ML20106A469

Requests Commission Notification of Encl Info Re Severe Accident Considerations in Pwrs.Staff Evaluation of Info Should Be Complete within Several Months Depending on Extent of Analysis Necessary

ML20106A469
Person / Time
Site:	Indian Point
Issue date:	08/24/1984
From:	Speis T Office of Nuclear Reactor Regulation
To:	Eisenhut D Office of Nuclear Reactor Regulation
Shared Package
ML20106A471	List: ML20106A469
References
RTR-NUREG-CR-3369 NUDOCS 8409070328
Download: ML20106A469 (131)

Text

.

4 UNITED STATES 7,

NUCLEAR REGULATORY COMMISSION

{

I wasmuciow. o. c.zosos s

y August 24, 1984 s

MEMORANDUM FOR: Darrell Eisenhut. Director Division of Licensing FROM:

Themis P. Speis, Director Division of Safety Technology

SUBJECT:

BOARD NOTIFICATION The purpose of this memorandum is to request that you notify the Comissior.

in connection with the Indian Point Hearing and any other licensing boards

associated with severe accident considerations in PWRs of new and possibly relevant information which has recently come to our attention. A description of this information is provided in the enclosure.

The staff is evaluating this infonnation to determine its safety signifi-cance and relevance.

In particular, we are evaluating how the new infor-mation affects our assessments of risk associated with core melt and early containment failure. We anticipate completing our evaluation within several months, depending on the extent of analysis necessary.

M

[

Themis P. Speis, Director Division of Safety Technology

Enclosure:

As stated cc:

H. Denton S. Varga t

G. Lainas R. Vollmer l

R. Bernero j

W. Houston J. Rosenthal R. Minogue, RES

0. Bassett, RES D. Ross, RES F. Rowsome J. Moore, ELD

. p a r 2 m === m 7,.,

L gy v,3 Y

~m w=w a===m wn 1

BACKGROUND Under some conditions a core melt accident may proceed such that a rapid transfer of energy from the melted core to the liquid coolant (water) might The liquid coolant could then rapidly boil producing large amounts occur.

of steam in what is known as a steam explosion.

To result in a significant safety concern the interaction must be very rapid I

(millisecond time scale) and must involve a large amount of the melted core.'.

' and coolant in certain spatial configurations.

If such events were to take '

place within the reactor pressure vessel (RPV) missiles could be generated which might penetrate the containment and allow early release of fission products. A missile capable of penetrating the containment must have con-siderable mass and velocity on impact. To simplify their calculations some analysts conservatively assume that, given a sufficiently large steam explosion within the RPV, the vessel top head bolts fail and the top head itself becomes the missile.

For this to occur a dense slug of material must be accelerated by the steam explosion within the RPV and it must impact the upper vessel head fairly coherently after traversing the structures above the core region. The head bolts must then fail essentially simultaneously.

In some cases even if the head could become a missile it must itself often penetrate a missile barrier located above the vessel. Thus many factors can influence both the likelihood of generating a missile and, given the genera-O tion of a missile, its characteristics, i.e., mass and velocity.

Clearly many factors are involved in estimating a probability for containment failure other than the probability of an energetic steam explosion. However, 1

considerable attention has been given to the steam explosion phenomenon.

There is a substantial amount of experimental data on certain aspects of the interaction of molten metals with cooler liquids notably on the pressure histories, final particle size distributions and energy conversion, estimates 4.

n.-

i

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

. for such interactions. Less is known about the detailed physics sf the mixing process itself (e.g., hydrodynamic breakup, droplet fluidi'zat40n, geometric effects, etc.) especially as affected by scale and proto-typicality. Thus one must, in general, apply analytical techniques (bench-marked to experiment where possible) to estimate the potential consequences of such events in the case of severe nuclear reactor accidents.

The calculation of such interactions involve a large number of parameters each of which can take on a range of values. Evaluation of the magnitude an'd-

' likelihood of such events has been the subject of considerable effort within the industry and at NRC for some time.

In the past the staff has concluded that the likelihood of such events is small enough that they need not be con sidered further in the overall evaluation of risk.

The staff has recently been informed of a new analysis in this area by the Sandia National Laboratory (SNL).

In this analysis the authors conclude that the uncertainties associated with estimates of the pctential consequences from molten fuel-coolant interactions is large. The SNL analysis is presently being evaluated by the staff.

PROBLEM The new analysis by SNL (a copy is attached as Enclosure 1) represents an attempt by the authors to systematically assess uncertainties in the evalu-ation of the potential for early containment failure from molten fuel-coolant interactions. The report presents a model of the entire process beginning with the steam explosion phenomenon and progresses through acceleration of a slug of material to the impact of the slug on the RPV upper head. The analysis goes on to model the failure of the upper head and its behaviour as a missile and the potential for the head to penetrate the containment.

k e

t' 4 The SNL study divides the relevant parameters in their model into.three ranges of values covering the spectrum of physical possibility.

It then* assigns relative likelihoods to each of these ranges. The authors assume a unifonn distribution in all cases. That is, all values are considered equally likely. The analysis then utilizes a statistical (Monte Carlo) sampling of the parameter space for selection of the values to be used in their model for estimating the potential consequences frem such events.

In this report the SANDIA investigators conclude that "tne conditional probability of containment failure, given core melt during a low pressure accident, is extremely uncertain". They then go on to say that "Indeed the results span the range of probability from 0 to 1".

It is also pointed out in this study that "this uncertainty estimate (i.e., the range of 0 to 1) is derived from the particular choice of distributions and combinations thereof used".

Based upon their assumption of uniform distributions in the lowe;, middle and upper uncertainty ranges the containment failure probabilities estimated for these three ranges are as follows:

lower range, P=0 middle range, P=10-4 upper. range, P=1 The authors caution that the middle range result, i.e., the 10~4 value,

'~[

"should not be used as a best estimate of the fraction of core melt accidents leading to containment failure by steam explosions".

These results are at variance with previous studies. Among these studies are the conclusions presented in WASH-1400 and, more recently, in the staff's Zion / Indian Point study reported in NUREG-0850. Both of these documents provide more limited ranges of conditional probabilities. The WASH-1400 i

study adopted a range for this conditional probability of 10-I to,10'4 with a 4

e e

r

~. -

. median value of 10-2 In the Zion / Indian Point study the staff cgncluded that the probability of a steam explosion induced failure of containment was at least two orders of magnitude lower than the 10-2 median value given in WASH-1400.

The authors of the SNL report identify. number of other studies which have, for the most par t, accepted the concept of a narrower range of conditional probabilities. These include the German Risk Study, the UKAEA PWR Degraded Core Analysis Report, the Report of the Swedish Government Comittee on Steam

' Explosions and studies by Fauske & Associates, Inc., Theofanous and Saito, Swenson and Corradini and Mayinger among others.

In its report on the potential consequences from core melt accidents in the Zion and Indian Point facilities (NUREG-0850, Volume 1, Preliminary Assess-ment of Core Melt Accidents at the Zion and Indian Point Nuclear Power Plants and Strategies for Mitigating Their Effects, November 1981) the staff adopted the analysis of Professors T. G. Theofanous and M. Saito of Purdue University to provide the basis for its conclusion that steam explosions need not be considered for purposes of evaluating accident mitigation requirements.

That analysis is reported in NUREG/CR-2318, LWR and HTGR Coolant Dynamics:

The Containmen't of Severe Accidents, October 1982 by T. G. Theofanous and M.

Saito of Purdue University. The approach used by these investigators is based largely on an evaluation of physical limitations in the fuel-coolant mixing process. A presentation on the subject was given to the ACRS by Professor Theofanous on January 11, 1984. A copy of his presentation is included as Enclosure 2.

The subject of steam explosions has also been part of the review currently in progress within the severe accident study program.

It was one of the subjects at a meeting between the NRC and industry's degraded core task I

e l

1

. (IDCOR) at Harpers Ferry, West Virginia on November 29 - December 1,1983.

I Presentations on this subject from various parties were discussed. Copies of the consensus reached by the participants are included as Enclosure 3.

While general agreement was reached on broad areas of the subject several details of the assessment of this phenomena remain to be agreed upon. In particular it was agreed that steam explosions can occur under appropriate initial and boundary conditions but that such details as the quantities of materials that can participate in the interaction, the propagation and conversion efficiency

, of the interaction, geometry and scaling require further consideration.

It ~?

'was also agreed that steam explosions large enough to fail containment are deemed unlikely but have not been demonstrated to be impossible.

The staff is evaluating information pertinent to these considerations to determine how it affects the assessment of risk associated with core melt and early contaimaent failure. This evaluation is expected to take several months.

l i

,=_A[

l

~

ENCLOSURE 1 HUREG/CR-3369 AN UNCERTAINTY STUDY OF PWR STEAM. EXPLOSIONS G

e 5

P

• m e, 9

L

=

Gn

WURIEG/CR-3369 SAND 83-1438 R1 Printed May 1984

~.

An Uncertainty Study of PWR Steam Explosions M. Berman, D. V, Swenson, A. J. Wckett Notenal Laboratores t

AtuperQue, New Meaco 87185 and Lee. CeWome 94550 l

for the thted States Decertnent of Energy under coneract DE-AC04 76DP00789 1

Prepared for U. S. NUCLEAR REGULATORY COMMISSION

a e

~

NOTICC This report was prepared as an account of work sponsor?d by an agency of the Onited Semese Government. Neither the Uniend States Government nor any agency thereof, or any of their em.

l P eyees, makes any warranty,"espressed or implied, or assumes anylegalliabilityor. f any informeelen, appasseus product or ity forany elded party's tese,or the results of such use, o process disclosed in this report, er sopresene that its use by such third party would not infringe privately owned rights.

Available from CPO Sales Program Divis6on of Technical Information and Document Control U.S. Nuclear Regulatory Commission Washington D.C. 20555 and National Technical Informanon Service Springfield. Virginia 22161 6

C

NUREG/CR-3369 SAND 83-1438 i

R1 j

g s

AN UNCERTAINTY STUDY OF PNR STEAM EXPLOSIONS M. Berman, D. V. Svenson* and A. J. Nickett+

MAY 1984 Sandia National Laboratories Albuquerque, NM 87185 Operated by Sandia Corporation for the U. S. Department of Energy

~

i '

Prepared for Division of Accident Evaluation i

Office of Nuclear Regulatory Research U. S. Nuclear Regulatory Commission Nashington, DC 20555 Under Memorandum of Understanding DOE 40-550-75 NRC FIN No. A1030

• Present address:

School of Civil and Environmental Engineering, Cornell University, Hollister Hall, Ithaca, NY 14853 t

+0n attachment from United Kingdom Atomic Energy Aut'llWr ity.

Safety and Reliability Directorate 1

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

._~___.__.__.____m_,

Abstract Some previous assessments of the probability of conhinnent failure caused by in-vessel steam explosions in a PWR have recognized large uncertainties and assigned broad ranges to the probability, while others have concluded that the probability is small or zero.

In this report we study the uncertainty in the probability of containment failure by combining the uncertainties in the component physical processes using a Monte Carlo method.

We conclude that, despite substantial research, the combined uncertainty is still large.

Some areas are identified in which improvements in our understanding may lead to large reductions in the overall uncertainty.

i

~;

l ee iii/iv

CONTENTS Section Page Summary 1

1.

Introduction 3

1.1 Background

3 1.2 Previous Assessments 4

1.3 Aims of this Report 6

2,, Methods 7

2.1 Uncertainty and Sensitivity Analysis 7

2.2 Monte Carlo Method 7

3.

Modeling Assumptions 10 3.1 General 10 3.2 Fraction of Core Molten 11 3.3 Pour Diameter 12 3.4 Pour Length (or Trigger Time) 15 3.5 Mixing Limitations 17 3.6 Fraction of Water that Mixes 19 3.7 Conversion Ratio 19

' 0.1 3.8 Heat Content of Molten Fuel 22 3.9 Fraction of Remaining Melt Above Explosion 22 3.10 Fraction of Remaining Water Above Explosion 22 3.11 Energy Dissipation by Botton Failure 23 3.12 Slug Composition

~*

23 3.13 Energy Dissipation by Core and Upper Internal 24 Structure 3.14 Slug Impact Model 27 3.15 Containment Failure 28 Y

CONTENTS (cont'd)

Section

~

Page 3.16 Summary of Modeling 30 4.

Calculations and Results 33 4.1 Outline of Calculations 33 4.2 Results of Main Calculations 34 4.3 Results of Additional calculations 38 5.

Other Areas of Uncertainty 42 5.1 The Effects of High Pressure 42 5.2 Uncertainty in Head Becoming Missile 43 5.3 Multidimensional and Geometric Effects 44 5.4 The Effect of Correlations 45 5.5 Effects of Model Parameterization 46 6.

Discussion 49 References El Appendix A:

Subjective Probability A-1

'.2.

Appendix B:

Finite Element Calculation of Vessel B-1 Failure Introduction B-2 Material Properties B-2 Failure Criteria B-3 Numerical Model B-4 Loading Conditions g

B-S e

Results 4:

B-5 Summary B-12 vi

o UNITED STATES

', ^

NUCLEAR REGULATORY COMMISSION n

r WASHINGTON, D. C. 20555

/

August 24, 1984 MEMORANDUM FOR: Darrell Eisenhut Director Division of Licensing FROM:

Themis P. Speis, Director Division of Safety Technology

SUBJECT:

BOARD NOTIFICATION The purpose of this memorandum is to request that you notify the Commission in connection with the Indian Point Hearing and any other licensing boards

. associated with severe accident considerations in PWRs of new and possibly relevant information which has recently come to our attention. A description of this information is provided in the enclosure.

~

The staff is evaluating this information to detemine its safety signiff-cance and rr.levance.

In particular, we are evaluating how the new infor-mation affects our assessments of risk associated with core melt and early containment failure. We anticipate completing our evaluation within several months, depending on the extent of analysis necessary.

Aow 1 yc^~^

Themis P. Speis, Director Division of Safety Technology

Enclosure:

As stated cc:

H. Denton S. Varga G. i.ainas R. Vollmer i

R. Bernero W. Houston l

J. Rosenthal l

R. Minogue, RES t

0. Bassett, RES D. Ross, RES F. Rowsome J. Moore, ELD

[

e l

l L

a

. BACKGROUND Under some conditions a core melt accident may proceed such that a rapid transfer of energy from the melted core to the liquid coolant (water) might occur. The liouid coolant could then rapidly boil producing large amounts of steam in what is known as a steam explosion.

To result in a significant safety concern the interaction must be very rapid (millisecond time scale) and must involve a large amount of the melted core,

and coolant in certain spatial configurations.

If such events were to take '

place within the reactor pressure vessel (RPV) missiles could be generated which might penetrate the containment and allow early release of fission products. A missile capable of penetrating the containment must have con-siderable mass and velocity on impact. To simplify their calculations some analysts conservatively assume that, given a sufficiently large steam explosion within the RPV, the vessel top head bolts fail and the top head itself becomes the missile. For this to occur a dense slug of material must be accelerated by the steam explosion within the RPV and it must impact the upper vessel head fairly coherently after traversing the structures above the core region. The head bolts must then fail essentially simultaneously.

In some cases even if the head could become a missile it must itself often" penetrate a missile barrier located above the vessel. Thus many factors can influence both the likelihood of generating a missile and, given the genera-tion of a missile, its characteristics, i.e., mass and velocity.

Clearly many factors are involved in estimating a probability for containment l

failure other than the probability of an energetic steam explosion. However, 1

l considerable attention has been given to the steam explosion phenomenon.

There is a substantial amount of experimental data on certain aspects of the l

interaction of molten metals with cooler liquids notably on the pressure I

histories, final, particle size distributions and energy conversion estimates i

t l

e.

l l

. for such interactions.

Less is known about the detailed physics of the mixing process itself (e.g., hydrodynamic breakup, droplet fluidt2ation, geometric effects, etc.) especially as affected by scale and proto-typicality. Thus one must, in general, apply analytical techniques (bench-marked to experiment where possible) to estimate the potential consequences of such events in the case of severe nuclear reactor accidents.

The calculation of such interactions involve a large number of parameters each of which can take on a range of values.

Evaluation of the magnitude an'd

' likelihood of such events has been the subject of considerable effort within the industry and at NRC for some time.

In the past the staff has concluded that the likelihood of such events is small encugh that they need not be con sidered further in the overall evaluation of risk.

The staff has recently been informed of a new analysis in this area by the Sandia National Laboratory (SHL).

In this analysis the authors conclude that the uncertainties associated with estimates of the potential consequences from molten fuel-coolant interactions is large. The SNL analysis is presently being evaluateo by the staff.

f PROBLEM The new analysis by SNL (a copy is attached as Enclosure 1) represents an

'~~

attempt by the authors to systematically assess uncertainties in the evalu-ation of the potential for early containment failure from molten fuel-coolant interactions. The report presents a model of the entire process beginning with the steam explosion phenomenon and progresses through acceleration of a slug of material to the impact of the slug on the RPV upper head. The analysis goes on to model the failure of the upper head and its behaviour as a missile and the potential for the head to penetrate the containment.

f n=

l

. The SNL study divides the relevant parameters in their model into.three ranges of values covering the spectrum of physical possibility.

It then* assigns relative likelihoods to each of these ranges. The authors assume a uniform distribution in all cases. That is, all values are considered equally likely. The analysis then utilizes a statistical (Monte Carlo) sampling of the parameter space for selection of the values to be used in their model for estimating the potential consequences from such events.

In this report the SANDIA investigators conclude that "the conditional probability of containment failure, given core melt during a low pressure accident, is extremely uncertain". They then go on to say that "Indeed the results span the range of probability from 0 to 1".

It is also pointed out in this

~

study that "this uncertainty estimate (i.e., the range of 0 to 1) is derived from the particular choice of distributions and combinations thereof used".

Based upon their assumption of uniform distributions in the lower, middle and upper uncertainty ranges the containment failure probabilities estimated for these three ranges are as follows:

Icwer range, P=0 middle range, P=10-4 upper range, P=1 The authors caution that the middle range result, i.e., the 10-4 value,

'~'

"should not be used as a best estimate of the fraction of core melt accidents leading to containment failure by steam explosions".

These results are at variance with previous studies. Among these studies are the conclusions presented in WASH-1400 and, more recently, in the staff's Zion / Indian Point study reported in NUREG-0850. Both of these documents provide more limited ranges of conditional probabilities. The WASN-1400 study adopted a range for this. conditional probability of 10'I to,10 with a d

c":

4

' 4 median value of 10-2 In the Zion / Indian Point study the staff concluded that the probability of a steam explosion induced failure of containment was at least two orders of magnitude lower than the 10-2 median value giv'en in

~

WASH-1400.

The authors of the SNL report identify a number of other studies which have, l

for the most part, accepted the concept of a narrower range of conditional probabilities. These include the German Risk Study, the UKAEA PWR Degraded Core Analysis Report, the Report of the Swedish Government Committee on Steam

. Explosions and studies by Fauske & Associates, Inc., Theofanous and Saito, l

Swenson and Corradini and Mayinger among others.

1 In its report on the potential consequences from core melt accidents in the ZionandIndianPointfacilities(NUREG-0850, Volume 1,PreliminaryAssess-ment of Core Melt Accidents at the Zion and Indian Point Nuclear Power Plants and Strategies for Mitigating Their Effects, November 1981) the staff f

adopted the analysis of Professors T. G. Theofanous and M. Saito of Purdue University to provide the basis for its conclusion that' steam explosions need not be considered for purposes of evaluating accident mitigation requirements.

f That analysis is reported in NUREG/CR-2318, LWR and HTGR Coolant Dynamics:

The Containment of Severe Accidents, October 1982 by T. G. Theofanous and M.

Saito of Purdue University. The approach used by these investigators is based largely on an evaluation of physical limitations in the fuel-coolant l.

mixing process. -A presentation on the subject was given to the ACRS by j

Professor Theofanous on January 11, 1984. A copy of his presentation is included as Enclosure 2.

i The subject of steam explosions has also been part of the review currently in progress within the severe accident study program.

It was one of the subjects at a meeting between the NRC and industry's degraded core task c=

l l

l

(IDCOR) at Harpers Ferry, West Virginia on November 29 - December,1,1983.

Presentations on this subject from various parties were discussed. Copies of the consensus reached by the participants are included as Enclosure 3.

While general agreement was reached on broad areas of the subject several details of the assessment of this phenomena remain to be agreed upon. In particular it was agreed that steam explosions can occur under appropriate initial and boundary conditions but that such details as the quantities of materials that can participate in the interaction, the propagation and conversion efficiency of the interaction, geometry and scaling require further consideration.

I t '.

' was also agreed that steam explosions large enough to fail containment are deemed unlikely but have not been demonstrated to be impossible.

The staff is evaluating information pertinent to these considerations to determine how it affects the assessment of risk associated with core melt and early containment failure. This evaluation is expected to take several months.

• @g i

e i

m.

e

a.-

_a_aop e

o ENCLOSURE 1 NUREG/CR-3369 AN UNCERTAINTY STUDY OF PWR STEAM EXPLOSIONS I

c=

i k

T NUREG/CR-3369 SAND 83-1438 R1 Printed May 1984 An Uncertainty Study of PWR Steam Explosions M. Berman, D. V. Swenson, A. J. Wickett S E 7 ion.it.o..i-,..

,T3.2."'#l.'tTO:,'M"""' "'"""

under Contr.ct DE-AC04 760P00789

-~

n.

Prepared for U. S. NUCLEAR REGULATORY COMMISSION

. -=

~

NOTICE This report was prepared as an account of work sponsored by an agency of the United States Government. Neitner the United States Covernment nor any agency thereof, or any of their em.

ployees, makes any warranty, ity forany third party's use,or the any legalliability or responsipe.,expreened or implied, or assumes l

results of such use, et any information, apparatus product or process disclosed in this report, or reprueenes that its use by su:h third party would not ininnge privately owned rights.

Available from GPO Sales Program Division of Technical Information and Document Control U.S. Nuclear Regulatory Commission Washington, D.C. 20555 and Nat6onal Technical Information Service Springfield, Virginia 22161 e

I C

3 Q

NU SAND 83-143 REG

\\

69

\\

AN R1 8

UN

\\

CERTAINTY

\\

STUDY OF PWR STEAM EXPLOSION M. Be rman. D.

V. Swenso*ad n

n A. J.

\\

MAY Wicke t t +

1984 Sandia Nati Albuqu ona erqu,l Labo e

Sandia Cooper ted NM r torie a

a 87 by 185 s

U. S. Departfor the rpo r tion a

ment of Energy U.icDivision Prepa Off e

of rd of e

Nu lear RegAc ident Under MemorS. Nuclear R c

c for ulatory ReEvalu tion andumWa hingtonegulato s

a ry Com is ion of

,DC sea Under tandi NRC m

rch 20555 F

i s

IN No.s

• Pres L

Cor ent A1030ng DOE n ll addr 40-5 e

Univ es:

50-75 s

er ity. Scho l s

'~~

o Hollister HalCivil of

+0n Safety andatta hme t

l. a d n

Env c

Ithac,ir n

NYonmental Reliabilifrom Unite a

d 14853 Engin ty Dir Kingdom Ato

ering, e

e to c r te a

mic Energy Au

\\

~

thority.

c

\\

E-(g

-w---

w.

