONC AREA RELEASES HEATING CONV. REEVOLUTION RELEASES CASE # INJECTION WATER IIN MCT YES NO N0 YES

030 0 --

j 2.4IN MCT YES NO N0 YES 130 0 --

IIN HCT YES YES NO YES 031 0 --

g7.y 2.4IN MCT YES YES NO YES y . ,[

I 131 0 --

('j s.

III j

Dk k

.i I

DRAyr w OON C to sr! rs O rs c3 CC O O

• O .-e

hcl O .-4. N. rN. rs. FN. FN. Po. . LO ro (c. P o. L C. r o. W.

4 I

e CD w O O rs O to wl es CC rs IN CC C) C') .-e XI Ce .-*e rs. FN. ts. ts rs. to L0. to. (c. en. on. to. to.

e .

LO LO LO LO LO t Ln i

LO t

Lo s

N t% NN to CO CO t i C e to ts ontoto 7 7W W W W W W W W WI O O O O O O O O. 7. FN. Co. CO. O. Lc .Ln.

JJ O. O. O. O. O O. O. 7 O Po CO eq CO to CO LO 7 m7

^ CO 7 ro N FN I e .-e LO 8 I M

g 7 'to to .T. Ln 7 LO rs .-4 N W W w NW W ZZ

• El O .-4 N ON OOO O LO. O. O O Lo. ro.

gg o C >l O. O. O. O. O O. O O. O. CO N O. O. CO N N& '

O O t.ht b [T N mN CO w mw an

, W cl WW

. *. C1 C3 C) CO O .-4 .-4 Nw N w N w N i

> g E e- y 3 El N7 to. ro eq 7 7. O. O. O O. O. O. O. O.

. . . .s g g -

d4W O*I s

N h jLa E CO to 1%

O N O.-4ro CO CO OO*  !

3 -- gg wl -e 7 O Po O O -e O N N O N ON C">

W g .-e )

Hj N. to N O. O. O. O. O. O. O. O.

gg y3 O. O. N . . . .

W -

O ,

f b 7 7 W <

&f H E O m7 s I W W 7 (.O 7 (D W CO LO

.-e 60 NW I ts N

.-4 C1 Z

H g

g O m CO. Po.

.-4 .-*.-4 N O. O O

.-e O. O. O. t% O. O. O.

.-4 N. OON y O C) N .-4O.O. O. O. . CO <

)

H

' H s H Z

7 7 H s,n 7 .

I I CO .-4 8 FN

• 8 .

La (O r0 W w ON l.L.

H WW 7 W w7 th s .-4 O

m .-4--*

(o O. O. O O 4 N. OwN O ts.

sn 19. eO. O. O. ^. O.

e co . s 'O. O. ts O. O. p. Z O

H

- >=-

O W

< 4 CD < <4< <

m O O -e .-e .-4 . * ' w-e N N N N NN x

< O O O .-* O .-*.-* O O .-e N. N OO .-e a s.L.

e j

O O .-4 OOOOOOw Ow O .-e O*

l .,

D .

• N.1 1 s

l *.

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

e  %

O N a:t N CO N tr M1 .-e .-e O. . O. .

e O N wl fN CD fN (D XI .-e .-e C. . O. .

D ee 1 O I O wNwN

$ O 8.=O 8 C/3 E O CD h< O CEl Col 7 N tn N O .-* O O. O. O. O.

.-4 oC g =d w

I

• 4

> 03

$w0 (O, to wwww

. 4, (O ,

-N-O pg, H 'cn N CO 7 N ZJ W M f% M CO O

E wl QOOO w kW g

H R

. O. O. O.

is a H i 2 g O 7 77 w }

O i e i en > .

I w w w .-e Z H

l O N . CD. - .O tn O .-* N CO O. J l

• C H

H H

Z 477 H .

I e 8 O H wwwN t.4-in .-e O O . 7 CO. O i t

LO e CO O. Z O

H d W

O w *C co OO .-e w  %

<[ M MM M La.

O O .-e C .-e e

~ -

Am

{

e' l l_ -. _-- .,. ._ _ ., . - _ - - 1