,m

,-----en-,--r--

Abstract Some previous assessments of the probability of containment failure caused by in-vessel steam explosions in a PWR -have recognized large uncertainties and assigned broad ranges to the probability, while others have concluded that the probability is small or zero.

In this report we study the uncertainty in the probability of containment failure by combining the uncertainties in the component physical processes using a Monte Carlo method.

We conclude that, despite substantial research, the combined uncertainty is still large.

Some areas are identified in which improvements in our understanding may lead to large reductions in the overall uncertainty.

f

]

E e

t lii/iv

CONTENTS Section Page Summary 1

1.

Introductio11 3

1.1 Background

3 1.2 Previous Assessments 4

1.3 Aims of this Report 6

2.

Methods 7

2.1 Uncertainty and Sensitivity Analysis 7

2.2 Monte Carlo Method 7

3.

Modeling Assumptions 10 3.1 General 10 3.2 Fraction of Core Molten 11 3.3 Pour Diameter 12 3.4 Pour Length (or Trigger Time) 15 3.5 Mixing Limitations 17 3.6 Fraction of Water that Mixes 19 3.7 Conversion Ratio 19 3.8 Heat Content of Molten Fuel 22 3.9 Fraction of Remaining Melt Above Explosion 22 3.10 Fraction of Remaining Water Above Explosion 22 3.11 Energy Dissipation by Botton Failure 23 3.12 Slug Composition 23 3.13 Energy Dissipation by Core and Upper Internal 24 Structure g

4.

3.14 Slug Impact Model 27 3.15 Containment Failure 28 V

CONTENTS (cont'd)

Section Page 3.16 Summary of Modeling 30 4.

Calculations and Results 33 4.1 Outline of Calculations 33 4.2 Results of Main Calculations 34 4.3 Results of Additional Calculations 3 r, 5.

Other Areas of Uncertainty 42 5.1 The Effects of High Pressure 42 5.2 Uncertainty in Head Becoming Missile 43 5.3 Multidimensional and Geometric Effects 44 5.4 The Effect of Correlations 45 5.5 Effects of Model Parameterization 46 6.

Discussion 49 References 51 Appendix A:

Subjective Probability i

A-1 Appendix B:

Finite Element Calculation of Vessel i

Failure B-1 l

Introduction B-2 Material Properties B-2 Failure Criteria B-3 Numerical Model B-4 Loading Conditions E -5 Results

[

I on B-5 Summary B-12 I

vi l

Illustrations Figure Title

.Page 1

Three Uniform Distributions of Fraction of Core Molten 12 2

Melt Pour into Lower Plenum by Failure of the Lower Core Plate 13 3

Melt Flow into the Lower Plenum by Sideways Penetration of the Core Barrel 14 4'

Three Uniform Distributions of Pour Diameter 16 5

Three Uniform Distributions of Pour Length 16 6

Three Uniform Distributions of Conversion Ratios 21 7

Three Uniform Distributions of Condensed-Phase Volume Fraction in Slug 24 8

Energy Dissipation in Upper Internal Structure 26 B-1 Bolt Stress Intensity Calculation B-4 B-2 Finite Element Model and Locations of Failure Evaluation B-6 B-3 Head Displacements for 80 MPa Ramp Loading I

(Magnification = 2)

.B-8 B-4 Average Axial Stress in Studs B-9 a

B-5 Average Effective Plastic Strain in Studs B-10 B-6 Average Effective Plastic Strain at Center l

of Head B-11 c.

i vii 4

l

Tables Table Title Page t

I Ranges of Poured Mass 17 II Thermodynamic Maximum Conversion Ratios 21 III Penetration Formulae 29 IV Main Calculations 35 V

Additional Calculations 39 '

B-1 Material Properties at 288'C B. 2 B-2 Loading Cases Analyzed Using Finite Element Model and Failure Evaluation B-7 I

e 4

e.

viii

1 List of Abbreviations BWR Boiling Water Reactor

~

~

CEA Commissariat a l'Energie Atomique EDF Electricite de France EPRI Electric Power Research Institute EXO-FITS FITS experiments conducted outdoors FITS Fully Instrumented Test Series LANL Los Alamos National Laboratory LWR Light Water Reactor NDRC National Defense Research Committee PWR Pressurized Water Reactor RPV Reactor Pressure Vessel UIS Upper Internal Structure UKAEA United Kingdom Atomic Energy Authority WASH-1400 Reactor Safety Study [ Reference 1]

ZIP Zion / Indian Point Study (References 2 and 3]

f b

l ne=

j ix l

I ACKNOWLEDGEMENT Many individuals with diverse backgrounds contributed to this study.

M.

L.

Corradini, R.

G.

Easterling, N.

A. Evans.. D. E.

Mitchell and J.

B.

Rivard all attended meetings of a Working Group and participated in discussions with the authors dealing with the phenomenological and statistical questions that arise in this kind of study.

Their contributions are gratefully acknowledged by the authors.

In addition to valuable assistance from reviewers within Sandia National Laboratories, we acknowledge helpful comments from R. S.

Peckover (UKAEA). D. Squarer (EPRI), and J. L. Telford (NRC) e

^

l i

4 e

t f

e.

X

Summary

\\

In the unlikely event of a core-melt accident in a PWR, molten l

core material may flow down into residual water in the lower plenum of the reactor pressure vessel, possibly resultihg in a steam explosion.

The probability that such an explosioh would be large enough to fail containment, for example by ejecting the vessel top sufficiently energetically, is of interest for probabilistic risk assessment.

Many studies have addressed this issue [1, 4-12], and have offered a variety of probability estimates covering several orders of magnitude.

One of these studies [11] investigated the probability by assigning probability distributions to various uncertain parameters in a simple model of the process, and then sampling at random in a Monte Carlo analysia.

The number of containment failures predicted by the model, out of 10,000 trials, gave the probability of containment failure.

The present study has two major goals.

The first is to provide an uncertainty estimate. for the conditional probability of direct containment failure by steam explosions (given core melt).

The second is to identify important contributors to this uncertainty, in order to provide understanding of the reasons for its magnitude and to indicate (by sensitivity studies) what additional information would be needed to reduce it.

This report offers several improvements over the previous analysis [11).

Uncertainties have been evaluated in a more complete and systematic manner.

The input distributions have also been chosen more consistently, and the uncertainty ranges have been explored in greater depth by sampling portions of the ranges with separate distributions.

Changes have been made in the modeling of the physical processes described.

Among the most important of these were:

changes in the calculation of the amount of melt that can participate in the explosion:

changes in the conversion ratio estimates; changes in the calculation of the energy absorbed by the internal structure; and changes in modeling slug impact and vessel and containment failure.

The distributions and uncertainty estimates also reflect some recent research results.

When distributions were selected from the lower thirds of the uncertainty ranges, the probability of failing both the PWR vessel lower plenum and the large, dry containment was zero.

When distributions from the middle thirds of the uncertainty ranges were used, the probability of containment failure was essentially zero (i.e.,

approximately 10-4).

The correspond-ing probabilty of failing the lower plenum of the vess was about 21%.

When distributions from the high thirds

the uncertainty ranges were used, the probability of failing both the vessel and containment was very nearly one.

The individual 1

l i

I i

l probability numbers calculated would have been different had a different parameterization of the problem been used.

However, the ranges of uncertainty calculated for these probabilities covers the entire range of possibilities.

Sensitivity studies indicated that among the most important uncer-tainties were those in the pour diameter (related to total mass of molten core mixed) and conversion ratio.

When the distribu-tions of each of these parameters were taken at the upper third of their ranges, all other parametric distributions remaining in i

the middle of their parameter ranges, the probabilities of vessel and containment failure increased significantly.

Similarly, when these parameters were sampled in the lower third of their uncer-tainty ranges, the failure probabilities were significantly reduced, even when all other parameters were sampled at the high parts of their ranges.

i

'The modeling described above refers to accidents in which the i

pressure in the reactor vessel is at or near atmospheric.

Exten-sions of these results to accidents that occur at higher ambient

~

pressures in the vessel introduces additional uncertainty.

There may well be differences in mixing, triggering, conversion ratio and vessel failure due to higher ambient pressure, but the current state of knowledge is insufficient to account for these differ-ences.

The calculations reported here assumed that strong enough loading of the reactor pressure vessel upper head would produce a large missile, rather than a more benign failure mode.

A large missile, if energetic enough, could penetrate the containment; if the explosion produced only small missiles or none at all, the large dry containment would very likely remain intact.

Thus the pres-sure vessel failure mode is another important uncertainty.

For consistency in these calculations, dimensions were taken fron l

the Zion PWRs.

The results calculated in this report may, in principle, be extended to other PWRs by accounting for differences in geometry, containment strength, missile shields, etc.

Applica-tion to BWRs entails a large uncertainty because of major plant differences.

Differences in BWR vessel geometry and strength could strongly affect the characteristics of the steam explosion.

l Differences in containment geometry and strength will also influ-t once the failure mode (penetration by missiles or overpressuriza-j tion due to rapid steam release).

n=

2

1.

Introduction

1.1 Background

Although accidents in light water reactors (LWRs) involving core melting, breach of containment, and release of radionuclides to the environment are unlikely events, they were considered'in the Reactor Safety Study of 1975 (1) and h;ve received renewed research emphasis since the accident at Three Mile Island Unit 2 in March of 1979.

The analysis of such accidents requires consideration of phenom-ena accompanying damage to the core which could cause a breach in i

containment allowing the release of radioactive materials outside the plant.

Severe damage to the reactor core will occur because of overheating by decay heat, even if the reactor is shut down.

if the core ceases to be covered with cooling water and remains'-

uncovered for longer than about 10 to 30 minutes (depending on accident scenario).

The core will be uncovered either if the primary circuit leaks and the lost coolant is'wot replenished (loss-of-coolant accident) or if the capability to remove decay heat - from the primary circuit is lost, in which case the coolant will boil off through safety valves uncovering the core unless, again, coolant is replenished.

Severe fuel damage will begin if the fuel temperature exceeds about 1300 K at which point rapid oxidation of the zircaloy cladding begins.

If reflooding with water does not occur, the core will continue to be heated by decay heat and exothermic clad oxidation until melting begins at about 2000 K (the melting point of zircaloy).

Gross fuel liquefaction may begin at this point (UO2 is soluble in liquid zircaloy under some circumstances) or it may be postponed as late as the time at which the melting point of UO2, about 3100 K, is reached.

When j

gross liquefaction occurs the melt will at first flow down and i

refreeze on cooler fuel below.

At this point a number of alterna-tives exist.

One possibility is that an impermeable crust forms at the bottom of the core, holding up subsequently formed melt in a pool; another possibility is that such a crust does not form in which case the melt will flow out from the base of the core as it is formed.

In the latter case the melt will be steadily quenched by residual water in the lower plenum of the reactor pressure vessel (RPV): the water will steadily boil off.

Whether these or other processes occur is highly uncertain.

The former case leads to the possibility that when the melt pool l

becomes so large that the crust can no longer hold it up, it may coherently flow down into residual liquid water in the lower plenum of the reactor pressure vessel, possibly causing.a steam explosion which might be large enough to eject a missile that could breach the containment building.

Such a sequence of events l,

would be of high significance to risk because it would hgeach two barriers to the release of radioactivity, the RPV and tJe_ contain-ment building, almost simultaneously.

If these barriers Were 3

,.,,,-,-.--.,m-y,-

-.-,p--__.-

,.,.wmg-

-~-v--wmv----,,w-,,-,----v---

---%w

+- rir

-vww-m-*-"-w-=r'+m***~-"--+-***

- - ' ~ ^ * * ^ - - - - ~ ^ ^ - - - - * - - - - - - ' - - - - - - - - - - - - - - - - - - - - - -

e breached simultaneously there would be a path for release of radionuclides to the environment potentially affording little i

attenuation.

These events may occur relatively soon

(~ 1 hour1.157407e-5 days <br />2.777778e-4 hours <br />1.653439e-6 weeks <br />3.805e-7 months <br />) i after reactor shutdown which also enhances the possibility that the release of radioactivity would be large.

A steam explosion is caused by the rapid transfer of

• thermal energy from a hot liquid (in this case liquefied core material) to water-on a time scale so short

(~ 1 as) as to produce effects associated with the more familiar chemical explosives. Industrial experience with these thermal explosions (13] has shown them to 1

be often capable of doing significant damage.

I An important issue which emerges from these considerations is the l

need to estimate the probability that, given the required condi-tions (liquefied core materials contacting liquid coolant), a i

containment-breaching steam explosion will occur.

This is the,

", conditional probability" discussed in this document.

A second issue which emerges is the need to assess the degree of certainty which may be attached to the probability estimate described above.

One way to perform such an uncertainty assess-ment is to provide bounds which express the range within which we

{

are confident the probability lies.

If the bounds turn out to be suitably narrow or skewed to the low probability side, we may conclude that a probability estimate within these bounds is satis-factory and that further effort to refine the estimate is unnecessary.

On the other hand, wide bounds that include the possibility of high conditional probabilities may signal the need for improved resolution of the issue.

1 This study addresses the steam explosion issue for a pressurized water reactor (PWR) located in a large dry containment structure (for convenience only, data were taken from the Zion PWRs (14].

The reader is cautioned not to apply the results given here to l

other reactor or containment types without appropriate reformula-tion of the problem.

i.

1.2 Previous Assessments The Reactor Safety Study [1] contains the first quantitative analysis of this problem.

Its authors concluded that "a broad band of uncertainty must be associated with a quantitative evalua-i tion of the likelihood of failure of the containment as a result l

of a steam explosion in the primary vessel. " To express such i

consensus as they had achieved, they adopted a range 10-1 to 10-4 for the conditional probability of containment failure, given core melt.

(These numbers were the fifth and 95th percen-tiles of a skew log-normal distribution whose median was 10-2,)

It was'thus clear in 1975 that further investigation was necessary.

{

[

e l

4 1

The German Risk Study [4] came to essentially the same conclusions in 1979: on the one hand "a destruction of the containment vessel as a result of a steam explosion is very unlikely," but on the other, problems of meltdown, fragmentation and heat transfer were "open",

leaving a degree of uncertainty reflected in the range 10-1 to 10-3 chosen for the conditional probability of*contain-ment failure (fifth and 95th percentiles, lognormal distribution, median 10-2), pending the results of further research.

The recent (1982) UKAEA PWR Degraded Core Analysis Report [5] took a similar view, stating "despite recent work the uncertainty pro-perly recognized by the Reactor Safety Study and the German Risk Study has not been significantly diminished and we see no reason to adopt a narrower range than that of the Reactor Safety Study."

The view expressed by the authors of these three reports [1, 4,

5]

is thus that, although containment failure is thought to be unlikely or impossible, physical uncertainties have so far pre-

~

vented this from being demonstrated.

A number of other studies, however, have argued that these uncertainties are less important and that a small or zero probability can be adopted with certainty.

The Report of the Swedish Government Committee on Steam Explosions

[6] advanced several arguments that containment failure is imposs-ible and concludes that "it is possible to exclude completely the possibility of steam explosions of such force that they could lead to rupture of the reactor vessel and containment."

Fauske and Associates. Inc. have argued [7] that large steam explo-sions are impossible at ambient pressures near to atmospheric because steam production would prevent formation of a large enough coarse premixture; and at higher pressures because triggering will not occur.

They conclude that containment failure by this mecha-nism is impossible.

Similar arguments have been advanced by Theofanous and Saito [8, 9) who conclude that "the steam explosion-induced containment failure probability is judged essentially incredible, i.e.,

at least two orders-ofmagnitude lower than the 10-2 estimate given in WASH-1400."

Mayinger [10] argued that molten core flows slowly out from a degraded core so that steam explosions involving a large amount of melt cannot occur in a reactor, and that even if such an explosion did occur, so small a percentage of the heat in the melt would be converted into kinetic energy as to preclude endangering either the reactor pressure vessel or containment building.

l E

< =.

5 l

I

i f

Squarer and Leverett (12) assigned the following probabilities in an event tree:

probability of large coherent melt mass, given core melt 10-1 probability of presence of subcooled water 10-1 probability of containment failure, given large melt mass and subcooled water 10-2 Taken together these give 10-4 for the probability of contain-ment. failure, given a core melt accident.

Despite acknowledging some phenomenological uncertainty, Squarer did not estimate the uncertainty in his probability.

1.3 Aims of this ReDort Two points of view are summarized in Subsection 1.2: that our understanding of the physics of steam explosions and their effects is, or is not, sufficiently certain to justify the conclusion that containment failure is impossible or very unlikely.

Although conflict can arise over single estimates of the proba-bility of containment failure, we are more concerned here with different perceptions of the degree of the uncertainty of this probability.

Resolution of the conflict concerning uncertainty lies in a proper clarification of the combination of the various component uncertainties into an expression of the overall uncer-tainty.

This report attempts such a clarification and resolution.

This report therefore has two aims:

The first is to provide an uncertainty estimate for the conditional probability of contain-ment failure by steam explosions (given core melt).

The second is to identify important contributors to this uncertainty, in order to provide understanding of the reasons for its magnitude and to indicate what additional information would be needed to reduce it.

This report offers several improvements over a previous analysis of this problem by Svenson and Corradini (11).

The uncertainty is evaluated in a more complete and systematic manner.

Also, modeling changes have been made.

One modeling change that influences the results strongly is that a more realistic, and lower, limit is placed upon the capability of the upper internal structure within the reactor pressure vessel to dissipate explo-sive energy.

Other changes were made-in several of the assumed input distributions, including those describing the amount of melt that participates' in an explosion and the fraction of thermal energy converted to mechanical energy (convers Q ratio).

6

• y-'q

-rwopv ar m-+u-M-*h.-e--

ene-------m-e'e--s---

2.

Methods 2.1 Uncertainty and Sensitivity Analysis We first distinguish between uncertainty and sensitivity analysis.

Uncertainty is a state of incomplete knowledge, and our uncer-tainty of the value of a quantity is expressed as the ran*ge within which we are reasonably sure that it lies.

Uncertainty analysis is the process of determining this range, and this can be accom-plished by finding the smallest and largest values of the quantity of interest that are obtained by varying the parameters upon which it depends over their ranges of uncertainty.

The sensitivity, Si, of dependent variable F to a change in one of the independent variables, xi, can be defined as 0F(x, x ' ***)

3 y

2

~

i :_

ax i Hence, a change in the function F,

i.e.,

6F, which results i

from a small change in the independent variable xi, i.e.,

6xi, is given by 6F = S 6x g

g g

However, sensitivity is used here in a more general sense, to include the effect of changes in F when the underlying variables are varied over their ranges of uncertainty.

This study incorporates both uncertainty analysis and sensitivity analysis to attain the aims set out in Subsection 1.3.

The sensi-tivity study will identify the important factors contributing to the overall uncertainty.

2.2 Monte Carlo M9thod The earlier study (11] used a simple parametric model of a steam oxplosion and its effects which determined whether or not contain-failure occurred as a function of values of uncertain para-cent teters (such as fraction of core molten and conversion ratio).

The uncertain parameters were sampled by the Monte Carlo method.

In this method probability density distributions are assigned to cach uncertain parameter.

A value for each parameter is sampled at random according to these distributions.

The model then deter-Cines whether containment failure occurs.

Such trials are repeated many times, each time with a newly sampled set of para-Ceters.-

The fraction of trials in which containment failure cccurs is then an estimate of the probability of failure.

E n =.

i 7

The uncertain parameters used in reference 11 and in this study have the following features in common:

1)

Each parameter is known (with greater or lesser certainty) to lie within an interval bounded by values based og physical arguments.

2)

It is not known whether each parameter takes the same value (or range of values) for different accident sequences.

3)

It is not known whether each parameter takes the same value for hypothetically repeated occurrences of the same accident sequence.

Some may well do so but others, being random variables, may not.

4)

If there are parameters which take different values for hypothetical repetitions of the same accident sequence, their probability distributions (conditional on accident sequence) are unknown.

In these circumstances, if probability distributions are assigned to the uncertain parameters, they must be interpreted as distribu-tions of subjective probability for consistency in calculations.

The concept of subjective probability has been extensively dis-cussed in textbooks; the points required for this study are suaraarized in Appendix A.

For consistency of method with the previous study [11), this study uses Monte Carlo sampling of subjective probability distributions of the parameters that the previous study found to be important.

i As explained in Appendix A it is necessary to vary these distribu-tions, within their possible ranges.

Each subjective probability I

distribution is therefore systematically varied within the range of its parameter.

A complete variation of such a distribution l

would include distributions of all possible shapes, widths and means.

The selection made here is of three flat distributions covering the high, middle and low thirds of each parameter range.

Evidence does not exist to determine the choice of distributions used, so it is essentially arbitrary.

The present choice of

-~

rectangular distributions was made because these distributions cover the whole parameter ranges used, they allow sampling from different parts of the ranges separately, they avoid giving the erroneous impression that they are derived from direct measure-ments, and they do not express any preference for different parts of the ranges.

Other choices of distributions could equally well have been used, however.

Subsection 5.5 below discusses the effect of this choice upon the conclusions of this study.

The numerical results show that the selection of distributions used did not cause underestimation of the uncertainty in the contain-ment failure probability.

This selection also cannot cause over-estimation of the uncertainty because it is only a susset of the possible distributions.

4%

8

The probability of containment failure due to in-vessel steam explosions may depend on the accident sequence.

Thus, with com-plete knowledge, the probability could be evaluated for each sequence; if desired, a weighted average could then be obtained for all sequences or a subset of them.

The calculations in this report refer to an unspecified sequence where the ambient pressure is near to atmospheric with water in the vessel.

The same uncertainty intervals for the uncertain parameters apply for each such sequence.

Thus our calculated uncertainty interval for the probability of containment failure will encompass the values corresponding to each of the above-mentioned sequences and their weighted average.

The fraction of core-melt sequences having no water in the vessel is unknown but it could be zero.

Including such sequences, in which steam explo-sions cannot occur, would reduce the weighted average probability.

Anticipating results obtained later in this document, the calcu-1ated range of containment failure probabilities includes the value zero.

Thus our calculated uncertainty interval will also apply to the weighted average Probability for all sequences near to atmospheric pressure.

Sections 5 and 6 discuss the effect of relaxing the restriction on pressure.

4 ao b

e i

c.

9

..-----,.------,n-,--

,,-n---

.I I

3.

Modeling Assumptions 3.1 General Section 3 describes the model of steam explosions and their I

effects used in this study.

It indicates which parameters are uncertain and over what ranges, and the distributions ind values that are sampled from these ranges.

l The core-melting process is characterized by the fraction of the core molten at the time of the steam explosion (Subsection 3.2).

This molten core is modeled as flowing out from the core region in a stream having a particular diameter; the melt mixes with i

residual water in the RPV lower plenum and after a delay, param-eterized by the length of the pour, an explosion is triggered (Subsections 3.3 and 3.4).

Subsection 3.5 discusses recently-proposed hypotheses that melt-water mixing is limited by steam,

production and concludes that while they are within the realm of possibility they have not yet been established well enough to justify an upper bound on mixing.

The quantity of the available water that is in the mixture is deduced from current data (Subsection 3.6).

i When a steam explosion occurs, a certain fraction of the heat "available" in the hot melt, i.e.,

of that in excess of the temperature of the water, is converted into kinetic energy.

This fraction is called the conversion ratio appropriate values of i

this are discussed in Subsection 3.7.

Subsection 3.8 discusses the heat content of the hot melt.

The kinetic energy produced in an explosion is shared among the materials thrown off.

In the present case the geometrical i

arrangement of melt and water immediately prior to the explosion affects the partition of this energy (Subsections 3.9 and 3.10).

This partitio'n is also affected by whether or not the RPV fails at the bottom; this can mitigate the effects of a steam explosion i

upon the top of the vessel by venting explosion products and kinetic energy downwardc (Section 3.11).

!.~

I The destructive potential of the upward-moving slug of material driven by the explosion depends on its density as well as its energy (a denser slug of the same mass and energy will exert a higher pressure on'an obstacle it encounters).

The slug density depends on its composition including any voids within it (Subsection 3.12).

As the slug traverses the upper internal structure (UIS) above the core region in the RPV it may damage it and be decelerated by it (Subsection 3.13).

If the UIS does not stop the slug, the slug next impinges on the vessel top head.

The resulting pres-sure loadings are discussed in Subsection'3.14, togethef with the criterion for failure of the bolts retaining the top *Wead and consequent miss.ile generation.

10

If a missile is generated, it may be stopped by a missile shield.

Alternatively the shield will reduce the missile energy.

The missile energy will be further reduced by gravity.

The missile may then hit the containment done and, depending on its speed at impact, fail it (Subsection 3.15).

Subsection 3.16 summarizes the modeling described in Se'cti~on 3 by listing the equations used.

This study only considers pressurized water reactors.

Physical dimensions have, where possible, been taken trom the Zion plants, Unita 1 and 2 [14].

The results should not be assumed to apply to other PWRs without a careful comparison of the important initial and boundary conditions. The calculations refer to acci-dents where the ambient pressure in the primary system is near to atmospheric.

In Subsection 5.1 below, we describe the rationale for this constraint, and possible effects of relaxing it.

3.2 Fraction of Core Molten This means the fraction of the reactor core molten at the time of

~

the postulated steam explosion or, if more than one explosion occurs, at the time of the largest one.

This quantity is deter-mined by the processes of core degradation and the sequences of core degradation and meltir,g which are at present not well under-4 stood.

On the one hand it is argued [10] that melt issues steadily from a degraded core as soon as it is formed over a period of many minutes and therefore the fraction molten at the time of any explosion will be small

(< 0.1%); on thr! other hand, i

the possibility remains that a self-herated pool of' melt will be retained and held up by a crust of refrozen melt until a large fraction of the core has melted [2].

Thus our range for the fraction of core molten is 0.0 to 0.75.

The higher value corre-sponds to the. whole core except a layer 160 mm thick over the side, top and bottom; it is difficult (but not impossible) to i

envisage a larger melt pool than this.

Subsection 4.3 describes a sensitivity study in which the effact of fractions of core i

molten in the range 0.75 to 1.0 was investigated.

For the cases l--

studied, these higher values made little difference; this is because the pouring parameters (Subsections 3.3 and 3.4) generally I

provide the strongest constraint on the mass of melt in an explo-sion.

The total mass of fuel elements in the core is 125.200 kg, so 0.75 of this is 93,900 kg.

Figure 1 shows three uniform proba-bility distributions used for the fraction of core molten (low:

0-0.25%: middle:

0.25-0.5: and high: 0.5-0.75).

i ch i

11 i

4

,,c,~,-n.- w

--n--

u-,-,,--w,.-.--,

w nn--

- - - -, - - - - - ~ ~ -, ~ - -

J L 4.0 WC

-dc LOW MIDDLE HIGH vm-

%KE ESE vs a.

I 0.00 0.25 0.50 0.75 FRACTION OF CORE MOLTEN 3

Figure 1:

Three Uniform Distributions of Fraction of Core Molten 3.3 Pour Diameter The quantity of molten core that mixes is parameterized here in a different way from that in the earlier study (11).

Instead of using the fraction of molten core that mixes, melt is assumed to pour out through a circular hole of constant diameter over a distance call'ed the pour length before an explosion is triggered.

The volume of melt participating in the explosion is the product of the area of the hole and the pour length.

The two ways melt can pour into the lower plenum are either down-ward penetration through the lower core plate (Figure 2), or side-ways penetration of the melt through the baffle plate, core barrel and thermal shield (Figure 3).

For failure of the lower core i

plate, the two conditions that limit the flow rate are the exit l

hole diameter and the flow area through the diffuser plate.

The i

exit hole diameter will be determined by the size of the initial failure and any ablation as the pour proceeds.

The 96 support columns provide redundant support of the lower core plate: thus initial local failure of the plate may well not lead a

e.

12

'I 'I

?

(

l CRUST, SINTERED RUBBLE b I FRACTURED FUEL, Zr0

~

2

~

INTACT FUEL RODS ?

6...

I i-l t' '

l l}v h I.:.

"h, ELT

,f

! llll. A

' $ l '.ll

/^

T Th l"E Y E &

ELT FLOW _

T INTO WATER g

y x

k

' LOWER CORE PLATE h

I DIFFUSER PLATE WATER LOWER SUPPORT PLATE

%ux

[

n Figure 2:

Melt Pour Into Lower Plenum by Failure of the Lower Core Plate 13

'I W$

l l

?

l CRUST, SINTERED RUBBLE r

j FRACTURED FUEL, Zr0 2

i INTACT FUEL RODS ?

I l ',.

f.

s c5 mtT l.

• i "g...,.....

._ J,

kl ::

l i,

/ WATER LEVEL t

,, f ELT FLOW / T 4 y q g.h Y k Mk]@

h INTO WATER l

p

'h gk 4

i WATER NNm Figure 3:

Melt Plow Into The Lower Plenum by Sideways l

Penetration of the Core Barrel A.

14

_ =

i

)

directly to a massive collapse.

On the other hand an initial j

small pour may cause a steam explosion large enough to disrupt i

i the lower core plate, causing a subsequent larger pour.

Multiple explosions have frequently been observed in Sandia's steam explo-i sion experiments [15, 16).

The second limiting flow area is the l

flow passages in the diffuser plate.

If the lower core plate f ails massively due to a steam explosion, the diffuser p1* ate would probably be disrupted at the same time.

If the diffuser plate remains intact the flow area would initially be limited by the j

open area in the diffuser plate - about 1/4 of the core area:

this would lead to an estimate for pour diameter of 1/2 the core l

diameter.

However the diffuser plate may be rapidly ablated, so l

this limitation on the pour diameter may not be effective.

Factors which can limit the effective pour diameter into the lower plenum due to sideways penetration include the size of the exit hole through the baffle, core barrel, and thermal shield, and the -

flow area available for the pour to enter the lower plenum by i

flowing down the downconer.

Because of the secondary core l

supports and the radial keys, it is unlikely that the entire circumference of the core barrel could fail simultaneously if the l

initial penetration is local.

This will be the case unless melt progression is highly symmetrical.

The size of the penetration i

will grow due to ablation, but the secondary supports will pre-I vent the lower support structure from cocking and further opening l

the hole.

The pour area is limited by the annulus between the core and the reactor vessel which, neglecting the thermal shield, l

is approximately 0.26 m wide.

Assuming the pour occurs over 1/4 l

of the circumference leads to a flow area of 0.04 m2 For the i

whole circumference it would be 3.4 m2 Thus, the maximum flow l

area for sideways penetration and pouring is 3.4 m2, correspond-ing to an effective diameter of about 0.3 of the core diameter.

1

'I The above arguments show that the upper limit of the effective i

pour diameter is the full core diameter.

Hence, the distribution used for this parameter are those shown in Figure 4.

1 j

3.4 Pour Lenath for Triacer Time)

.i.

Experimental data at intermediate scale indicate that a steam cxplosion can be spontaneously triggered at almost any time after l

Celt entry into the water and up to about 30 ms after the mixture contacts the bottom of the vessel (16-18).

Furthermore, the melt front in the mixture appears to fall through the water with an

}

approximately constant velocity (19, 20).

All the melt in the i

water at the time of triggering is assumed to be mixed and to participate in the explosion.

This is consistent with the method l

used at Sandia to calculate conversion ratios, which empl6yed the same assumptions.

(Note that some experiments [21] have been cnalyzed by subdividing melt in the water into " mixed" and

" unmixed" fractions).

The possibility of limitations q$ the i

oxtent of mixing is discussed in subsection 3.5 below.

4:.

}

l l

15 i

i

L 0.38 A

W 5 '5 ggg LOW MIDDLE HIGH W2 EE" mmE i

0.00 1.13 2.27 3.40 (CORE DIAMETER)

DIAMETER (m)

Figure 4:

Three Uniform Distributions of Pour Diameter

~.

In the reactor vessel, a likely trigger location would be the lower support plate, about 1.8 m below the core.

It is also:

possible that the melt will pass through the support plate and be triggered at the bottom of the vessel, for example because the support plate may have been damaged or moved by the first of two steam explosions.

Thus, the effective upper limit of the pour length is the depth to the bottom of the vessel, 3.0 m.

The probability distributions used for this parameter are shown ir Figure 5.

Thus, the middle distribution includes the case where triggering occurs preferentially at the lower core support plate.

The high distribution corresponds to triggering at the vessel base.

Recent experimental data indicate that the ease of triggering of melt-coolant mixtures may increase with increasing scale (18);

i.e.,

larger macccc might tend to be triggered at shallower de.pths.

The low distribution allows for this possibility.

J i 1.0 a'

w C 's yy[

LOW MIDDLE HIGH u m.-

W2G mo=

5EE 0.0 1.0 2.0 3.0 (DISTANCE TO VESSEL BOTTOM) i c:.

LENGTH (m) i Figuro 5:

Three Uniform Distributions of Pour Length 16

i Squarer has suggested that the water has to be subcooled for i

spontaneous triggering of an explosion [12).

This idea is called into question by two explosions observed in one test with satu-rated water by Buxton and Benedick (15] and by three spontaneously j

triggered explosions observed by Krein and Berman with hot water (subcooling only 2-5K) [22-24).

4 j

3.5 Mixina Limitations The volume of melt participating in the explosion is assumed to i

be the product of the area of the pour and the pour length, or 1

the total volume of the core molten, whichever is the smaller.

i The mass of melt is calculated using a density of 7000 kg/m3 1

Table I indicates the ranges in melt mass available that result from grouping the various distributions together.

For low distri-4 butions, the largest mass that could be mixed is 7000 kg.

For 1

the middle distributions, the mass range is 7000-56000 kg.

For'.

i the high distributions, the range is 56000 to 94000 kg.

The upper limit of 94000 kg corresponds to 0.75 of the core and is also the l

maximum that was used in the previous study (11).

TABLE I.- RANGES OF POURED MASS 4

l i

Core Pour Pour Pour Pour Pour l

Molten Diameter Area Length Volume Mass (1000 kg)

(m)

(m2)

(m)

(m3)

(1000 kg) j Low 0-31 0.0 -1.13 0.0-1.0 0.0-1.0 0.0-1.0 0-7 j

Middle 31-63, 1.13-2.27 1.0-4.0 1.0-2.0 1.0-8.0 7-56 f

High 63-94 2.27-3.40 4.0-9.1 2.0-3.0 8.0-13.4*

56-94*

i.

• Limited by mass of core molten i

It has been suggested [7, 8] that large coarse mixtures of liquid

+

fuel and water cannot form because the resulting steam production would drive the mixture apart.

This would preclude steam explo-sions involving large quantities of melt.

Mhere these arguments i

are developed quantitatively [7, 25), they depend strongly upon a i

number of simplifying assumptions, notably those of a ste'ady state and one-dimensional flow pattern (5).

The Henry-Fauske model [7, j

25] predicts that only a few hundred kilograms could mix in the i

lower plenum, and that this quantity is independent oqiwater depth.

Theofanous has postulated that 2-3% of the coratWould i

17 i.

l represent the maximum mass that could mix in-vessel [8, 9]: this would correspond to 2500-3750 kg.

The Corradini model [19, 20, 26-31] predicts that the amount mixed increases with water depth and with coarse particle size.

The latest formulation of this model (30, 31] predicts mixing of 3000 to 5000 kg corresponding I

to particle diameters of 50 to 100 mm.

Because of the simplifying assumptions in each of the models, all l

of the predicted upper limits to the extent of mixing are uncer-tain.

For example one assumption made by Henry and Fauske is that ll melt particle diameters less than 1 cm are necessary for a steam explosion.

However, although there is a wealth of data indicating

}

that mixtures of cm-sized particles can explode (for example j

reference 17) it has been suggested [8, 9] that much larger par-ticles can also participate in steam explosions.

Another assump-tion made by these models is to ignore the potential of steam flows to enhance mixing as well as to suppress it.

Two recent~.

experiments, in which initially stratified configurations with water on melt exploded, demonstrated the possibility of such enhancement, because melt-water mixing appeared to take place

]

spontaneously (23, 24, 32].

i l

A complete mechanistic uncertainty analysis does not exist for

{

any of these mixing models.

Two examples show that they can be very sensitive to changes in assumptions, however.

Corradini

[30, 31] showed that changing the limiting criterion in the Henry-Fauske model from a critical heat flux criterion to a fluidization criterion changed the maximum fuel mass mixed from 100 kg to 550-750 kg for 10 mm particles or 5300-12000 kg for 100 mm par-ticles.

The ranges given in each case are due to uncertainty in the effective water particle size for fluidization.

The current Corradini model [30, 31] assumes that the melt is initially in a spherical configuration.

If, as in some earlier formulations (26]

it is assumed that a cylinder of the same diameter.and having length equal to the water depth can mix, this changes the range i

of upper limits from 3000-5000 kg to 14000-20000 kg.

i These variations, although they cause large changes in model pre-l-

dictions, by no means take account of all uncertainties or span the whole range of possible mixing limitations.

Corradini's model is the most detailed.

It allows for transient break-up of the l

melt [29-31]; however the formulation used for this process is I

itself uncertain.

None of the models allows for the possibility

{

of the transient existence of mixtures which would be unstable I

due to large steam flows in a steady state.

An extreme, but often observed, transient mixture of this kind is that formed by the first of two explosions.

Direct experimental evidence is inconclusive in distinguishing 1

l between large-scale mixing models because the bulk of the world's s

e data was taken for melts of mass about 20 kg or less and is consistent with all proposed limits to mixing.

Some reEint i

i j

18 i

t experimental data (33, 34) imply that the Fauske [25] model may underestimate' the size of the mixtures that can be formed, but this has not been conclusively demonstrated.

I Regardless of the accuracy of.any of the models, there may well be a tendency for mixing to become more difficult with* increasing scale.

.This could strongly influence failure probtbilities.

However, at present, evidence does not exist to allow an upper limit to be imposed on the mass of melt mixed that is less than i

all the available melt.

The distributions used in this study effectively cover the whole range of possibilities.

As shown in i

Table I,

combinations of distributions that include " low" pour i

diameter or length allow for the possibility of very small amounts i

of coarse mixing.

In addition, the maximum value in the low range is less than twice the Theofanous estimate of a maximum of 3,800 kg [8).

The uncertainties in the current coarse mixing models do j

not preclude any of the distributions studied here.

3'. 6 Fraction of Water That Mixes The zones of mixed fuel, water and steam in the FITS experiments

~

were observed to be roughly paraboloidal (17).

The data for the volume and depth of the mixing zones [19, 20) show that the dia-meter of the mixing zone tends to about four times the initial fuel diameter.

The mixing zones contain 40-60% by volume of liquid water.

Here we estimate the water mass mixed by assuming a cylindrical mixing zone with four times the diameter of the pour diameter.

Thus, the masses of melt and water are approximately i

equal.

For this analysis, the mass of water that mixes is set l

1 equal to the fuel mass or the total water mass (28000 kg), which-4 ever is smaller.

In these calculations, this mass only affects the partition of material between upward and downward moving slugs.

A sensitivity study, described in Subsection 4.3, showed that the reaults were insensitive to changes in this partition.

3.7 Conversion Ratio By conversion ratio, we mean the fraction of the heat in the melt participating in the explosion (assumed to be all the melt that mixes with water) above the water temperature, that is converted to kinetic energy.

1 Various estimates of the. conversion ratios in reactor accidents have been made.

Mayinger [6, 10) and Becker [6] refer to a report by Haag and Korber [35) which suggests that conversion ratio falls with' increasing melt mass.

This conclusion appears to arise from limited experimental data.

Squarer [12] predicts conversion ratics of 1% or less.

Theofanous and Saito [8, 9) estimate an explosion energy of 600 MJ which, together with their estimates of the mass mixed and of the heat in the molten core, implies a conversion ratio of 15%.

Ultimately they expect that will be possible to demonstrate a reduction of a factor of 5 the l

l 19 J

I

,--.--._,-.---..-.,.~....-.___-,..,-_..-.-_n-_-_2-_.,------,--.,---,-----.-,---

i value, giving 3%.

The Gittus Report [5] similarly estimates an upper limit of 4% subject to a factor of 4 uncertainty either way.

In the Zion / Indian Point study [3], calculations indicated the I

potential for a significant increase in conversion ratio with increasing scale.

When an overlying molten pool was assumed to exist, steam expanding away from the explosion was furthen heated in passing through the molten region.

Effective conversion ratios as high as 14% were calculated, much higher than the experiment-ally measured results discussed below.

The difference was due to the calculated heating of the steam, compared to experiments where the steam is cooled by expanding through cold water, and possibly also due to increased inertial confinement.

The predictions of these calculations however, are uncertain because of several simplifying assumptions used [5,6].

However they do indicate i

effects that should be taken into account when assessing the uncertainty introduced in extrapolating to larger scales.

M&ny experiments have been conducted at Sandia over the last few years to measure conversion ratios.

Fifty-nine intermediatescale

(< 20 kg) experiments were conducted in a cylindrical steel tank using iron-alumina and corium melts [15, 36].

The largest conver-sion ratio measured was 1.34% when a cover plate was used to increase the degree of confinement.

In one other test, the con-i version ratio was estimated to be nearly it.

For the other 57 l

tests, it was 0.6% or less.

Fifty-five tests have been conducted i

in the FITS and EXO-FITS facilities using corium and iron-alumina melts ranging from 1 to 20 kg [16, 17).

The largest conversion ratio measured was about 2.5%; as with the earlier tests, many of i

the explosions resulted in conversion ratios in the range of 1 to 2%.

The relevance of these experiments has been questioned by Fauske and Henry [37).

Their criticisms have been answered by Corradini and Berman [23, 24].

The largest conversion ratio ever measured at Sandia National Laboratories was about 4.4% for a single-droplet explosion at an ambient pressure of 0.96 MPa [26, 1

38).

The accuracy of all these conversion ratio measurements is i

probably better than a factor of two.

Guided by these data, the range of conversion ratios from 0 to 5%

was used for most of the calculations in this study, Figure 6.

This range does not, however, fully account for the uncertainty introduced by extrapolating from kilogram-scale experiments to accidents at the scale of thousands of kilograms.

At larger scales the conversion ratios may decrease as suggested by Haag and Korber, remain within the range currently observed, or increase (perhaps due to increased inertial confinement).

Estimates of the largest conversion ratios possible may be made by considering calculations of maximum work thermodynamically possible.

Such calculations have been made by Corradini and Svenson [11, 39] and by McFarlane [40).

Both of these assume that stated masses of molten core and water mix and came to thermal equilibrium at constant volume and then, with ne further 8

20

heat transfer, expand up to the volume of the reactor vessel.

The two calculations differ mainly in that Corradini and Swenson assume that the initial state has a 50% volume fraction of steam, while McFarlane does not.

Table II gives the results of these two calculations.

Corradini and Swenson predict lower maximum conversion ratios because their mixtures expand from* larger l

volumes.

Based on these calculations, we choose 16% as a repre-sentative upper bound on the conversion ratio.

This value is also consistent with the upper limits of references 5 and 8.

Additional flat distributions over the lower, middle and upper thirds of the range 0-16% were used to examine the sensitivity of the results of the study to the possibility that conversion ratio increases with scale.

J i 0.6 WO

f C LOW MIDDLE HIGH S3G 2o{

5E 0.0 1.7 3.3 5.0 CONVERSION RATIO (%)

Figure 6:

Three Uniform Distributions of Conversion Ratio TABLE II.

Thermodynamic Maximum Conversion Ratios (%)

Mass Mass of Water (1000 kg) of Fuel (1000 kg) 5 5

10 10 20 20 Ref 39 Ref 40 Ref 39 Ref 40 Ref 39 Ref 40 5

13.1 13.2 4.8 6.3 3.8 2.3 10 15.9 19.3 9.3 10.3 2.7 4.6 20 17.9 22.2 11.4 15.9 6.3

'5. 3 40 10.7 16.9 12.8 19.2 8.0 13.2 80 7.3 10.9 7.6 16.2 7.0 16.9

[

c l

21

3.8 Heat Content of Molten Fuel The heat content of the molten fuel depends on the course of melt progression and the constituents of the melt.

Because of uncer-tainty in core melt progression, the heat content is uncertain.

In this study, we have assumed a base water temperature of 400 K.

The lowest temperature at which core material may liquefy can be estimated at 2000 K. when liquid Zircaloy begins to dissolve solid U0.

Using a specific heat of 500 J/kg-K for solid UO, and 2

2 neglecting a small contribution from the latent heat of Zircaloy, gives 0.8 MJ/kg in the melt over 400 K as an estimate of the lower limit of the melt's heat content.

An upper limit can be esti-mated by considering UO2 heated up to its melting point, approx-imately 3100 K, and then melted with latent heat 0.27 MJ/kg.

This implies a total latent plus sensible heat above 400 K of 1.6 MJ/kg.

For most of this study the single value of 1. 2 MJ / kg iP used for the melt heat content; a ransitivity study investigates

,the effect of the values 0.8 and 1.6 MJ/kg as well.

~

The kinetic energy produced in the explosion is the product of the mass of melt in the water at time of triggering, the heat content of the melt, and the conversion ratio.

1 3.9 Fraction of Remainino Melt Above ExDlosion For calculations of events after the explosion we need to know the position of the melt that did not participate.

In the pre-vious study [11] the fraction of the remaining melt above the explosion was regarded as an undetermined parameter in the range 0.0 to 1.0, the rest of the melt was assumed to be below the explosion.

For simplicity, we have assumed that all the melt that did not participate in the explosion remains above the explosion in most of the calculations in this study.

Alternative assumptions used are discussed in Subsection 3.10.

3.10 Fraction of Remainino Water Above Explosion Similarly to their treatment of remaining melt, Swenson and Corradini (11] assumed that, of the water not participating in the explosion, a fraction between 0.0 and 1.0 could be located l

above the explosion.

As in (11), we assume that all the water i

not participating in the explosion lies below the explosion for most calculations.

Both the assumptions in this Subsection and Subsection 3.9 may not be right; but they counteract one another in determining the slug energy and so do not represent an extreme combination.

The possible extreme combinations in which all the melt and all the water not participating in the explosion are either bot) above or c:r.

22

4 t

both below the explosion are investigated in a sensitivity analy-sis discussed in. Subsection 4.3.

This shows the results to be insensitive to these assumptions.

3.11 Enerav Dissipation by Botton Failure

+

l In common with the previous analysis [11) we assume that the base i

of the RPV fails if the explosion energy exceeds 1000 MJ.

t

}

Reference 3 placed this threshold in the range 1000 to 1500 MJ.

}

In Subsection 4.3, a sensitivity study shows that the results are insensitive to the value of this threshold within the range 500 j

to 1500 MJ.

Swenson and Corradini [11) assumed that failure caused dissipa-1 tion of between 0.0 and 0.5 of the explosive energy:

the remain-der being kinetic energy of the upward moving slug.

Here we i

assume that, if the bottom of the vessel fails, two masses are accelerated:

a downward moving one consisting of the vessel base (30,000 kg), the water below the explosion and one-half of the

+

water and melt participating in the explosion; and an upward

~

moving slug being the other half of the melt and water in the explosion and the melt above the explosion.

On the assumption i

that there is no not transfer of momentum from the body of the vessel to the slugs, these share the explosion kinetic energy in

)

inverse proportion to their masses.

mass below i

KE,3y, mass above + mass below KE,,pg,,g,,

=

4 i

These simplifying assumptions neglect the delay before bottom

)

failure (which would increase the upper slug's energy) and side-j ways venting of steam and work absorbed in failing the bottom (which would reduce it).

3.12 Sluo Composition i

If the vessel bottom does not fail the upward moving slug is assumed to consist of all of the melt and water participating in the explosion and the melt above the explosion.

If the botton does fail, only half of the exploding materials are assumed to move upwards.

Steam formed in the explosion may impregnate the upward moving slug with bubbles or break it up into a spray of droplets.

This would change its mechanical effects by altering the momentum flux in the slug and hence its stagnation pressure.

d

?

We treat the volume fraction of condensed phases (liquid plus

}

solid) in the slug as an uncertain parameter.

On the one hand the volume f raction of condensed phases might be large.,'as steam might not penetrate forward into overlying melt, and affat be 4

1 23 i

i

l compressed during interaction between the slug and upper internal structure.

We take 1.0 as an effective upper limit.

As a lower limit we take a volume fraction of 0.25.

Figure 7 shows the dis-tributions of this parameter used.

In calculations of the effects of slug impact (Subsection 3.14) the slug density is calculated from the masses of water' and fuel and volume fraction of steam in the slug.

J L 4.0 h

g-LOW MIDDLE HIGH saa 2e$

aE 0.25 0.50 0.75 1.00 FRACTION Figure 7.

Three Uniform Distributions of Condensed-Phese Volume Fraction in Slug fuel mass water mass vol slug" fuel density +

water density volume fraction of condensed phases slua pslug =

yng slug 3.13 Eneroy Dissipation by Core and udder Internal Structure The remains of the reactor core after a core-melt accident are unlikely to be able to withstand substantial forces and'honce to be the cause of significant dissipation of energy following a large explosion.

Cffects of the order of the gravitational poten-tial energy of the core, ~ l MJ, are to be expected buf this is negligible in comparison with the explosion energies c6nsidered here.

Thus we ignore absorption of slug kinetic energy in the residual core.

24

. ~

The space between the upper core plate (at the top of the core) and the upper support plate (at the level of the vessel flange) is 3.22 m high.

It contains 40 support columns which hold the core down against hydraulic friction forces in normal operation, and control rod guide tubes and drive shafts and any control rods in the withdrawn position.

The space above the core ic*needed to accommodate withdrawn control rods.

For our present purposes we call the components between the upper core plate and the upper support plate the " upper internal struc-ture" (UIS).

Interaction between a fluid slug and this structure will dissipate energy produced by a steam explosion (2, 11).

The extent of this dissipation is difficult to estimate because of the uncertain material properties of the slug and the trans-ient nature of the slug loading and the structural response.

Reference 11 assumed, by analogy with experiments in which water" was forced through an undeformed scale model of the upper internal structure of a fast reactor, that the whole UIS would absorb 90%

~

the slug's kinetic energy; and that this factor would be pro-of portionally reduced if part of the UIS had been melted away.

Squarer proposes a similar formulation in which 75% of the kinetic energy is absorbed (12).

We retain the description in reference 11 here with two modifica-tions.

First, the UIS is assumed to be fully intact.

This is because it might initially be protected from high temperatures and heat fluxes (radiative and convective) from the center of the degraded core by the upper layer of the core.

However, recent calculations have demonstrated that, particularly in accident sequences in which the RCS pressure is high, convective heat transfer from the core can cause melting in the UIS before core melting begins (41].

The error introduced if the assumption that the RCS remains intact is wrong is small, as discussed at the end of this Subsection.

Second, the energy absorption in the UIS is limited by its capa-bility to withstand the corresponding forces.

These can be estimated simply and roughly.

We assume the coupling between the slug and undeformed UIS can be described by a constant friction factor.

Then the resultant retarding force exerted on the slug by the UIS will be proportional to its kinetic energy.

This force, F, reduces the kinetic energy E, as a function of distance

traveled, ff--F=-CE i

x. n,e-c*

nl I

25

Since E is presumed to fall by a factor of 10 over a distance 3.22 m (corresponding to absorption of 90% of the kinetic energy),

C-0.715 m-1 Thus the restraining force on a slug of given energy can be estimated.

The force decelerating the slug is transmitted through the UIS to the upper support plate and thence to the RPV.

We now roughly estimate the capacity of the UIS to transmit this force.

Its only components designed to withstand forces in this direction are the 48 support columns.

These are slotted steel tubes of outside dia-meter 190.5 mm and thickness 12.7 mm and total cross-sectional area, taking account of the slots, of 0.155 m2 Assuming a yield stress of 468 MPa implies a maximum sustainable force of 72.5 MN.

This will be a substantial overestimate of the average force during crushing because slight asymetries in plastic yielding will produce buckling which will be enhanced by the slots and will reduce the force required to collapse the columns.

However it will be a be,tter estimate of the force required to initiate crushing.

From the calculation of the frictional forces above, this crushing threshold corresponds to a slug energy of 101 MJ.

Thus we presume the UIS to absorb 90% of the kinetic energy of slugs less energetic than this.

More energetic slugs will be presumed to crush the UIS against a constant force of 72.5 MN (probably an overestimate as noted above) until their kinetic energy falls below 101 MJ.

Thus the maximum energy absorbed is 233 MJ when this force acts over the whole length. 3.22 m, of the UIS, and this energy will be absorbed from the slugs of initial energy greater than 233 + 101 -

334 MJ.

Slugs of intercediate energy will first be decelerated by a constant force and then an exponentially falling one, but for simplicity we interpolate the energy absorption linearly between the two extreme cases as shown in Figure 8.

m 300 5

E h

200 g _3 mm*

3 100 h

W hl O

0 100 200 300 400 INITIAL SLUG ENERGY (MJ)

[

c =.

Figure 8:

Energy Dissipation in Upper Internal Structure 26

1 Because of the possibilities that the UIS is weakened by heating and that its crushing strength is less than calculated here, the dissipation calculated is an upper bound.

Since the greatest possible dissipaticia_ calculated, 233 MJ, is approximately a factor 10 smaller than the ' slug energy cequired to fail the vessel top head, extra uncertainty caused by overestimation of this dissipa-tion will not be considered further.

3.14 Sluo ImDact odel The six degree ofc!reedom slug impact model described in Ref. 10 is one dimensional and hence lessPedalistic than calculations [3, 4] which demonstrate the importanch of two-dimensional effects.

The slug exerts a pressure on the RPV top head of approximately puc where p is the slug density, u its speed and e its sound speed.

This correspondsjto the plane reflection of a sound ways of velocity u.

Additionally the values of c used depend upon the particular prescription assumed for the spsed of sound in a compos-ite medium (11).

/

Instead here we approximate the pressure to be expected in two-dimensional flow, in which fluid m' oves up, across the vessel head and down again, by the stagnation prehnp,re pu2 This is equal to the flux of momentum across a plare through which the slug passes.

We take the total force on tas,RPV head to be this pres-sure multiplied by the cross-sectional' area of the vessel, 14.1 2

m.

Thus we assume.that the part of the vessel's area not occu-pied by upward flowing material"contains material flowing downwards whose acceleration makes a centribution toJthe total force on the head.

~

It may be that the upper support plate,which[Ahans the vessel at the level of the top head flange and which is' reinforced by a web of cross members, is strong enough to withstand $the pressures exerted by the slug.

This will not change our af,riyais because if this plate does not fail it transmits the f orce' ax'orted on it directly to the upper head just above the bolts.

In Appendix B,

we discuss sotkr of the possible? Jusel failure modes resulting from slug impact.

.The,most damaging Jailure would occur if the studs fr'actured and alloyed the head ~to'tly off.

The criterion used in these,calchlatione foi failure of the top of the RPV is that the force on,it exceed 2 the f ailure tension of the bolts.

Multiplying t he i t' combined cross-sectional area, 1.341 m2, by the failure strean'bf 870 MPa gives a total balt failure tension of 1170 MN.

Not'e that bolt fracture occuth while the bulk deformation is elastic; 'cef ore plastic def ormati on' of the bolts occurs.

l e

f

/

As shown in Appendix B,

the period of natural vibration of the vessel is short enough relative to'.the loading that ydk loading can be assumed to be static.

We can then calculdte tWe~ loading pressure and compare it to the failure pressure totevaluate vessel

failure, i

...~ r

/

27 f

/

Y

I Subsection 5.2 discusses the effect of~other possible failure modes of the top of the RPV.

3.15 Containment Failure In this report, we are interested in containment failure resulting from impact by the head.

The Zion containment structure is in the shape of a cylinder with a shallow, domed roof and a flat foundat-i 1

ion slab.

Some approximate dimensions of the reactor containment I

are:

inside diameter 42.7 m; inside height 64 m; containment dome l'

height above reactor 45 m: vertical wall thickness 1.07 m; and done thickness 0.81 m.

The entire structure-is post-tensioned and lined with 6.35 mm-thick welded steel plate to provide vapor tightness.

In addition to the barrier provided by containment, a concrete I

missile barrier is positioned above the reactor vessel to block i

any missiles generated by the failure of the control rod housings.

i Approximate dimensions of the barrier are a radius of 2.5 m and a' thickness of 1.3 m.

The approximate mass is 65,000 kg.

Other equipment above the missile shield includes the polar crane.

~

The sequence of events leading to hypothesized containment' failure starts with failure of the studs, which allows the taad to rise

]

and impact the missile shield.

Impact with the missile shield absorbs some of the head energy.

The head than continues to rise and impacts containment.

Some additional energy is then absorbed in breaching containment.

{

In this calculation, we have estimated the velocities to perforate l

the missile shield and containment using both the NDRC formula i

modified for low-speed impact (42, 43] and the CEA-EDF formula

[44).

These perforation velocities then give the energy absorbed i

during these impacts.

These formulae are listed in Table III.

Alternately, we ammumed that impact'with the missile shield reduced the head velocity in half.

This is equivalent to an inelastic j

collision between the head and shield assuming that a part of the shield, equal in mass to the head, continues to travel with head.

For missile shield perforation, the NDRC equation gives a required i "

velocity of 39 m/s while the CEA - EDF equation gives 55 m/s.

For j

the containment, the NDRC equation gives a velocity of 23 m/s, while the CEA - EDF equation gives 29 m/s.

We now can calculate the range of possible required initial veloc-ities for containment failure.

For the smallest velocity, we assume the head perforates the missile shield with energy loss calculated by the NDRC formula and impacts the containment with a small velocity that nevertheless damages it.

Summing the kinetic l

l l

[

s 4%%

l 28 l

l

Table III. Perforation Formulae [5]

Nomenclature:

d missile diameter (m) m missile mass (kg) 3 p

concrete density (2400 kg/m )

6 a

concrete compressive strength (28.6 X 10 p,)

V missile velocity (m/s) x target thickness (m)

Modified NDRC Formula [42, 43):

G(Z) = 2.55 x 10 ' K N do.2 I*

DV 4

9x 0

K=

= Concrete Penetrability Factor "c

N = 1.0 for spherical-nosed missile (dimensionless) 3 D = "y (kg/m ) = Calibre density d

(fg)

G(Z)

=

CEA - EDF Formula [44):

-3/8

-1/8 a 1/2 y /4 3

l x = 0.82 o p

e d

20 < V < 200 0.3 < x/d < 4 l

AkN-i l

l-29

energies absorbed in perforation and the gravitational potential energy needed to rise through 45 m gives a required initial veloc-ity of 49 m/s.

For the largest velocity, we assume first an inelastic collision between the head and missile shield in which the head loses 3/4 of its kinetic energy, and subsequent perfora-tion of containment with energy loss calculated by the CEA-EDF formula.

Then the required initial velocity is 83 m/sec.

In Section 4, we tabulate missile velocities of 50 m/sec and 90 m/sec to include the range of these results.

This range of missile velocities should not be regarded as a fully justified uncertainty interval because some extrapolation from the experimentally tested ranges of the correlations was used and because of the possibility of effects, such as spinning, which differ from the ideal vertical missile trajectory assumed.

3.16

_ Summary of Modelina This sub3ection summarizes the modeling described in Section 3.*

F'irst the meanings of the symbols used are listed.

Then the i

equations defining the model are set out, with references to the We Subsections where detailed discussion can be found.

Nomenclature:

i Ab total cross sectional area of bolts Ay cross sectional area inside vessel d

pour diameter p

Eb threshold explosion energy for vessel bottom failure Ed slug energy dissipated in UIS E,

explosion energy E

residual slug energy after dissipation in UIS r

E initial upward slug energy u

El initial kinetic energy of top head

~

E2 kinetic energy of top head after missile shield impact E3 kinetic energy of top head at containment impact i

F volume fraction of condensed phases in slug e

F fraction of core molten m

g acceleration due to gravity h

height from missile shield to containment dome

~~

H heat content of melt l

pour length p

Mb mass of vessel base M

mass of core c

Md mass of downward-moving slug Mh mass of vessel top head M

mass of melt mixed with water m

M mass of melt poured out from core p

Mt mass of water Mu mass of upward-moving slug M

mass of water mixed with melt y

P pressure exerted by slug on top head c

R conversion ratio nd(

30

.-m...

~.

Pm density of melt Pu density of upward-moving slug pw density of water ob failure stress of bolts u

velocity of slug at impact on top head V

volume of condensed phases in upward-moving slug.

c V

volume of upward-moving slug u

Mass of melt in explosion (Subsections 3.2 to 3.5):

2 r

ud y

p"3 l

M, - min M F,,

4 j

c Mass of water in explosion (Subsection 3.6):

M,-

min (M, M,).

g Energy of explosion (Subsections 3.7 3.8):

~

Ee - MmHR.

Condition for vessel bottom failure (Subsection 3.11):

Ee>Eb bottom failure.

Mass and volume of upward moving slug (Subsections 3.9 and 3.10):

Vessel bottom intact Vessel bottom failed Water mass Mw 1/2 Mw Melt mass

.M F MFc m - 1/2 M cm m

Total mass Mu " Mw+MFcm Mu = 1/2 Mw+MFc m - 1/2 Mm f M, M F, - f M, M

F w

c, Total volume V"

+

c c

p p

c" p

p (condensed w

m w

a phases)

(Different assumptions are used in a sensitivity study -

Subsection 4.3)

Mass of downward moving slug if vessel bottom fails (Subsection 3.11):

Water mass Mt 1/2 M w

Melt mass 1/2 M m

Total mass Md -Mt - 1/2 Mw+ 1/2 Ma+Mb e

(Different assumptions are used in a sensitivity stuBy -

Subsection 4.3) 31

Kinetic energy of upward moving slug (Subsection 3.11):

i d

Eu"M

+M d

Volume and density of upward moving slug (Subsection 3.12):

Vu = Ve/Fe Au = Mu/Vu Energy dissipation in UIS (Subsection 3.13):

Ed depends on Eu as shown in Figure 8.

Er - Eu-Ed Impact velocity and pressure of upward moving slug (Subsection 3.14):

f2E r My P = Pu u2 Condition for bolt failure (Subsection 3.14):

bolt failure P Av > ob Ab Initial kinetic energy of top head:

1 u"h 1"2 (Mu+

h)

Energy reduction due to missile shield (Subsection 3.15):

Inelastic collision:

E2= 1/4 El Penetration formula:

E2=E1 - kinetic energy needed to perforate Condition for containment failure (Subsection 3.15):

E3 =E2-Mhgh E3 > kinetic energy needed to perforate failure.

[

n.

32

4.

Calculations and Results 4.1 Outline of Calculations Subsection 1.3 of this report listed its aims.

These are to pro-vide an uncertainty estimate for the conditional probability of containment failure by steam explosions (given core melt) and to identify important contributors to this uncertainty.

Subsection 2.1 explained that these aims would be attained by uncertainty analysis (to find bounds on the probability) and sensitivity analysis (examining the dependence of the probability on various samplings of the uncertain parameters, to determine which para-meters have the greatest effect).

Section 3 described the uncertainties in modeling the various processes involved.

Our modeling of these processes is relatively simple.

Nevertheless, the number of different uncertain para-meters in this simple model makes a fully comprehensive sensitiv-ity study difficult.

We have used one of many possible sampling schemes and selected calculations so that our conclusions are, as c'

far as possible, independent of the particular cases studied.

Since the selection of the sampling scheme used was essentially arbitrary, the reader is cautioned against attributing special significance to any individual calculated probability number.

More attention should be given to the way in which the calculated probabilities depend upon the different parameters varied.

Subsection 5.5 below discusses the effect of an arbitrary choice of a sampling scheme upon the validity of the conclusions that we draw.

As explained in subsection 2.2, it was desired to use a Monte Carlo sampling technique (with an adequate sample size) in order to make explicitly clear that any differences between the results of this study, and the previous Monte Carlo study [11) were not due to a difference in statistical method.

The five uncertain parameters judged to have the most important influence on the overall uncertainty were sampled by the Monte Carlo method as described in subsection 2.2.

These five parameters are called here the "first set."

The first set parameters are the two found to be important-in the previous study [11), conversion ratio and slug condensed phase volume fraction; and the three parameters which determine the amount of melt participating in a steam explosion, namely the fraction of core molten and the pour diameter and length.

The importance of the amount of melt in the explosion is potentially high but it was not explicitly investigated in reference 11.

The remaining parameters are called the "second set."

For each of the first set parameters, three alternative flat distributions of subjective probability were assigned f y :overing the low, middle and high thirds of the uncertainty range"of the 33

parameter.

In the main study different combinations of these distributions were selected systematically as set out in Table IV.

Single values of the second set parameters were used.

The second set parameters were the heat content of the gelt, the location of melt and water not participating in the explosion, and the explosion energy required to fail the vessel base.-

As well as these parameters, three upper limits of first set para-meters were varied in additional calculations:

fraction of core melted, conversion ratio and pour diameter.

The second set parameters were varied over their ranges of uncer-tainties in additional calculations in which one or two additional single values of them, selected to cover their ranges, were used.

Each of these values was combined with all the low, all the middle and all the high distributions of the first set parameters.

This sampling thus spans the whole range of each of the first set parameters.

These additional calculations are set out in Table V.

They show that the second set parameters generally had, as expected, less important ur.cer tainties than the first set para-meters.

4.2 Results of Main Calculations In these calculations the different distributions of the first set parameters were combira;d while keeping the second set constant at the following values:

Melt heat content:

1.2 MJ/kg Position of unmixed melt:

over explosion Position of unmixed water:

under explosion Explosion energy needed to fail vessel base:

1000 MJ The cases calculated and the results obtained are set out in Table IV.

The entries in the Table are now explained, using Case 1 as an example.

In this case each of the first set parameters was given a flat distribution of subjective probability, over its whole range.

These full ranges are Fraction of core molten:

0 - 75%

Pour diameter:

0- 3.4 m Pour length:

0- 3.0 m Slug condensed phase fraction:

25 - 100%

Conversion ratio:

0 - 5%

For all cases other than Case 1 distributions labelled L, M and H are used for these parameters.

These mean L:

flat, low third of whole range M:

flat, middle third of whole range H:

flat, high third of whole range.[

c r.

34

a Sl 4

-l TAllt.F IV.

Main Calculationn

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

__._........__..._z Fall.UNES INPUT cal.CUI.ATIONS (per 10.000 t r ials )

Case Fraction Pour Pour Sluq*

Convession Meae.

Mean Sluq' Mean Mean Sluq*

Vessel holts large. Large Molten Diameter length Condensed Ratio Emplosion impact Slug

• Mass botton Misstle Missile 4

(m)

(m)

Phase Fraction (4) f:ncrgy Energy Volume' (2000 kg)

V >50 V>90 (4)

(MJ)

(MJ)

(m3) m/s e/s

. -. _ _ ~ _ _..... _

Full 1

0-75 0.0-3.4 0.0-J.o 25-100 0-5 504 2HJ 31.5 53.1 2087 466 460 267 es t dt h All low 2

L(0-25)

L(0.0-1.13)

'L(0.-l.)

1.(25-50) 1.60.0-1.7)

I 9.3 16.7 0

0 0

0 All middle 3 M(25-50)

M(1.13-2.27)

M(1.-2.)

M(50-75)

M(1.7-3.J) 732 400 40.2 61.6 2126 1

1 0

All high 4

H(50-75)

H(2.27-3.4)

H(2.-3.)

H(75-200) II( 3. 3-5. 0 )

382R 2008 22.6 53.8 10000 9987 9987 9959 All 5

L M

M M

M 407 211 25.2 28.6 172 0

0 0

Middle 6

L 106 24 16.6 50.3 0

0 0

0 g

g Encept 7

L 247 98 24.2 55.1 4

0 0

0 Indav.

8 L

735 404 69.3 62.0 2087 0

0 0

Low 9

L 248 100 45.8 68.5 5

0 0

0 A!!

10 H

M M

M M

735 364 47.3 92.6 2136 0

0 0

Eiddle H

1352 683 32.9 43.8 8272 185 185 2

Except 12 H

1078 570 39.0 54.3 5335 62 62 0

Indiv.

13 H

722 396 28.8 62.1 1977 68 68 0

High 14 H

1203 595 31.3 51.8 5884 384 384 84 All 15 L

H H

H H

779 434 14.6 22.3 3479 1719 1557 0

High 16 L

293 147 19.5 84.0 79 0

0 0

Encept 17 L

1116 433 25.1 88.5 5155 460 460 162 l

Indiv.

18 L

3820 2003 54.6 53.8 10000 4110 4110 4110 Low 19 L

780 392 37.2 87.4 3524 5

5 0

A!! Low 20 H

L-L L

L 12 1

34.2 79.4 0

0 0

0 Execpt 21 H

118 31 38.5 27.3 0

0 0

0 Indiv.

22 1 H

49 8

19.8 20.5 0

0 0

0 High 23 lD H

11 1

3.8 16.7 0

0 0

0 24 H

56 9

9.3 16.7 0

0 0

0 e

3Upward moving slug.

8

i The distributies.s used in each case are listed in the five columns under " INPUT."

In the text these distributions are referred to as " low," middle" and "high."

The next four columns give the mean values, out of 10000 trials randomly sampled from these distributions, of four cal'culated parameters.

These are the mean steam explosion energy,*the slug impact energy, and the volume and mass of the slug.

Thus in Case 1 the mean steam explosion energy was 584 MJ.

The last four columns in Table IV give the number of failures of different kinds calculated to occur out of 10000 trials.

Vessel bottom failures are listed first followed by failures of the top head retaining bolts.

The last two columns give the number of containment failures if the threshold values for the initial velocity of the top head to cause containment failure is 50 or 90 i

m/s.

This estimate of the uncertainty range for this parameter -

was calculated in Subsection 3.15.

Thus the numbers in these two' columns, divided by 10000, estimate the. range of containment failure probability,~ conditional on the input distributions licted under " INPUT" and the values of the second set parameters listed above.

Caution should be used when the numbers in the last four columns are small, as they are subject to a sampling error approximated by their square roots.

Cases 2, 3 and 4 group all the low, middle and high distributions.

l The low distributions cause no failures.

The middle distributions give 2126 base failures, and one bolt failure leading to a large missile with velocity greater than 50 m/s and less than 90 m/s..

This result is very similar to the nominal PWR1 case of Reference

'll which gave 26% base failures and no bolt failures.

This is coincidental because of the different assumptions used.

The Case 3 result however differs from Case 1, in which the input distri-i butions have the same means but larger widths.

Case 1 permits parameter combinations leading to larger explosions than the largest possible in Case 3,

and so leads to more vessel top and containment failures.

Case 4, grouping all the high distributions gives 10000 base failures, 9987 bolt failure and 9959 with missile velocity greater than 90 m/s.

The sharp rise in failure proba-bilities between Cases 3 and 4 is at first sight surprising as it appears to indicate a chance coincidence of a threshold with the boundary chosen between these cases (this boundary corresponds to an explosion of energy 2218 MJ).

However it should be noted that the explosion energy is less densely sampled near its extreme values in each case because these extremes correspond to the coincidence of extremes in pour diameter, pour length and conver-sion ratio.

As indicated in Section 3 all these distributions are within the bounds of possibility.

So also are all their combinations.

Since these combinations cover such a wife range of calculated probabilities it is necessary to investi @ further combinations to see which parameters are most influential.

36

-~

- ~ ~ - - - - - - - - - - - - - - -

i.

i The next ten cases, 5 to 14 may be regarded as perturbations about Case 3 in which all the middle distributions were used.

Cases 5 to 9 change one distribution from middle to low at a time, and Cases 10 to 14 change one from middle to high.

Cases 5 to 9 show that any low distribution suppresse bqlt fail-ure if combined with the other middle distributions.

The changes have markedly different effects on bottom failure (2126 in Case 3).

Pour diameter has the greatest effect, giving zero failures, followed by pour length and conversion ratio (4 and 5, insignifi-cantly different) and then fraction of core molten with 172.

Finally, changing the distribution of slug condensed phase frac-tion has no significant effect (2087 failures).

These results are easily understood because bottom failure only depends on the explosion energy which is proportional to pour diameter squared, the pour length and the conversion ratio.

It is unaffected bg slug composition and is affected by core fraction molten only to the extent that this imposes a cutoff on the melt mass calculated from the pour geometry.

Table I shows that changing to the low

~

distribution of core molten changes the range of melt mass in the explosion from 7000 - 56000 kg to 0 - 31000 kg whereas changing to the low distribution of pour length restricts the melt in the explosion to 0 - 7000 kg.

Cases 10 to 14 perturb from Case 3 in the direction of greater damage.

The ordering of importance, measured by the change from the base case value in Case 3, for bottom failure, is similar to that from Cases 5 to 9.

Pour diameter has the greatest effect followed by pour length and conversion ratio close together; and fraction of core molten and slug composition have no significant effect.

Fraction of core molten is now less important because it only makes a small change in cutoffs imposed on the mass in the explosion (see Table I).

The relative importance of these changes for vessel top failure is different however.

The largest changes are now caused by changing the conversion ratio distribution, followed by pour dia-meter, followed by slug condensed phase fraction and pour length.

l "'

Changing the distribution of fraction of core molten had no sig-nificant effect.

This provides an illustration of the fact that the importance ranking of uncertainties can depend on the particu-lar quantity that is of interest.

Cases 15 to 19 are perturbations from Case 4 in which all the high distributions were combined yielding nearly 100% failures in all categories.

One low distribution at a time is now used.

For vessel bottom failure the largest change is caused by. changing the pour diameters, then fraction of core molten and conversion ratio (insignificantly different), then pour length.

Slug compo-sition caused no change.

These changes can all be understood by considering the explosion energy.

For-bolt f ailure,ajd missile velocities above 50 m/s changing the pour diameters stf11 produces 37

the largest changes, followed now by conversion ratio, pour length and fraction of core molten, with slug composition still producing the smallest change.

For missile velocities above 90 m/s, the changes in pour diameter, conversion ratio and fraction of core molten completely suppress missile formation, with pour length yielding the next largest change from Case 4 and slug cpaposition still yielding the smallest change.

Cases 20 to 24 perturb from Case 2 in which all the low distribu-tions were combined, leading to no failures of any kind.

These cases show that this result is unchanged by using any one high distribution.

Summarizing the main calcula,tions, the relative importance of the

^

various parameters was found to depend on which particular kind 4

of failure was investigated, and on which base set of distribu-tions was perturbed.

Generally the parameters directly defining'.

the explosion energy, pour length and diameter and conversion ratio, were most important.

Often the pour diameter had the largest influence, because it enters aguared into the expression i

for explosion energy.

The fraction of core molten often turned

~-

out not to be important because with the modeling and distribu-tions used it acts as a cutoff on the mass of melt in an explo-sion; and in many of the cases sampled was either not effective 3

or not the dominant restriction on the mass of melt in the explo-sion.

To some extent this effect is an artifact of the way this model is parameterized.

The slug condensed phase fraction does not affect vessel bottom failure; it sometimes significantly affected top failure but always ranked low among the five para-meters investigated.

4.3 Results of Additional Calculations In these calculations, values of the second set parameters were varied one at'a time within their uncertainty ranges and combined with each of all the low, middle and high distributions of the first set parameters.

Thus the effect of these changes over the whole range of the first set parameters is explored.

Comparison with cases 2, 3 and 4 allows the importance of changes in the

~

second set parameters to be compared.

Additionally, the effect l

of changing the upper limit of three first set parameters, frac-tion of core molten, conversion ratio and pour diameter, was investigated.

These calculations are set out in Table V.

I e.

38

]

I I

TASI.E V.

.Actditional Calculations INPUT Fall.URES CAI.CULATIONS (per 10,000 trials)

Case Comparer Fraction Pour Pour Slug

• Conversion He.a n Mean Mean Mean Slug

• Vessel thalts Large I.arge with Molten Diametes Length Cosplensed Ratin Emplosion Slug

• Slug

• Mass Botton Missile Missile Case (1)

(m)

(m)

Ph.ese (5)

Energy Impact Volume (1000 kg)

V >50 V>90 hunt.c r Fractson (MJ)

Enesgy (m3) m/s e/s (MJ)

Heat 25 2

L L

1.

L 1.

7 1

9.3 16.9 0

0 0

0 Content = 26 3

M M

M M

M 484 258 45.3 67.7 220 0

0 0

0.8 MJ/kg 27 4

H H

18 at H

2759 1320 22.6 53.8 10000 9394 9394 5799 Hett 28 2

L L

L L

L 15 2

9.3 16.9 0

0 3

0 Contrnt = 29 3

M M

M M

M 965 498 35.1 56.1 4118 89 89 to 1.6 MJ/kg 30 4

H 11 H

H H

$109 2865 22.6 53.8 10000 10000 10000 10000 All Un-31 2

L L

L L

L 11 1

9.3 16.8 0

0 0

0 cised Melt J2 3

M M

M M

M 106 24 16.6 50.3 2094 1

1 0

cnd Water 33 4

H H

H 98 H

3826 2086 22.6 55.8 10000 9993 9993 9966 Abow?

A11 Un-34 2

L L

L L

L 11 1

9.2 16.7 0

0 0

0 Cated Melt 35 3

M M

M M

M 732 424 39.9 59.8 2099 2

2 0

W cnd Water 36 4

H H

H H

H 3830 2123 22.4 52.4 10000 10000 10000 9999

' e Solow V ri:: tion 37 20 75-100 L

L L

L 112 3

46.8 111.0 0

0 0

0

$f Frac-38 10 75-100 M

M M

M 749 344 54.6 124.0 2252 0

0 0

tton 39 4

75-100 H

H H

H 4888 2518 26.0 74.4 10000 9753 9753 9677 Molt:n V*ristion 40 2

L L

L L

0-5.3 36 5

9.3 16.9 0

0 0

0 cf Conver-41 3

M M

M M

5.3-10.7 2325 1136 26.2 46.4 9284 4304 4304 3053 clon Ratio 42 4

H H

H H

10.7-16.0 32292 7211 22.7 53.8 10000 10000 10000 10000 tower 43 2

L L

L L

L 11 3

9.3 16.9 0

0 0

0 P13 rum 44 3

M M

M M

M 729 243 29.3 49.6 7101 0

0 0

Failure 45 4

H H

H H

H 3838 2094 22.6 53.8 10000 9996 9996 9962 500 W A

Lower 46 L

L L

L L

11 1

9.3 16.9 0

0 0

0 Plenum 47 3

M M

M M

M 733 490 45.3 67.7 254 0

0 0

Failrre 48 4

H H

H H

H 3821 2004 22.6 5J.8 10000 9994 9994 9953 1500 MJ small 49 16 H

0.0-0.075 H

H H

1.3 0.1 12.9 78.2 0

0 0

0 Pour 4

0

Cases 25 to 30 explore variation of the melt heat content.

This enters into the equations modeling the explosion on exactly the same footing as the conversion ratio, that is, only in a product of terms defining the explosion energy.

Thus it is expected to be rather important.

Case 26 shows a drop from 2126 in. case 3 to 220 vessel bottom failures caused by reducing the heat. content from 1.2 to 0.8 MJ/kg.

In Case 27 the 10000 vessel bottom fail-

~

ures are unchanged, the 9987 missiles with v > 50 m/s only fall to 9394 but the 9959 missiles with v > 90 m/s fall to 5799.

In the Presence of the threshold effects in this problem, the effect of relatively small changes such as this may or may not affect a result depending upon whether they cause a large number of trials to move from one side of the threshold to the other, t

Increasing the heat content to 1.6 MJ/kg. Cases 28 to 30, increased the bottom failures to 4118 (from 2126) and produced 89 bolt failures with missile velocity > 50 m/s instead of 1 in Case'.

3.-

Ten of these had missile velocitiec over 90 m/s (zero in Case 3).

The increase in heat content was enough to produce 10000

'l missiles over 90 m/s in Case 30 compared with 9959 in Case 4.

J~

The isomorphism of the problem to conversion ratio and heat con-tent means that the calculated results can be used to make further predictions; for example, results similar to Case 29 would be expected if a heat content of 1.2 MJ/kg were combined with a flat distribution of conversion ratio in the range 2.2 to 4.4%.

Cases 31 to 36 examine the effect of changing the assumption that the water that does not participate in an explosion lies below I

the explosion, and any unmixed melt lies above.

In Cases 31 to 33 all the unmixed melt and water is located above the explosion, and in Cases 34 to 36 it is all below.

Neither change alters the results of Cases 2,

3 and 4 significantly.

This is because for explosions large enough to cause bolt failure there is little or no unmixed water; for the middle distributions, explosions strong enough to cause bolt failure will involve almost all the melt; and for the large distributions, again most of the melt is mixed.

This insensitivity to. the partition of material between the upward and downward moving slug means that the results are also insensitive to the assumed mass of water participating in the explosion.

In the model used here, this water mass only affects the up/down partitioning.

Cases 37 to 39 explore the effects of fractions of core molten higher than 75%.

Cases 37 and 38 which use a flat distribution from 75 to 100% show no significant difference from Cases 20 and 10 in which the range is 50 to 75%.

Case 39 shows a very small reduction in bolt failure compared with Case 4 probably caused by increased tamping by unmixed melt over the explosion leading to lower slug velocities.

a e

40

Cases 40 to 42 examine the effect of increasing the conversion ratio upper limit from 5 to 16%.

These three cases use the same distributions as Cases 2, 3 and 4 except that the low, middle and high thirds of the range 0-16% are used for the conversicn ratio.

Case 40 shows that conversion ratios up to 5.3% are not sufficient to overcome the combined effect of the other small distnibutions.

This is consistent with Case 24.

Case 41 shows that a substantial number of failures of all kinds - 9284 vessel bottom and 3053 bolt f ailures with missile velocity greater than 90 m/s - are produced by combining all the middle distributions with conversion ratios from 5.3 to 10.7%.

These are the highest numbers obtained in this study frcm any single change from Case 3 (all middle dis-tributions).

Case 42 using all large distributions and conversion ratios from 10.7 to 16.0% qives, as would be expected, 10000 failures in each category.

Cases 43 to 48 examine the effect of using different values f or'-

the energy required to fail the vessel bottom.

As would be expected, this affects the number of vessel bottom failures where this is not 0 or 100%; Case 44 with a 500 MJ threshold gives 7101 c-failures, compared with Case 3 using 1000 MJ giving 2126 and Case 47 using 1500 MJ yielding 250 failures.

The lack of any effect on the numbers of bolt failures is presumably because explosion energies up to 1500 MJ would not cause bolt failure even without vessel bottom failure.

Case 49 explores the effect of restricting the pour diameter to the size of one of the holes in the lower core plate.

The maxi-num melt mass implied is 93 kg which by a wide margin is insuffi-cient to damage the vessel.

This mass is also similar to the limit proposed by Henry and Fauske [7, 25).

No failures were predicted.

To summarize the results of the additional calculations, varying the position of unmixed melt and water, varying the maximum frac-l tion of core molten and varying the vessel bottom failure thres-l hold did not significantly affect bolt failure or missile forma-tion; varying the melt heat content had significant effects; and varying the maximum conversion ratio had a substantial effect.

I c.

41

5.

Other Areas of Uncertainty 5.1 The Effects of Hich Pressure The calculations and models described above all refer to steam explosions at ambient pressures at or near atmospheric.. However, many important PWR accident sequences involve pressures up to about 17 MPa, the set point of the primary system safety valves.

For example-in the Zion Probabilistic Safety Study the frequency of core melt following a large break loss of coolant accident is calculated to be 1.15 x 10-8 per year [45).

These are the sequences in which the pressure in the RPV is expected to be near to atmospheric.

The total calculated core melt frequency for Zion is 4.21 x 10-5 per year (45).

Thus, the low pressure sequences are calculated to be 27% of all core melts for Zion.

This percentage is uncertain and plant-specific.

The experimental data on steam explosions at elevated pressures are very sparse and inconclusive.

Single droplet experiments indicate that, for.vnstant water temperature, the triggering of explosions becomes easier for pressures above 0.1 MPa until about J-0.8 MPa [33, 38, 46].

At 1.0 MPa, explosion triggering is com-parable again to the 0.1 MPa case.

At 1.1 MPa (the limit of the apparatus), triggering becomes slightly more difficult than at 0.1 MPa.

At intermediate scale, 5.4 kg delivered to the water at an ambient pressure of 1.09 MPa did not explode spontaneously

[17].

An explosion was, however, triggered with a detonator.

Experiments have been conducted at Ispra which resulted in externally-triggered explosions under ambient conditions as high as 3.0 MPa (47].

Based on these and other data and models, it has been assumed that spontaneous triggering of steam explosions I

becomes less likely as the pressure increases, although explosions can still be induced by sufficiently large external triggers.

While some external triggers, falling objects for example, may be found during reactor accidents, it is not known what trigger.

strength is required as a function of ambient pressure, nor what

(

triggers will be available with what frequency.

Although the extrapolation of smalland intermediate-scale data at relatively low ambient pressures to large-scale events at much higher pressures seems plausible, it could conceivably be quite wrong.

Single-droplet experiments show that explosion triggering becomes more difficult if noncondensable gases are present (hydro-gen, oxygen, air) in the film around the droplet, or if the water subcooling is low.

Intermediate-scale tests indicate that these suppressive mechanisms are not operative when the volume of the melt delivered is above a certain threshold (18].

It.is not inconceivable that the suppressive effects of high ambient pres-sure might also be overcome at larger scales.

There are simply no reliable data in this regime.

d.

42

The following summarizes the effects that must be considered in any study accounting for high ambient pressure:

1.

Trigger strength required as a function of pressure and scale.

i 2.

At higher pressures, the volume production rate of steam in a coarse premixture will be lower than at low pressure, so that any limitation on mixing caused by steam generation may be weaker.

3.

Small-scale results indicate that conversion ratio increases i

with ambient pressure (26, 38].

This may also be true at i

larger scale.

4.

If the primary system is under pressure, the additional pressure increment to reach the failure threshold will be lower.

s j

5.

If lower plenum f a'. lure occurs, additional blowdown forces may contribute to the vessel's subsequent motion. [5]

J-a 6.

Variation of material properties of water as a function of pressure.

i Item 1 above would have the effect of reducing the calculated probabilities, possibly to zero, because of the possible improba-I bility or impossibility of triggering steam explosions at high pressure.

Items 2 through 5 on the other hand, have the poten-tial to increase the calculated probabilities of failure.

Thus the effects of uncertainties in steam explosion behavior and j

effects at high pressure may be either to increase or decrease the probabilities of vessel and containment failure calculated in Section 4.

This is similar to the position adopted in Squarer's probabilistic analysis; he did not assign a probability for sup-l pression of steam explosions at high pressure [12].

1 5.2 Uncertainty in Head Becomino a Missile In this calculation, we have assumed containment failure due to impact by the vessel head.

This failure mode requires that the head become a missile with a > 50 m/s velocity.

-If the head is to become a miosile, failure must occur at the bolts rather than at the top of the vessel top head.

As discussed in Appendix B, it is uncertain whether the actual failure location is at the bolts or the top head.

A second necessary condition is I

efficient coupling of the slug energy to the head.

This requires that all the studs fail at approximately the same time.

If this does not happen, the head may "can open" and the slug will con-tinue, leaving the head behind.

d 43

!~

4

,i If failure at the top of the vessel top head occurs before, or 1

instead of, bolt failure, it is very uncertain what effect this would have on the failure probabilities calculated in Section 4.

1 If the threshold for top head failure is lower than that for bolt failure, top head failure would generally have higher probabili-ties than those indicated for bolt failure.

If there is to be i

any possibility of direct containment failure, a missile ~ is required.

This could either be a fragment of the top head, or the slug.

Fragmentation of the top head, as distinct from the f ormation of flaps (open can-lid), might occur because of the substantial nonuniformity of the top head.

The size and speed of any such fragments would be difficult to estimate.

This would make their penetrating capability very uncertain.

A similar situation could occur if, instead of the studs failing at approxi-mately the same time, the studs fail in a zipping pattern.

If this happens, the head could be spinning as it flies upward.

In this situation, it is uncertain how much of the slug energy would'-

4 be. transferred to the head and what would be the consequence of a j

spinning head that may fly sideways to impact containment.

The potential for the slug itself to be a damaging missile would 4

1 appear to depend on whether it remains coherent or spreads out.

This depends on details of the failure, and the slug flow pattern and so also is very uncertain.

The uncertainties in large missile formation that are due to uncertainty in the details of the failure processes at the top of the vessel may thus be bounded by two possibilities.

On the one hand, formation of a large missile with penetrating power suffic-

)

ient to breach containment may occur according to the criteria in l

Subsections 3.14 and 3.15.

On the other hand the alternative mechanisms discussed in this Subsection and in Appendix B may always prevent the formation of missiles capable of damage.

5.3 MM1tidimensional and Geometric Effects The modeling described in Sections 3 and 4 of this report only I

accounted for the gross geometrical features of the reactor pres-sure vessel and its internals.

In particular, mixing of melt-flowing from the core into the lower plenum with residual water there was assumed to be unimpeded except for the uncertain influ-ence of steam production.

Also the model of slug formation and i

propagation was one-dimensional.

Geometrical features of the vessel can be identified which might affect the correctness of these assumptions and which contribute uncertainty.

These are discussed in this subsection.

1.

Almost all steam explosion experiments to date have been conducted in relatively uncluttered vessels.

The lower plenum region of a PWR is relatively cluttered compared with these experiments.

l (The bottom of a BWR vessel is much more cluttered than g PWR.)

This clutter may tend to inhibit the coarse mixing proep_s prior i

to an explosion, by restricting lateral mixing [33].

ThW~

l r

l 44 I

--.v.____-_,--,mm._,_m - _,,._,..,_.-_, _ _,,,.,,, _.

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

I increased surface area might also tend to trigger a number of (smaller) steam explosions, rather than a single large one.

These possibilities have not yet been investigated experimentally.

It is also conceivable that the presence of clutter (control rod tubes, instrumentation tubes, grates, etc.) could enhance mixing.

and increase the amount of mass mixed and the ultimate explosion energy.

The diffuser plate in a PWR might increase the degree of mixing of any melt passing through it.

Furthermore, turbulent wakes and vortices might develop as the melt passes over and by various surfaces.

This turbulence could increase mixing.

In premixed gas phase combustion, obstacles and clutter can greatly increase the burning rate because the turbulence generated by those structures enhances mixing in front of a flame [26, 33].

Because the effects of lower plenum clutter are not known, they were not modeled in this study.

The actual vessel geometry is much more complex than the simple'-

one-dimensional approximations employed in this study.

Under some conditions, an explosion in the vessel lower plenum could vent up the downcomer annulus as well as up the core barrel.

It is pos-

_~

sible that such venting would ameliorate the forces on the upper

head, but a detailed multidimensional calculation would be required to quantify this effect.

Two-dimensional calculations have been performed with both the SIMMER [3] and CSQ [33] codes.

4 The SIMMER calculations identified important effects in the down-i comer; water speeds of ~200 m/s were associated with explosions

~[

of peak energy ~1000 MJ and water slug impact peak pressures at the top of the downcomer were 30-100% of those calculated at the j

top head.

I In the CSQ calculations "a

small portion of the water slug" was forced up the downcomer [33].

Any difference between these results and those in the ZIP study is probably caused by differ-ent assumed boundary conditions.

Further calculations of this l

kind would be needed to investigate the implications of downcomer flow more fully.

It is thus clear that multidimensional and geometric effects have the potential for both aggravating and mitigating the consequences of in-vessel steam explosions.

Their neglect is thus a potential cause of underestimation of uncertainty, although in this study it is not important because of the wide range already identified.

5.4 The Effect of Correlations The sampling from distributions described in Section 4 assumed that all the sampled parameters were independent; that is to say that knowledge of one of them does not change our knowledge of any other.

If this assumption is wrong the affected pagameters I

would be correlated.

Our state of knowledge about th$dh para-meters is consistent with either the presence or the a6sence of correlations.

45 l

The qualitative effect of some potential correlations may be discussed by simple arguments.

For example, it is possible that the heat content of the melt is correlated with fraction of core molten because a larger core melt may take longer to accumulate and hence may also accumulate more decay heat per unit mass.

In i

comparison with Case 3 in Table IV, in which the melt heat content was constant, the range of possible explosion energies would be widoned if heat content and fraction molten were correlated in i

this way.

This is because massec of melt in explosions that were j

limited by small fractions of core molten would be combined with 4

small melt heat contents; and explosions involving the largest melt masses permitted in Case 3 would have high melt heat con-tents.

This effect on the tails of the explosion energy distribu-tion would be particularly significant for the very low proba-bility failures, increasing their probabilities to values inter-mediate between those of Cases 3 and 32 (in which the highest value of the heat content has used throughout).

The effects of potential correlations between parameters would be either to increase or

~

to decrease the probabilities calculated in Section 4.

Thus omitting potential correlations from our numer-ical calculations caused a potential understatement of the over-all uncertainty in the probabilities.

5.5 Effects of Model Parameterization The numerical results of the calculations described in Section 4 might be used to draw the following conclusions:

that the proba-bilities of vessel failure and containment failure are uncertain over the range from 0 to 1; that estimates of'these probabilities obtained from " middle" assumptions (Case 3) are about 21% and i

10-4 respectively; and that the most important contributing uncertainties are those in the pour diameter and conversion ratio.

These potential conclusions may depend on the arbitrary choices made of model parameterization and distributions.

It is there-4 fore.necessary to consider whether the same conclusions would have been obtained, had different parameterizations or distributions been chosen.

First we consider the uncertainty ranges for the failure proba-bilities.

Obviously a different parameterization, or a different l

set of combinations of distributions, could produce different ranges.

For example, if all input distributions extended to the lower end of the parameter uncertainty ranges, having different upper limits, every combination of such distributions would include explosions of low energy which would cause no damage.

Hence, all calculated damage probabilities would be less than one.

This was illustrated in some preliminary calculations for this study published in ~ ref erence 48.

In those calculations lower j

limits of zero were used in all distributions for some pagameters.

4 Additionally the conversion ratio distributions were tI1 angular (as in reference 11).

Failure probability ranges of 0 't'o'~51% for 46

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

t I

the vessel and 0 to 33% for the containment were calculated.

However if such a set of calculations were the only ones available it would be necessary to consider whether the full range of proba-bilities had been obtained.

The ranges of probabilities set out in the present paper show that any narrower range would not include the full range of possibilities.

Second, we consider whether the " middle" probabilities obtained in Case 3 would be expected to change if different distributions or parameterizations were used.

It is clear that they could change substantially in either direction.

In comparison with the values in this paper (21% for the vessel and 0 to 10-4 for con-tainment), the corresponding preliminary calculations in reference 4

48 yielded 1.5% for the vessel and O for containment, from a different " middle" set of distributions.

Additionally, a differ-ent parameterization of the problem has been suggested [49] in -

which the area of the melt pour out f rom the core is used instead-of its diameter.

If the area parameterization is used, the middle third of the range of pour areas is 3.0 to 6.0 m, and if this 2

is combined with the other middle distributions used in this

~

report a mass range of 21,000 to 63.000 kg is obtained.

This should be compared with the middle mass range of 7000 to 56,000 kg in Table I.

The main effect of this parameterization change would therefore be to eliminate smaller melt masses (from 7000 to 21,000 kg) from the combination of middle distributons.

The result would be intermediate between Case 11 (mass range 28,000 to 63,000 kg) and Case 12 (14,000 to 63,000 kg) in Table IV, and so calculated failure probabilities between 53 and 83% for the vessel and 0 and 1.8% for containment would be expected.

Thus changes in distributions or parameterizations can substantially i.

change the probabilities calculated from a " middle" combination, in either direction.

Such probabilities must therefore be con-j sidered essentially arbitrary.

Finally, we consider whether the uncertainties in pour diameter and conversion ratio would continue to have the highest impor-tance under a different choice of parameterization or distribu-tions.

Here, two points need to be made.

First, under some i "'

different parameterizations the particular parameters discussed here might not be used; for example pour diameter is not expli-citly included if the area representation described in the pre-vious paragraph is used.

In that case, the corresponding area parameter would assume high importance.

More generally, the mass of melt participating in the explosion is important, because together with the conversion ratio it strongly affects the total explosion energy.

The second point about relative importance of I

f e.

47

=

parameter uncertainties is that such importance, if measured by the probability changes caused by changes of the parameters from one base case, will in general depend on the choice of the base case.

The present calculations may be described as variations about three base cases (Cases 2,

3 and 4) that are rather widely distributed over the whole parameter space.

Additionally, the important parameters found in these calculations are the same as those found in the preliminary calculations [48].

However, it is possible that in some unexplored part of the overall parameter space other parameters (like slug void fraction, or melt heat content) may assume high importance.

It is therefore necessary to qualify the important uncertainties identified in the current calculations by noting that other uncertainties might be shown to be also important in parts of the parameter space not examined.

e w

l a

e 48

6.

Discussion The calculations in this report refer to in-vessel steam explo-sions at ambient pressures near to atmospheric.

Numerical values were taken from.the Zion reactors.

They show that the c,onditional probability of containment failure, given core melt during a low-pressure accident, is extremely uncertain.

Indeed the results span the range of probability from 0 to 1.

This uncertainty estimate is derived from the particular choice of distributions j

and combinations thereof used.

Adequate evidence does not exist at present to exclude any of these combinations, or the proba-bilities calculated from them, and such evidence would be required in order to establish a narrower range of probability.

If all the

" middle" distributions in this study are combined, a value of ~

10-4 is obtained.

This however should not be used as a best estimate of the fraction of core melt accidents leading to con-tainment failure by steam explosions, because a different para '.

meterization of the problem could give a completely different number and because it is derived from single assignments of sub-jective probabilities.

The effect of alternative assignments needs to be considered, and this leads to the range of results

~

calculated here.

Examination of Tables IV and V shows that the criterion of an explosion energy > 1000 MJ for vessel failure at the base led to a significant probability of such failure for many of the cases 1

sampled.

The uncertainty in this probability also covers the range from 0 to 1.

The possibility of explosive vessel failure should be taken into account when planning action in response to core-melt accidents that still have the potential for recovery to a coolable state in-vessel.

1-1 Extension of these results to higher pressures would in principle require reformulation of the problem to account for the different 4

characteristic's of triggering and possibly mixing.

However in i

practice the range of uncertainty can be explored by qualitative

'~

arguments:

on the one hand steam explosions may be impossible in.

reactors above some value of the pressure,'in which case the pro-bability of containment failure by this mode would be zero.

On the other hand effective external triggering may be probable, in which case the current calculations would have to be modified to take account of the effects listed in subsection 5.1.

Some of these effects, namely possibly easier mixing, possible conversion ratio increases increased ease of vessel failure, and blowdown forces from vessel failure at pressure, have the potential to increase failure probabilities.

Thus extension to higher pres-sures introduces effects that may reduce, and others that may increase failure probabilities.

The uncertainty inte'rvals estimated for the probabilities would therefore be unchanged.

e 4

r j-

[

49 i

-m,-__-_,~.

m

. -...-m..

l Extension of the results to other plants may in some cases in practice be possible by rather simple comparisons of dimensions to determine whether any significant differences exist, and if so, whether they are large enough to affect the results of the current calculations materially.

The extensive sensitivity study presented here shows that the uncertainties in two parameters out of the ones used in the model are highly important:

pour diameter and conversion ratio.

The prominence of the first of these is to some extent an artifact of the model, in that it appears squared in the expression for explosion energy (other parameters appearing linearly).

A more 1

general statement would be that the mass of melt participating in an explosion is highly important.

This mass is in turn determined by two highly uncertain processes:

the process of core melting which may or may not produce and release a large pool of melt coherently; and the process of melt

- water mixing which may or may not be effectively self-limiting due to steam production.

The conversion ratio is uncertain because it is not known whether this parameter decreases, remains within the bounds of current measurements or increases as the melt mass increases from kilo-j grams to thousands of kilograms.

An additional factor influencing the probability of containment failure, that was not accounted for in the sensitivity study, is the question whether the interaction of a slug with the top of 4

the vessel can produce damaging missiles or not.

Since the uncertainty in the vessel failure modes that determine the answer to this question can reduce the containment failure probability to zero, this uncertainty is of high importance.

Thus four of the most important contributors to the uncertainty i

in the probability of containment failure due to steam explosions are the conversion ratio, the mass of melt participating in the i

explosion, the likelihood of triggering at high pressure and the i

failure mode of the vessel top head.

Because this study is based on a finite sampling from a parameter space, other uncertainties may also be important.

Substantial reduction of any of these l

important uncertainties would, if the result were favorable, sub-stantially reduce the uncertainty in the probability of contain-ment failure due to steam explosions.

For a significant contain-ment failure probability, either a significant probability of conversion ratios higher than currently measured or a significant probability of large masses of molten core actively participating i

in an explosion would be needed.

Additionally, triggering in the l

pressure range of importance and large missile formation would have to be possible.

a 50

REFERENCES 1.

N.

C.

Rasmussen (Ed.), Reactor Safety Study:

An Assessment of Accident Risks in US Commercial Nuclear Power Plants.

WASH-1400, NUREG 75/014 (Washington, D.

C.,

October, 1975).

2.

W.

B.

Murfin (Ed.), Report of the Zion / Indian Point Study

~

Vol.

I, NUREG/CR-1410, SAND 80-0617/1, (Albuquerque, NM, August, 1980).

3.

M. G. Stevenson (Ed.), ReDort of the Zion / Indian Point Study!

Volume 2.

NUREG/CR-1411.

LA-. 8 3 0 6 -MS, (Los Alamos, NM, April, 1980).

4.

The German Risk Study Summary (Cologne, Germany:

}

Gesellschaft fuer Reaktorsicherheit abH, 1979).

5.-

J.

H. Gittus (Ed.), PWR

Dearaded Core Analysis,

NDR-610(S),

(Springfields, UK. April, 1982).

6.

Steam Explosions in Licht Water Reactors, Report of the Swedish Government Committee on Steam Explosions, Ds1 1981:3 i

(Stockholm, 1981).

i-7.

R.

E.

Henry and H.

K.

Fauske, Required Initial Conditions for Eneraetic Steam ExDlosions, presented at ASME Winter Meeting, (Washington, D.

C.,

November, 1981).

8.

T.

G.

Theofanous and M.

Saito, An Assessment of Class 9 (Core-Melt) Accidents for PWR Dry Containment Systems Nucl.

i Eng. Des. if., 301-332, (September, 1981).

9.

Preliminary Assessment of Core Melt Accidents at the Zion i

and Indian Point Nuclear Power Plaats and Stratecien for Miticatina Their Effects.

U.

S.

Nuclear Regulatory Commission, NUREG-0850, Vol.

1 (Washington, D.

C.,

November, 1981).

10.

F.

Mayinger, Kernschmelzunfall:

Wie sind Damoflexolosigge_n in Licht neuerer Erkentnisse zu beurteilen? Zu hoealichkeit.

Ablauf und Wirkuna, Atomwirtschaft, pp. 74-A1, (February, l

1982), English translation:

UKAEA Risley Trans 4636.

11.

D.

V.

Svenson and M.

L.

Corradini, Monte Carlo Analysis of LWR Steam ExDlosions, NUREG/CR-2307, SAND 81-1092, (Albuquerque, NM, October, 1981).

m.

51

o 12.

D.

Squarer and M.

C.

Leverett, Steam Explosion in Perspective, Proc.

Int. Meeting on LWR Severe Accident Evaluation Cambridge, MA, August 1983, pp. 6.1-1 to 6.1-9.

l 13.

7.

A.

Dullforce, The Spontaneous Tricaerina of Small Scale Vapour Explosions, Ph.D.

Thesis, University of Asto,n in Birmingham, (October, 1981).

14.

Commonwealth Edison Company, Zion Station Final Safety Analysis Report, Docket 50295-16 and 50304-16, (Chicago, IL, December, 1970).

15.

L.

D.

Buxton and W.

B.

Benedick, Steam E..Dlosion Efficiency Studies.

NUREG/CR-0947, SAND 79-1399, (Albuquerque, NM, November, 1979).

16.

D.

E.

Mitchell and N.

A. Evans. Intermediate Scale Steam '-

Explosion Phenomena FITSB

Series, NUREG/CR-0000, SAND 83-1057, (Albuquerque, NM, 1984, to be published).

17.

D.

E.

Mitchell, M.

L.

Corradini and W.

W.

Tarbell,

~

Intermediate Scale Steam Exolosion Phenomena:

ExDeriments and Analysis, NUREG/CR-2145 SAND 81-0124, (Albuquerque, NM, September, 1981).

(

18.

Memorandum from M.

Berman and N.

A.

Evans to R.

W. Wright,

[

USNRC, Effects of Scale On Steam ExDlosions March 31, 1983.

19.

M.

L.

Corradini, Analysis of Molten Fuel-Coolant Interactions.

Proaress ReDort for the Period Antil to l

September 1982, UWRSR4 (Madison, WI, 1982).

l 20.

M.

Berman (ed), Licht Water Reactor Safety Research Procram Semiannual Report. April September 1982 NUREG/CR-3407, SAND 83-1576, (Albuquerque, NM, 1983).

21.

M.

J.

Bird, Thermal Interactions Between Molten Uranium Dioxide and Water. An Experimental Study Usina Thermite i

Generated Uranium Dioxide, Presented at ASME Winter Meeting.

l (Washington, D.

C.,

November, 1981).

22.

M.

Berman and R.

K.

Cole, Status of Core Melt Procrans -

May-June 1983. Memo to T.

J.

Walker and S.

B.

Burson, USNRC (Albuquerque, NM, October 1983).

23.

M.

Berman and R.

K.

Cole, Status of Core Melt Procrans -

July-Auaust 1983 Memo to T.

J.

Walker and S.

B.

Burson, USNRC (Albuquerque, NM, October 1983).

24.

M.

Berman (ed), Licht Water Reactor Safety Research Procram Semiannual ReDort. April-SeDtember 1983

( Albuquy~ glue, NM, forthcoming).

~

52

25.

H.

K.

Fauske. Scale Considerations and Vapor Explosions (Rapid Phase Transitions), Presented at LNG Safety Workshop, MIT, (March, 1982).

26.

M. Berman (Ed.), Licht Water Reactor Safety Research Procram Semiannual Report.

October.

1981

March, 1982, NUREG/CR-2841, SAND 82-1572, (Albuquerque, NM,

Decedber, 1982).

27.

M.

L. Corradini, A proposed model for Fuel-Coolant Mixinc.

Trans. ANS 11, 415-416, (1982).

28.

M..L.

Corradini, Proposed model for Fuel-Coolant Mixina durina a

Core-Melt

Accident, Proceedings of the International Meeting on Thermal Nuclear Reactor Safety.

Chicago, IL, August 1982, NUREG/CP-0027, Vol.

2, pp.

1399-1408.

29.

M.

L.

Corradini and G.

A.

Moses, A Dynamic Model for Fuel-Coolant Mixinc, Proc.

Int. Meeting on LWR Severe Accident Evalution, Cambridge, MA, August 1983, pp. 6.3-1 to 6.3-8.

30.

M.

Berman and R.

K. Cole. Status of Core Melt Procrans -

November-December 1983, Memo to J.

L.

Telford and S.

B.

Burson, USNRC (Albuquerque, NM, February 1984).

31.

M. Berman (ed), Licht Water Reactor Safety Research Program Semiannual Report. October 1983-ADril 1984 (Albuquerque, NM, forthcoming).

32.

G.

A.

Greene et al, Some Observations on Simulated Molten Debris - Coolant Layer Dynamics, Proc. Int. Meeting on LWR Severe Accident Evalution, Cambridge, MA. August 1983, pp.

12.2-1 to' 12.2-7.

f 33.

M. Berman (Ed.), Licht Water Reactor Safety Research Proaram l

Semiannual Report. ADril SeDtember. 1981, NUREG/CR-2481, SAND 82-0006, (Albuquerque, NM, February, 1982).

34.

A.

T.

Chamberlain and F.

M.

Page, An Experimental Examination of the Henry-Fauske Voidina Hvoothesis (University of Aston, Birmingham, UK, 1983).

I i

35.

R. Haag and H. Korber, Zusammenstelluna wichtiaer Eroebnisse und Ableituna von Kenntnislucken zum Problenkreis Kernschmelzen, BMFT-150-400 (Gechingen, Germany, 1980).

_E 53 i

36.

L.

D.

Buxton, W.

B.

Benedick and M.

L.

Corradini, Steam Explosion Efficiency Studies, Part II:

Corium Experiments, NUREG/CR-1746, SAND 80-1324 (Albuquerque, NM, October, 1980).

37.

H.

K.

Fauske and R.

E.

Henry, Interpretation of Laroe Scale Vapor Explosion Experiments with Application to Li6ht Water Reactor (LWR) Accidents, Proc. Int. Meeting on LWR S1evere Accident Evalution, Cambridge, MA, August 1983, pp. 6.5-1 to 6.5-5.

38.

L.

S.

Nelson and P.

M.

Duda, Steam Explosion Experiments with Sinole Drops of Iron Oxide Melted with a CO3 Laser.

Part II.

Parametric Studies, NUREG/CR-2718 SANDil2-1105 (Albuquerque, NM, To be Published).

39.

M.

L.

Corradini and D.

V.

Swenson, Probability of Containment Failure Due to Steam Explosions Followino a Postulated Core Meltdown in an

LWR, NUREG/CR-2214 SAND 80-2132, (Albuquerque, NM, June, 1981).

~

40.

K.

McFarlane, Conservative (Hicks-Menzies) Results for the Efficiency and Work Capacity of an In-vessel Steam Ex_olosion, NST/PWR9(81)l7 (Culcheth, UK, 1981).

41.

V.

E.

Denny and B.

R.

Sehgal, Analytical Prediction of Core Heatup/Liouefaction/Slumpino, Proc.

Int. Meeting on LWR Severe Accident Evaluation, Cambridge, MA, August 1983, pp.

5.4-1 to 5.4-9.

42.

R.

P.

Kennedy "A Review of Procedures for the Analysis and Design of Concrete Structures to Resist Missile Impact Effects," Nuclear Encineerino and Desion, Vol.

37, pp.

183-203 (1976).

43.

R.

P. Kerinedy " Local Missile Impact Effects," Extreme Load Post Conference Seminar, SMIRT 5 Engineering Decision Analysis Co.,

Inc. (1979).

44.

C.

Berriaud, A.

Sokolovsky, R.

Gueraud.

T.

Dulac and R.

Labrot, " Local Behaviour of Reinforced Concrete Walls Under Hard Missile Impact," Paper J 7/9, 4th International r

Conference on Structural Mechanics in Reactor Technoloov, (SMIRT 4, 1977), Nuclear Engineering and Design 45, 457-469 (1978), English translation:

SAND 77-6021 (Albuquerque, NM, 1977).

45.

Commonwealth Edison Co.,

Zion Probabilistic Safety Study, Chicago, 1981.

a

<=

54

- > ~

5-

/, 4 j

y,'

/

46.

L.

S.

Nelson and P.

M. Duda, Steam Explosions of Molten Iron Oxide Drops:

easier initiation at small pressurizations.

Nature 296, 044-846 (1982).

47.

R.

Hohmann, H.

Kottowski, H Schins and R.

E.

Henry.

Experimental Investications of Spontaneous and Triacered Vapour Pixplosions in _ the Molten Salt /WatcJ System.

Proceedings of the International Meeting on Thermal Nuclear Reactor ~ Saf ety, Chicag0.s Vol. 2,.pp.962-971.

'I L, August 1982, NUREG/CP-0027 48.

M.

Berman,-D.

V.

Swenson and A.

J.

Wickett, Trans ANS 45 378-380 (1983).

c.

49.

J.

B.

Rivard, PersScal Communication (1983).

e

'{I i

t

-(

s

/

r I

l

(

of 4)

A f

i 8

N 4

n$

S l

4=

4 f

?

f.

55 f

~

-~,._.s__,.-.

n

~-.--r-

APPENDIX A SUBJECTIVE PROBABILITY 5

Cn A-1

A subjective probability is a numerical expression of an indivi-dual's degree of _ partial belief in the truth of a propbsition.

(A-1 A-2]

In the current case, degrees of belief in propositions such as "the conversion ratio lies between 1.4% and 1.5%" are the basis of the probability distributions.

Textbooks provide opera-tional definitions of subjective probability similar to the follow-ing.

"If an individual would offer betting odds for small bets of 1 to n that a proposition were true and n to 1 that it.were false, then his subjective probability of its truth is 1/(1 4 n)."

The following properties of subjective probabilities follow from the definition:-

1).They comply with the usual laws for combining probabilities.

2)

If sufficient data or evidence exist to justify a classical frequentist probability (fraction of successes out of a large number of trials) the subjective probability must be consis-tent with-it.

3)

If a frequentist probability statement cannot be justified, j

different individuals aware of the same evidence may quote l

different subjective probability values.

This last property, non-uniqueness, means in the circumstances of-the current problem, that any subjective probability distributions of the uncertain parameters are uncertain and must, in an uncertainty study, themselves be varied within the ranges of uncertainty of the parameters that they describe.

'O@g

+

i y=

A-2

REFERENCES A-1 D.

V.

Lindley, Introduction to Probability and Statistics from a Bayesian Viewpoiht (Cambridge University Press, UK, 1965).

A-2 D.

H.

Mellor. The Matter of Chance (Cambridge University Press, UK, 1971).

e e

t n

e f

4 A-3

9 8

6 e

APPENDIX B e

FINITE ELEMENT CALCULATION OF VESSEL FAILURE i

l

\\

I C.

B-1

1

==

Introduction:==

In the main body of this report, a fracture criterion was.used to evaluate failure of the bolts.

To do this, the slug impact pressure load on the head was assumed to be transmitted to the bolts.

The resulting stress was compared with the fracture strength of the bolts.

In this Appendix, we justify this approach and provide additional background into failure of the reactor vessel due to an internal steam explosion.

Similar calculations and a more detailed discussion are given in Ref. B-1.

The goal is to determine the sequence in which failures will occur and hence to provide a basis for choosing failure locations in the vessel.

However, failure prediction under high strain dynamic conditions for as complicated a structure as the reactor vass'el is very uncertain.

We have approached this problem using a simplified finite element model of the reactor vessel.

Calcu-lated stresses and strains were then compared to either a strain

-~ or fracture failure criterion, as appropriate for different parts of the vessel.

Material Properties:

Typical material properties were used in the analysis to obtain ostimates of vessel response.

The vessel is constructed of A533 steel and the bolts are made of SA-540 steel.

Valucs of the caterial properties were obtained from References B-2 and B-3 and o

are listed in Table B-1.

i TABLE B-1 Table B-1:

Material Properties at 2881C (Typical Values from References B-2 and B-3)

YOUNG'S YIELD ULTIMATE STRAIN

• DENSITY FRACTURE 4

~~

MATERIAL MODULUS STRESS STRESS

• AT (kg/m3)

TOUGHNESS 9

6 6

(lO Pa)

(10 Pa)

FAILURE (106 N-a-3/2)

(lO Pa) t VESSEL 177 422 598 0.20 8000 275

'A-533 BOLTS 177 892 1052 0.19 8000 175 SA-540

• These are the Engineering stress and strain, that is, the force divided by the initial area and the deflection dividedy~ty the initial length.

The logarithmic strain at failure, that is, the natural-logarithm of the current length divided by the initial length, is 0.18 for the' vessel.

B-2

~.

~

1 Failure Criteria:

In this analysis, two failure criteria were used:

strain failure and brittle fracture.

Strain failure occurs if a matepial is excessively deformed until voids form and coalesce, leading to loss of strength.

Brittle fracture occurs as a result of flaws in the structure.

If the energy released due to crack growth is greater than the energy to extend the crack, brittle failure occurs.

Based on previous calculations [B-1), the two locations of.likely failure are the top of the head at the centerline and the bolts.

The strain criterion was used at both locations, while the fracture criterion was used only at the bolts.

-(Brittle -fracture of the head is not expected because the head material is more ductile than the bolts.)

Strain failure was evaluated by comparing the calculated effec--

tive plastic strain to the uniaxial failure strain.

The effective plastic strain is defined by:

pg =

[(c1-2) c

(*2 - *3)

+ I'3 - *1} I c

+

l For uniaxial loading, the effective plastic strain is the equal to-the uniaxial strain, but for biaxial loading conditions (as experienced in the vessel head), the use of effective plastic strain leads to failure at biaxial strains smaller than the uni-axial failure strain.

This is consistent with experimental observation [B-4).

The second failure criterion was based on fracture mechanics calculations.

.(For a discussion, see a standard text such as Ref erence B-5. )

For this analysis, the stress intensity factor was calculated using linear elastic-f racture mechanics and a design flaw size recommended by the Pressure Vessel' Research

^

Committee [B-6).

This is a 7.6 mm deep circumferential crack for the bolts.

As shown in Figure B-1, the stress intensity is'a function of the bolt diameter, D,

the unflawed diameter, d.

and the applied load, p.

(or alternately, axial stress in the bolt, c) [B-7).

fD

[1.72 ( ) - 1.27]o K

=

y Substituting values appropriate for the bolt diameter, D = 0.1778 c, and the'unflawed diameter assuming the design flaw, d = 0.1625 0,

we obtain:

e K1 = 0.2020 N-m-3/2 g

Knowing the fracture toughness of the bolt material, Kfrac 175 MN-m-3/2, we can solve for the stress in the bolts to give failure:

B-3 l

1 frac - 870 MPa

1. fracture " 0.202 For the bolts, the fracture stress is below the yield stress.

It can also be shown that the size of the plastic zone at the crack tip is small compared to the bolt diameter.

Because plane strain linear elastic fracture mechanics assumptions are satisfied, the calculated fracture stress is a reasonable estimate of the true fracture stress.

Therefore, it is expected that the studs would fail by brittle fracture if the assumed flaw was present.

If smaller flaws were present (d > 0.165 m), the yield strength of the bolts is exceeded before the fracture stress is reached, and the bolts would likely fail by plastic deformation.

l

_~

t JL Nr

[A D d

~P P*--

U4 n- /'

Kg =

1.72 (h)- 1.27 D3/2 Kg = { h (D)1/2 1.72(h)- 1.27 o Figore B-1:

Bolt Stress Intensity Calculation (B-7]

t l

Numerical Model:

A finite element model was used to evaluate the response of the closure head to impact by material accelerated from below.

The structural model, which represents the reactor vessel above the nozzle center lines, is shown in Figure B-2.

This model was developed using the HONDO II [B-8] computer code which can calcu-late the large deformation, dynamic response of axisymmetric solids.

Because failure of the bolts could lead to a large mass missile (the top head), the bolts were modeled separately from the flanges.

The bolt material properties were reduced go account for the difference in area between-the solid ring in the axisym-metric model and the actual bolt area.

Sliding interfMcas were used between the flanges and between the top flange and the bolt p

nut to give a fairly accurate representation B-4

i

.of bolt / flange behavior during impact.

Based on the Zion FSAR

.[B-9), the bolts were pretensioned to a stress of 290 MPa.

The model did not include the effects of the penetrations at'the top 5

of the closure head.

These penetrations would be expected-to reduce the strength of the top of the head and to increase the possibility of head failure, j

Loading Conditions:

i As described in the main body of this report, we have modeled the i

slug impact as applying an approximately uniform pressure to the vessel head.

This loading is similar to the loading calculated by the Los - Alamos National Laboratory (LANL) using the SIMMER code [B-10).

Using the calculated bolt fracture stress. the bolt -

'orea, and the vessel dimensions, the static pressure in the head req 0 ired to fracture the bolts is approximately 80 MPa.

4 Four finite element calculations were made near to the 80 MPa

'~' fracture pressure.

These included ramp loadings of 60 80 and 100 MPa, and a step loading of 80 MPa.

For the ramp loading, the pressure was ramped to the peak value over 5 ms, held constant

'for-8 ms, and then ramped down to zero in 5 ms.

The step loading was applied for a period of 13 ms.

The purpose of the step load-

}

ing was to examine the effect of dynamic overshoot, since a ramp of 5 ms is sufficiently long relative to the period of natural vibration of the head that it can be considered an essentially static loading.

Because the slug will likely be somewhat diffuse by the time it loads the head, it seems reasonable to expect the loading to be closer to a ramp.

Once again, we should note that these loading conditions are similar to those calculated by LANL

.using SIMMER (B-10).

Results:

r Figure B-3 shows displacement plots for the 80 MPa ramped loading initially and after the pressure has been applied for 0.0013 sec.

~ ~'

Figures B-4 through B-6 show plots of the results used to evaluate fracture for the 60 and 80 MPa ramp cases and the 80 MPa step case.

A summary of all fracture evaluations is given in Table B-2.

For the 60 MPa ramp loading, only small plastic strains occur.

.The bolt stresses do not overshoot the static stresses signifi-contly,. confirming that the ramp loading is essentially static.

.No failure is predicted for this loading case.

l Increasing. the pressure to 80 MPa with a ramp loading causes j

Oignificant plastic strain in the head as shown both ingthe i

displacement plots (Figure B-3) and head strain plot'(Figure B-6).

I i

B-5 i

~

' TOP OF HEAD

FAILURE l

EVALUATJON 2.0 -

STUDS AND NUTS (PRETENSIONED) 1.0 -

E

.A

.---=::

STUDS x

E 0.0-i::::::

s

-r '>

,,l.}.,

s.

p

....r

- 1.0 -

-2.0 -

i i

i 0.0 1.0 2.0 R-AXIS (m) e Figure B-2:

Finite Element Model and Location ofa.

Failure Evaluation B-6

t Table B-2:

Loading Cases Analyzed Using Finite Element Model and Failure Evaluation i

Stud Plastic Head Plastic Bolt Stress

• Strain +

Strain +

CASE Max t,gy Max t,gy Max t,gy g

g g

(MPa)

(ms)

(as) 60 MPa Ramp 690 None 0.0 None 0.0001 None

_ 80 MPa Ramp 950 9

0.0008 None 0.12 None 100 MPa Ramp 1350 4

0.085 None 0.38 9

4 80 MPa Step 1100 2

0.0075 None 0.16 None

• True stress

+ Logarithmic strain (see footnote to Table B-1)

However, failure is not predicted in the head.

The stud stresses Cxceed the fracture stress (Figure B-4) and a relatively small caount of plast.ic strain occurs in the studs.

Based on these results, it appears possible for fracture of the studs to occur without failure at the top of the head.

Whether this will occur is not exactly clear since the penetrations in the head could waaken the top of the head.

Previous calculations [B-1), pre-dicted failure of the head rather than stud failure. The differ-

.once between these calculations is that a more spatially uniform

-loading of the head is assumed here, rather than loading which l

was biased towards the center of the head.

For the 100 MPa ramp loading (Table B-2), failure was predicted i

et both the. studs and the head.

The stud fracture criterion was l

ottained before the head failure criterion.

l Finally, the effect of step loading can be seen by comparing the 80 MPa step loading results to the 80 MPa ramp results.

Step i

Icading of the head causes higher stud stresses and greatep plas-i tic strain in the head.

However, as for the 80 MPa ramp Deading, only fracture of the studs is predicted for the 80 MPa9tep j

leading.

t B-7 l

t

a. INITIAL CONFIGURATION EEE

==

p N

5 lc

~

C ll:::=

b. AFTER 0.013 see LOADING

- ~,

,/,~g

/,

//

\\

\\

/

INITIAL

\\

~/

CONFIGURATION

\\pg

\\'I E

U E

E:::'

6 3

1::::

l2.

lllll llg lll::

ll.

m-Figure B-3:

Head Displacements for 80 MPa Ramp Loading (Magnification = 2)

B-8

W 1

I 80 MPa STEP 1100

/

%/

'\\

\\

/

g f-1000 I

\\

_ j- _ _g l I

I 7

STUD FRACTURE 900 1

J--

-l

\\_pI

-- STRESS 800 f

t 80 MPa RAMP

^ 700 I

-2 j~

ys STUD STRESS DUE TO i

E 600

,7 - -

^

~

-~

-t s

60 MPa STATIC LOAD i

500 i

60 MPa RAMP g 400 4

,'t l

s i

i f

I

\\

300 l

\\ -

.,s.-

l 200 100 I

l 0

l O.000 0.004 0.00s 0.012 0.01s 0.020 TIME (sec)

Figure B-4:

Average Axial Stress in Studs I

4 B-9 i

i 0.008

,___s-----

0.007

/

l 80 MPa STEP 0.006 I

E 0.00s I

s E

I

{ 0.004 l

e l

• 0.003 l

-l 0.002 l

'l 80 MPa RAMP l

/

0.001

/

I ! '

MPa RAMP 60 0.000 8

O.000 0.004 0.008 0.012 0.016 0.020 TIME (sec)

Figure B-5:

Average Effective Plastic Strain in Stud's s'

Cn B-10

l 0.20 1

0.18 HEAD FAILURE STRAIN 0.16

/---

/\\

0.14

/

80 MPa STEP

? 0.'12 l

E

/

z 0.10

/

q

$ 0.08 j

80 MPs RAMP

=

/

0.06 f

l l

l 0.04 7

0.02 6

I '0 MPa RAMP 0.00 0.000 0.004 0.008 0.012 0.016 0.020 TIME (sec)

Figure B-6:

Average Effective Plastic Strain at Center of Head

[

n=

B-11

= -. -

i Summary:

The loading conditions we have examined are morc spatially. uni-form than those arising from impact by a solid water slug (B-1) which tended to concentrate the loading to the center of the head.

Instead, they approximate the LANL loading conditions predicted using SIMMER [B-8].

For this more spatially diffuse loading, the location of failure is uncertain.

This should be expected, since the loading is similar to a static pressure loading for which the vessel is designed.

Good design implies approximately equal strength for all failure modes.

Because of the change in loading, bolt failure is more likely in this study than in reference B-1.

Assuming flaws exist in the bolts, bolt fracture is predicted to'-

occur before head failure.

Thus, it is plausible that the bolts could fail and the head become a missile.

This is the assumption we have used in Sections 3 and 4 of this report.

Subsection 5.2 discusses the effects of the alternative possibility of head failure before bolt failure.

• W4 n-.

B-12

i REFERENCES B-1 M.

L.

Corradini and D.

V.

Svenson,

" Probability 'of

• Containment Failure Due to Steam Explosions Following a Postulated Core Meltdown in an LWR "

NUREG/CR-2214, SAND 80-2132, Sandia National Laboratories. (Albuquerque, NM, June, 1981).

B-2 Letter from R.

R.

Seeley (Babcock and Wilcox Research and Development Division) to D. V.

Swenson, (February, 1980).

B-3 W.

L.

Server and W.

Oldfield, " Nuclear Pressure Vessel Steel Data Base," EPRI NP-933 Electric Power Research Institute,

. Palo Alto, CA, Project 886-1, Topical ReDort, (December, 1978).

~_. B-4 A.

K.

Ghosh, "A Criterion for Ductile Fracture in Sheets Under Biaxial Loading," Meta 11urcical Transactions, Vol. 7A.

(April, 1976)

B-5 D.

Broek, Elementary Enoineerino Fracture Mechanics (Martinus Nijhoff. The Hague, 1982.)

B-6 WRC Bulletin

175, "PVRC Recommendations on Toughness Requirements for Ferritic Materials "

Pressure Vessel Research Committee, (August, 1972).

B-7 G.

C.

Sih, Handbook of Stress Intensity Factors. Institute of Fracture and Solid Mechanics, Lehigh University, (Bethlehem.,PA, 1973).

B-8 S.

W.

Key, Z.

E.

Beisinger and R.

D.

Krieg "HONDO II, A Finite Element Computer Program for the Large Deformation Dynamic Response of Axisymmetric Solids " SAND 78-0422, Sandia Laboratories, (Albuquerque, NM, October, 1978).

B-9 Commonwealth Edison Company, Zion Station Final Safety Analysis Report, Docket 50295-16 and 50304-16 (Chicago, IL, 1970)

B-10 M..G.

Stevenson (editor) " Report of the Zion / Indian Point Study Vol.

II,"

NUREG/CR-1411, LA-8306-MS, Los Alamos National Laboratory, (Los Alamos, NM, April, 1980).

i a-B-13

Distribution:

U. S. NRC Distribution Contractor (CDSI) 7300 Pearl Street Bethesda. MD 20014 275 copies for R1 U.

S. Nuclear Regulatory Commission (7)

Office of Nuclear Regulatory Research Washington, DC 20555 Attn:

R. M.

Bernero M. A. Cunningham R. T.

Curtis C.

N. Kelber J. T.

Larkins D.

F.

Ross L.

S.

Tong U.. S. Nuclear Regulatory Commission (5)

Office of Nuclear Regulatory Research Washington, DC 20555 Attn:

B.

S.

Burson M.

Silberberg J.

L. Telford T.

J. Walker R. W. Wright U.

S. Nuclear Regulatory Commission (6)

Office of Nuclear Reactor Regulation Washington, DC 20555 Attn:

G. Quittschreiber U.

S. Nuclear Regulatory Commission (5)

Office of Nuclear Reactor Regulation Washington, DC 20555 Attn:

W.

R.' Butler J.

Rosenthal Z.

Rosztoczy F. H. Rowsome T.

P.

Speis U.

S. Department of Energy Operational Safety Division Albuquerque Operations Office P.O. Box 5400 Albuquerque, NM 87185 Attn:

J.

R.

Roeder, Director n-DIST-1

i l

i Swedish State Power Board El-Och Vaermeteknik Sweden Attn:

Eric Ahlstroem

~

Berkeley Nuclear Laboratory (3)

~

Berkeley GL13 9PB Gloucestershire United Kingdom Attn:

J. E. Antill S.

J. Board N.

Buttery Gesellschaft fur Reakforsicherheit (GRS)

Postfach 101650 Glockengasse 2 5000 Koeln 1 Federal Republic of Germany Attn:

Dr. M. V. Banaschik Battelle Institut E. V.

Am Roemerhof 35 6000 Frankfurt am Main 90 Federal Republic of Germany Attn:

Dr. Werner Baukal UKAEA Safety & Reliability Directorate Wigshaw Lane, Culcheth Warrington WA3 4NE Cheshire United Kingdom Attn:

F.

R. Allen J. H. Gittus (6)

A. N. Hall M.

R. Hayns K. McFarlane R.

S.

Peckover British Nuclear Fuels, Ltd.

Building 396 Springfield Works Salwick, Preston Lancs United Kingdom Attn:

W. G.

Cunliffe AERE Harwell Didcot Oxfordshire OX11 ORA United Kingdom a

Attn:

1. Cook (2)

J.

R. Matthews, TPD DIST-2

Kernforschungszentrum Karlsruhe Postfach 3640 75 Karlsruhe Federal Republic of Germany Attn:

-Dr.

S.

Hagen (3)

Dr. J. P.

Hosemann Dr. M. Reimann Simon Engineering Laboratory University of Manchester M13 9PL United Kingdom Attn:

Prof. W.

B.

Hall (2)

S. Garnett Kraftwerk Union Hammerbacher strasse 12 & 14 Postfach 3220 D-8520 Erlangen 2

_ - - Federal Republic of Germany Attn:

Dr.

K. Hassman (2)

Dr. M.

Peehs Gesellschaft fur Reaktorsickerheit (GRS mbH) 8046 Garching Federal Republic of Germany

-Attn:

E. F. Hicken (2)

H. L. Jahn Technische Universitaet Muenchen D-8046 Garching Federal Republic of Germany Attn:

Dr. H. Karwat McGill University 315 Querbes Outremont, Quebec Canada H2V 3W1 Attn:

John H.

S.

Lee National Nuclear Corp. Ltd.

Cambridge Road Whetestone, Leicester LE83LH United Kingdom Attn:

F. P. O. Ashworth (2)

R. May Director of Research, Science.& Education CEC Rue De La Loi 200 f

1049 Brussels en Celgium Attn:

B. Tolley DIST-3

Northwestern University Chemical Engineering Department Evanston, IL 60201 Attn:

S. G.

Bankoff Brookhaven National Laboratory Upton, NY 11973 Attn:

R. A.

Bari (3)

G. A. Greene T. Pratt UCLA Nuclear Energy Laboratory 405 Hilgard Avenue Los Angeles, CA 90024 Attn:

I. Catton Argonne National Laboratory 9700 South Cass Avenue Argonne. IL 60439 Attn:

R.

P. Anderson (2)

Dae Cho Sandia National Laboratories Directorate 6400 P. O.

Box 5800 Albuquerque, NM 87185 Attn:

R. Cochrell (20)

University of Wisconsin Nuclear Engineering Department 1500 Johnson Drive Madison, WI 53706 Attn:

M.

L. C6rradini Los Alamos National Laboratory P. O.

Box 1663 Los Alamos, NM 87545

..s Attn:

W. R.

Bohl M. G.

Stevenson (2) i Battelle Columbus Laboratory 505 King Avenue Columbus, OH 43201 Attn:

P. Cybulskis (2)

R. Denning Offshore Power Systems 8000 Arlington Expressway Box 8000 f

Jacksonville, FL 32211 4%;

Attn:

D. H. Walker DIST-4

Electric Power Research Institute 3412 Hillview Avenue Palo Alto, CA 94303 Attn:

D.

Squarer (4)

G.

Thomas R. Vogel I. W. Wall Fauske & Associates 627 Executive Drive Willowbrook, IL 60521 Attn:

R. Henry General Electric Co.

175 Curtner Avenue Mai'l Code 766 San Jose, CA 95125

_ Attn:

R. Muraldiharan NUS Corporation 4 Research Place Rockville, MD 20850 Attn:

G. W.

Parry (2)

R.

Sherry Westinghouse Corporation P. O. Box 355 Pittsburgh, PA 15230 Attn:

N.

Liparulo (3)

J. Olhoeft V.

Srinivas TVA 400 Commerce Ave W56178C-K Knoxville, TN 37902 Attn:

Dave Simpson EG&G Idaho Willow Creek Building, W-3 P. O. Box 1625 Idaho Falls, ID 63415 Attn:

Richard Moore (2)

Server Sadik Applied Sciences Association. Inc.

P. O. Box 2687 Palos Verdes Pen., CA 90274 f

Attn:

D.

Swanson 4:n DIST-5 s

Purdue University School of Nuclear Engineering West Lafayette, IN 47907 Attn:

T. G. Theofanous UKAEA SRD Wigshaw Lane Culcheth, Warrington WA3 4NE England, UK Attn:

A. J. Wickett (20)

UKAEA Culham Laboratories, Abingdon, Oxon OX14 3DB, England, UK Attn:

F. Briscoe (2)

B. D. Turland School of Civil and Environmental Engineering Cornell University

~__

Hollister Hall Ithaca, NY 14853 Attn:

D. V.

Swenson (10)

Technology for Energy Corporation, One, Energy Center, Pellissippi park 1way, Knoxville, TN 37922 Attn:

A. M. Kolaczkowski, (2)

M. Fontana University of Aston in Birmingham, Department of Chemistry, Gosta Green, Birmingham B4 7ET, England, UK Attn:

A. T. Chamberlain (2)

F. M. Page University of Loughborough Department of Physics Loughborough, England, UK i

Attn:

J.

Sturgess H. M. Nuclear Installations Inspectorate.

Thames House North, Millbank London SW1 England, UK Attn:

R. D. Anthony (2)

[

S. A. Harbison c:-

DIST-6

DECPU/SEFCA C.E.N. Cadarache BP No.

1.

13115 Saint-Paul-LEZ-Durance France Attn:

Patrick Herter Dr. Albert B.

Reynolds D0pt. of Nuclear Engr. and Engr. Physics Thornton Hall University of Virginia Charlottesville. VA 22904 H. Jacobs KfK-INR.

P0stfach 3640 D-7500 Karlsruhe 1 FCderal Republic of Germany

- Lorry E. Hochreiter M0stinghouse Corporation Nuclear Energy Systems P.O. Box 350 Pittsburgh, PA 15230 Prof. E. Mayinger Tcchnische Universitaet Hannover 3000 Hannover 1 Fcderal Republic of Germany M. Georges Berthoud L3boratoire de Thermohydraulique des Metaux Liquides Sorvice des Trankferts et Conversion d'Energie C mmissariat a l'Energie Atomique Contre d' Etudes Nucleaires de Grenoble B. P. N' 85 X - Centre de Tri F-38041 Grenoble Codex France Dipl. Ing. Manfred Burger Institut fur Kernenergetik und Energiesysteme Universitate Stuttgart Pfaffenwaldring 31 P:stfack 801140 D-7000 Stuttgart 80 (Vaihingen)

FCderal Republic of Germany Dr. Ing. Leonardo Caldarola Institut fur Reaktorentwicklung Kornforschungszentrum Karlsruhe GmbH f

Pastfach 3640 c:.

D-7500 Karlsruhe 1 FCderal Republic of Germany D187-7

Mr. Katsuro Takahashi Research Associate FBR Safety Laboratory Carai Engineering Center P.N.C.

Oarai-Machi. Higashi Ibaraki-Gun. Ibarake-ken Japan Ing. Giovanni Scarano Il Direttore Laboratorio Studi Dinamica e Sicurezza del Nocciolo Divisione Ricerca e. Sviluppo Dipartimento Reattori Veloci Comitato Nazionale per l'Energia Nucleare Centro di Studi Nucleari della Casaccia S.

P. Anguillarese km 1+3400 C. P. No. 2400 I-001000 Roma, Italy

_~

Mr. Anthony J. Briggs /Mr. R. Potter (2)

UKAEA Atomic Energy Establishment Winfrith. Dorchester Dorset DT2 SDH United Kingdon Dr. Ing. Heinz Kottowski Head of Liquid Metal Section Heat Transfer Division Commission of the European Commmunities Euratom-Joit Research Centre Ispra Establishment C. P. No. 1 1-21020 Ispra (Varese). Italy David Yue P.O. Box Y Bldg. 9104-1 Oak Ridge TN 37830 Bill Parkinson SAI 5 Palo Alto Square. Suite 200 Palo Alto, CA 94304 Mr. Jacques Pelce Department de Surete Nucleaire (DSN)

Institut de Protection et de Surate f

Nucleaire (IPSN) ot:m Commissariat a l'Energie Atomique Centre d' Etudes Nucleaires de Fontenay-aux-Roses Boite Postale 6 F-92260 Fontenay-aux-Roses, France DIST-8

Dr. Hermann Rininsland H ad Projekt Nukleare Sicherheit (PNS)

K0rnforschungszentrum Karlsruhe G. M. B. H.

Pcstfach 3640 D-7500 Karlsruhe 1 Fcderal Republic of Germany Mr. Carlo Zaffiro Director, Safety Studies Division ROsearch & Development Sector Direzione Centrale della Sicurezza Nucleare e della Protezione Sanitaria (DISP)

ENEA Viale Regina Margherita, 125 ccsella Postale N.

2358 I-00100 Roma A.D.

Italy Dr. Kazuo Sato

_. Deputy Director Division of Nuclear Safety Evaluation Jcpan Atomic Energy Research Institute Tokai Establishment Takai-mura N2ka-gun Ibaraki-ken 319-11 Jcpan Mr. Lars G. Hobgerg Director, Office of Regulation and Research Swedish Nuclear Power Inspectorate Statens Karnkraftinspektion P.O. Box 27106 S-102 52 Stockholm Sweden Mr. Andrew C. Mu11unti

._ Director, LWR Safety Research and Development Office of Converter Reactor Deployment Mail Stop B-107 U. S. Department of Energy W2shington, D.

C.

20545 Mrs. Paola Fasoli-Stella Manager, Reactor Safety Programme C;mmission of the European Communities Euratom Joint Research Centre Ispra Establishment COsella Postale'l 1-21027 !spra

[

Italy c =,

DIST-9

Dr. Jacques Royen Nuclear Safety Division OECD Nuclear Energy Agency 38 Boulevard Suchet F-75016 Paris.

France Sandia National Laboratories Distribution:

1524 W. N.

Sullivan 6400 A. W.

Snyder 6410 J. W. Hickman 6411 V.

L.

Behr 6411 A.

S.

Benjamin 6411 A. L. Camp 6411 F.E.

Haskin 6412 S. W. Hatch

'6415 J. M. Griesmeyer 6420 J. B.

Rivard 6420 J. V. Walker

-~~~

6421 K. Maramatsu 6421 T. R. Schmidt 6422 D. A. Powers 6423 P.

S. Pickard 6425 W. J. Camp 6425 W. Frid 6425 R. J.

Lipinski 6425 K. Schoenefeld 6425 S. D. Unwin 6425 M. F.

Young 6427 M. Berman (20) 6427 M.

S.

Krein 6427 B. W. Marshall Jr.

6427 L.

S. Nelson 6427 O.

Seebold 6427 M. P. Sherman 6427 C. C. Wong 6440 D. A. Dahlgren 6442 W. A. von Riesemann 6444 L. D. Buxton 6444 S. L. Thompson 6449 K. D.

Bergeron 6449 D. C. Williams 7223 R. G.

Easterling 8424 M. A. Pound 3141 C. M. Ostrander (5) 3151 W. L. Garner I

n =.

DIST-10

\\

. g.

.....,s.......

~...

.. s....,

U',"3 BIBLIOGRAPHIC DATA SHEET NUREG/CR-3365 SAND 83-1438 Sit ist,.uce Oss os '=t.tvt.$t

,.,6 0.go gw..

6 8 Jtt.64 6.se AN UNCERTAINTY STUDY OF PWR STEAM EXPLOSIONS 3.ve g.c., d..,gegg t..

s.-

. D. ' E.tN=' is$wt 3 M. Berman, D. V. Swenson, and A. J. Wickett l 1984 May

,....o...su.a.si...ou....sa....m..a.

~.

,,c

...m.c,........s,s.....

Scndia National Laboratories

.,,sc,. a..

.s,....

Albuquerque, NM 87185 A1030

.om..m.:..s,...eu....s..

6,sa.:a...

,.c

.......o....a..

Division of Accident Evaluation Offic'e of Nuclear Regulatory Research Technical U.S. Nuclear Regulatory Commission

• • '"mo"*3

' ~. * * *

- -Wrchington, DC 20555

,,.......s,...u...

,,.. s,.. c, ix.

S:me previous assessments of the probability of containment failure c0used by in-vessel steam explosions in a PWR have recognized large uncertainties and assigned broad ranges to the probability, while cthers have concluded that the probability is small or zero.

In this rcport we study the uncertainty in the probability of containment failure by combining the uncertainties in the component physical processes using a Monte Carlo method.

We conclude that, despite sub-ctential research, the combined uncertainty is still large.

Some areas cro identified in which improvements in our understanding may lead to lcrge reductions in the overall uncertainty.

,. t oc..... s......

..<,-a si

....o..

...,.,g..,..

etcam explosions failure modes vcpor explosion Unlimited PWR uncertainty analysis

,',',**[,'"""'

..e.s......o.iseseece...s Unclassified a

c ynclassifie:1

, s..e.. o...<., s 89

.....a Q u e. nove na.st at paentemo ortset -

s ees.o.its eatiest e

. =-...-.- _ - -

ENCLOSURE 2 VU-GRAPHS FROM PRESENTATION TO ACRS BY PROF THEOFAN005 ON JANUARY 1,1, 1984 REGARDING THE POTENTIAL FOR STEAM EXPLOSIONS

\\

s v

i y

c s

s a

• ~

To bt* provided ;

v,.

f f

e

'(

ENCLOSURE 2 VU-GRAPHS FROM PRESENTATION TO ACRS BY PROF. THEOFAN0US ON JANUARY 11, 1984 REGARDING THE POTENTIAL FOR STEAM EXPLOSIONS 4

I 1,

l f

c, To be provided

tTE A M EXPLOdlOU E M E R GET/ CJ IN LWR G EO N E 7aiF IE t-THE MEY t W G RE b lE UTJ t E uplosivHy of Reactor thoffs.

Propaga Hon an d Eflicien cy

~

G uan tities in Premixing v

Inectia condrain t.

Ref:

PAIE.RI-14e

Nard Eng Dn g, 201 (1984)

TG.TJeoSehouj t

Pur.lue" 5

,.ew--,

e,-.,,..e-.,.r...,--,,.,.---,,,a,,n_.-.,.,,,,n,_.,,,-..-n,,

.M@.,,T..

w;..l

,.4 w K

~'~

,f

. ;. ym.

. e,

~

O nr.

m

.x..

gw..

e., m.. v.

4

...w.w m.

'c'.g c

j

.(*. v.. ;

.,%. gd.~.'i. ?

i

.i

).6 x,..:. 1..

g

. ?.' ".

'D,@&;j+i g

~

-4 ?%. : +:.-

• '..r.

.. : i..:.

,,,.s...@

z s 4, O

(b k

.;l.;;:

O

..,..f. >..;:n

. 2_. s... y...,.

a.

n :. -,-

w I

3..,,

m a_

'Q f

t. U g)

,2 e %T N <. '

• s.. /

I o

a

=

.7 o

~

i i

1. u~gt < pas s W

Vf C

k as I

I g

4 m

c 8

G o

g#

a e

u

+

k O

E q

e I

3 O

F-(_

u i

I l

E X Plott VITY OF RE ACTOR m A TE gieu hot (ow q

ll 0

hpr 8fo n kels eJ.\\.s>

ee g

• 4l" E e cial CondiNOMJ p

=~

/

e er y

p o

e y

E xy fortont l

\\

founda me ntal Guedient :

4 Tkr mal Ey prosion.1 ?

ll ll II cy Vapor or ct) tonfiyurah af premixfure

.u.

erquirel + rigger

?

R&e o f Ey pe rime nti

{ frale

~

l E.

l

1' E x nosi v/ T Y OF RERCTOR MRTE9/nLt l

E aue matniaine,I hant to fn'.79er Purdue hvf tued Py ~ 15' due lo uncerfttinfits and eack o f funda nenfat undertfandry.

Eorey experiment.9 iloved harJ fo fr/m

' fa ndia Recent experimenft demonstrole P,. ~.L.

cpon k neous +riyperi f.e.

i inony detalfed argumed for Idcor__

non-expforirify,uleof ex;oloslon.c.a eflouf o

=

!dat cannof r

I Sfafus Rcacfor Ma feria /t ae infler-exptbsire

.tstue closed.

Iny fica kons Exfend o f premixty limifed.

1 ee

j

.. g EVALUATION OF P(i.ff; mg x m )

i l

e i

I I

LENGTM7 (BPEE97 i

AYtsNWAtl0N?

P P850BACKt

?

l FRAGMENTATION?

HEAT / MASS TRANSFER?

t i

A >

a san e.

im

P R O P & 6 9 71 O A?

4 E FFICIE A)C Y Purdue

Opper fuei frayment tire in preututure eiq Upper bound conversior) effictency

[D=4 b1 My un cerla/r>.

Bott 7

yery too elftciency per experinenft a rg J'o n dia Recenf up dofe :

e /[ g.

[dcot-

fu el frag menf ir>

perintxfare ~l cm.

Very lo v efficiency per landia.

2ecenf updale : D o not cere.

[lalus funda menlah of fragnenfafion unertalo fo

fem, ettaWisA 'I coupeig fenytt scafev ' o & tlas "e ff/cien cier.

Imeficaifont..

Furfter studiet rc<fuired h tupjert fouer efffclencies and fie lo prenf xfure enflyurqNoy, an.I fo defernine effect af ofler verfoi fet t.e. void fra cffo6), inerffQ contfrctinf dc'.

k2ctofuhon :

In,,eyf feu neonfAt, fleordbcaffy

+ie-Joen recen & Jeuefor _nenk.

GURA)TITIEJIA) PRENIXTORE

=

w ~.

~

~_A v

(+f p@

W

. ^*" W f

~

=

Rpproacier: o)

I rantien/ brea kup 1 nixin o 1-p 6) 5-S 1-b lAcenal limils m)

S-S t.b +lermal fitnifs G)

(ra ncien f 5-b frea kup asJ Tltrnal L.imilf.

.fcafe Effech are Crocial ifere.

E n =.

l l

.. m

e GUR uTITIES I AI PRE MIXI A16 Purdue :

fuel breakp n a role proa cs. Ested on hydrodynanta and geometric contfyninft used a 3 yl o f com for to-ressef a 10 g a f core for ex-resvel.

Ouafiloffvcly ackneu eedepe./ teff-Ptatffey clarackr k k due fo heaf fr<ntfer.

"3%, ~ )O% -

fondia:

Scafe-up from experimenft uring tiryfe drop approack*/

1.e. 8 drody non (ct 9

3-5%,-/A%

Zdcor :

Zdcalized,4imtfiny ca re to cpusofih effect of tfet mIof. Exfentt'ons 44 Ikir;l Rankoff and ly Corredini sup ort

-He genetul concept, but CHtt 4 i AG t de n liked.( Z-S,.L-b, eicy 10-100 fiah t.- ee sorufron

? Reatih confan/nj Lon cowponenft. Re f/neneeft nealed. nnaQ.cn nesf feu nonfU fadoued by

\\

o re.

L ndi cafez ex pert m e,,fi.

INP f/ cak' Obit

[l Dou,U eppar- (e&c1VC So <d* MontfrQlC

$10f a f infetcVL.y ff Mi/t'd si Jeverel extenal. a f p teniarfart1

~

for LV9 condiffon t

t 1

Co ucL US 10 US 4 3,0 DO $]

OC'Y. O h 40 M *b SO){ CY/ElCnus

~ 3 m'.F Rul u

"erceae.ym

=

unety[

l Cumd ruuls:

e L % e.

made

// [o ntideensWAt-erN SU'd M lLAf s'l UW l

Le perdWe k

dou haf ife c o n t e m,'

negri e've o i-9 9

e

)

Y>

l ENCLOSURE 3 j.

SUMMARY

OF CONSENSUS REACHED BY PARTICIPANTS IN THE SUBTASK ON STEAM EXPLOSIONS AT THE NRC/IDCOR MEETING ON SEVERE ACCIDENTS AT HARPER'S FERRY WEST VIRGINIA DURING NOVEMBER 29 THROUGH DECEMBER 1,19 1

.~.,

T,

,l, 4

1...

r t

e c=

i y

/

4'd m-e r

ia e, -,.

-,-c---,,,e,-,r.

y-*en e,x r+-

w-ew..,-

,w.-y:--.

4

- +

-, -- - - +-. -

m,-

6 STEAM EXPLOSION PHENOMENA THIS PRESENTATION REPRESENTS A CONSENSUS OF THE FOLLOWING CONTRIBUTORS:

T. G. THE0 FAN 00S, PURDUE UNIVERSITY T. GINSBERG, BNL R. W. WRIGHT, NRC J. TELFORD, NRC fl. L. CORRADINI, UNIV. OF WISCONSIN

f. CATTON, UCLA K. BERGERON, SANDIA M. BERMAN, SANDIA ET AL.

l l

l I

n=

i STEAM EXPLOSION PHENOMENA STEAM EXPLOSIONS (AND ALL RELATED FCI PHENOMENA) h D5 PEND STRONGLY ON SCALE.

THE CONCEPT OF A LIMIT TO MIXING BASED ON BOILING / HYDRO-DYNAMIC EFFECTS IS CONSIDERED TO BE OF MAJOR IMPORTANCE IN THE SEVERE ACCIDENT ARENA.

THE RAMIFICATIONS OF THIS CONCEPT G0 FAR BEYOND DIRECT

~

FAILURE DUE.T0 STEAM EPL0SIONS.

STEAM AND HYDROGEN GENERATION RATES (BOTH IN-AND EX-VESSEL) AND FUE'l DEBRIS CHARACTERISTICS ARE STRONGLY INFLUENCED BY THIS CONCEPT.

~

• eu O

e up

DIRsCTCONTAINMENTFAILURE(A-MODE)

KEY ELEMENTS:

EXPLOSIBILITY OF PROTOTYPICAL MATERIALS.

QUANTITIES ~0F PARTICIPATING MATERIALS, PROPAGATION AND CONVERSION RATIO.

FAILURE. PROBABILITY.

e oe.

e 4=

j EXPLOSIBILITY OF PROTOTYPICAL MATERIALS STAFF:

SPONTANE0US EXPLOSIONS CAN OCCUR UNDER APPROPRIATE INITIAL AND BOUNDARY CONDITIONS IDCOR:

ALTHOUGH MANY DETAILED ARGUMENTS FOR NON-EXPLOSIBILITY ARE SUGGESTED, IT IS ACKNOWLEDGED

~

THAT EXPLOSIONS CAN OCCUR.

SUMMARY

NO MAJOR BOTTOM-LINE D'ISAGREEMENT.

o 9

• e.

(

e9

-an--m---

+,t.s-,m-4.

-i p--

m.

QUANTITIES OF PARTICIPATING MATERIALS IDCOR:

SEVERELY LIMITED ( 10.100 KG) BY ARGUMENTS BASED ON

. HYDRODYNAMIC STABILITY;AND ENERGY REQUIRED FOR FRAGMENTATION, STAFF:

MAY BE LIMITED BY ARGUMENTS BASED ON DROPLET FLUIDIZATION, GE0 METRIC EFFECTS AND HYDRODYNAMIC BREAKUP PROCESSES TO 1000-4000 KG IN-VESSEL, AND 10,000-20,000 KG EX-VESSEL.

MAJOR' UNCERTAINTIES RELATED TO THE POSSIBLE MULTI-DIMENSIONAL AND TRANSIENT NATURE OF THE EXPLOSIONS,

SUMMARY

MAJOR DISAGREEMENT, ADDITIONAL RESEARCH REQUIRED i

L

PROPAGATION AND CONVERSION RATIO IDCOR:

LARGE FRA&ENTS PECLUDE AN " EFFICIENT", " ENERGETIC" EXPLOSION PROPAGATION AND CONVERSION RATIO AE BASICALLY IRRELEVANT BECAUSE "NECESSARY PEMIXTUE" IS PECLUDED BY OiF AND MIXING ENERGY ARUGENTS.

STAFF:

PREMIXTUES CAN INCLUDE PARTICLES LARGER THAN IDCOR ASSTES. EXPLOSIONS CAN OCCUR IN STRATIFIED GE0ETRIES, BUT EERGETICS UNidOWN. CURRENT l@0WLEDGE PERMITS CON-VERSION RATIOS FROM 0 15%. MAJOR UNCERTAltiflES EXIST 00NCERNING THE EFFECTS OF LARGE SCALE, CONFINEENT, PROTOTYPICAL STRUCIUES, AND HIGH AMBIENT PESSUE.

h STNARY: MAJOR DISAGEEENTS. ADDITIONAL RESEARCH EQUIED f

. =.

~

~

DIECT FAILUE PROBABILITY IDO)R:

ESSENTIALLY Ilf0SSIBLE.

STAFF:

IT IS GEERALLY bel.IEVED THAT EXPLOSIONS IN EXCESS

~

OF 2000 MJ h0VLD BE REUQIRED TO FAIL CONTAlffENT, EXPLOSIONS OF SU0i HIGH ENERGY ARE DEEE D UNLIlEl.Y, BUT HAVE NOT BEEN DEM)NSTRATED TO BE IMPOSSIBLE.

Slfl%RY: ADDITIONAL ESEARCH MAY BE NEEDED.

• * =.

I c.

IsRC FORu 256 C e nACT m m 1 0 18)

NRC-33-85-311 DivtSION OF CONTRACTS MODIFICATION NUMBER U.S. NUCLE AR REGULATORY COMMisslON WASHINGTON. D.C. 20555 UNEW D MODIFICATION OTHE R 15pecofri NOTIFICATION OF CONTRACT EXECUTION CONTRACT BASED ON:

TO:

Edwin G. Triner, Director AUTHORIZATION NUMBER Division of Budget and A lysis nnerly OM-8M09)

Office of Resource Management R M ORM-85-311 DATE toroenuntson) 3/5/84 A

i CONTRACT CHANGES PER THis ACTION FROM:

h y

(

f tDate)

UoyCe Bazin, Contract Negotiator Administrative Contracts Branch i

DIVISION OF CONTRACTS, ADM

]

CONTR ACTOR (Name a Location)

EXECUTION DATE United Engineers and Constructors, Inc.

12/.2V /84 30 South 17th Street P. O. Box 8223 ME OF CONTR ACT Philadelphia, Pennsylvania 19101 CPFF/ Task Order PROJECT TITLE PERIOD OF PERFORMANCE Engineering Cost Analyses of Proposed Changes in Two years

/

f- / 4 A7

)

NRC Requirements Having Economic Consequences PRINCIPAL iNvEsTiG'ATOR i

John F. Crowley/E. Ziegler NRC AUTHORIZED REPRESENTATIVE i

Gor' don Fowler B&R NUMBER FIN No.

Appropriation No.

80-19-06-03 D1216 31X0200.805 AMOUNT NEW NRC FUNDS s 258,845.00 FUNDING TOTAL FY _5 FUNDING s 258,845.00 j

8 TOTAL NRC OBLIGATIONS s 258,845.00 GOVERNMENT PROPERTY ATTACHMENTISI: NRC-33-85-311

)

i 1 Copy

{0NTR ACT DOCUME NT I y

senC FORM 2es 1 0 784 8501090077 841224 PDR CONTR NRC-33-85-311 PDR

p r

c..

A

_4 n

AWARD / CONTRACT i CmiFiED ron utiON AL ot rENSE P ' "'

" '[

j!

8 uNora soSA atc : aNDrOa DuS af c i i

7toNi a. Ao e.

sa., se,a,, e,o a ten (ci.vaoAit e a c c,v s.,,o

, evac-a u a t outsi,eaow %

s NRC-33-85-311 1/7/85 ORM-85-311 (fomerly ORM-84-409) s essuto e' COoE I

' ' " ' ' ' ' " " ' ' ' ' " * ' ' ' ' ' ' ' ' ' " ' ' ' cOof I U.S. Nuclear Regulatory Commission Division of Contracts, AR 2223 Washington, D.C.

20555

7. NAMt ANo A oDa tss of cont R ACT om (No. esmes. cary. coanars. ssese end DP Cedes aotuwtav United Engineers and Constructors, Inc.

30 South 17th Street fos ORicis OTHEa<se.

~

Post Office Box 8223

.. oiscovNi r oa a o-*T Av =t~1 Philadelphia, Pennsylvania 19101 NET

{TEw 10 SUBuiT INVOICES

<e so,.ee.aie-e, e' 12 wene e,ecihedl TO THE CODE F ACitlTY CODE ADDRESSSHOWNIN 3 3. smo Totuaan r oa j aa. Pav utNT weLL et uAot av U.S. Nuclear Regulatory CESMssion COoE Attention: Gordon Fowler, MS MNBB 12217 Refer to G.8.

Washington, D.C.

20555 i s AovE=T SE o

• 'd^*

^**" * * ' ' " * ^'^

is Tais i

B"&R"cc "0-19-06-03 APPN: 31XO200.805 ACousmON'

8 s e.tc.W e AitJ punsvaN1 to

• ^5 C'"'

u s'se see>>so usc 23aamt i x 41 u$c 252 en 10, FIN: 01216 OBl.lMTED: $258,845.00 15A iTE M NO '

15e SUPPolES' SERVICES 15C OuANTITY 15D UNIT l 15! UNIT PRICE

'15F AM 0 d'_

The U.S. huclear Regulatory Comission hereby acceptsUnited,EnginehrsandCons tructors Inc.'s e

proposal dated July 23, 1984 as revised Septenber 21,1984 which is incorporated into this contract by reference, to perform the effort specified $erein on a task order basis.

See attached pages for incorporation of admini strative changis.

This is a fully funded cost-plus-fixed-fee tas k order type c ntract.

I i

i i

isc ToTAt AMOUNT or CoNTR ACT > S258,845.00 16 TABLE OF CONTENTS VI lSEC l CE SC p tPTION lPAGE fS) l Vi j$EC l DE SC RirTION IF A ! '.S oAn :- tat scatoest Pant es - CONTE Ac1 CLausts X l A l SOLeCETATION CONTR ACT FORM l }

Xl l l CONTR AC7 clauses

! 29 l

l X

F

,S.M. i ' :: Si e..CEi Au 8 3.:ES 07,515 2

past m - List os ooc vet Nis. s a-se.t s aN:, cT at a a ti n.

X {C j DESCRiPTsON/SPE CS / WORK STATEMENT 4

'XlJ l LIST OF ATTACHMENTS

80 X jC i P Acc AS,N; aNC. e:mic iNC

/

past av - stentst N anoNs &NC inst auctions l

A E

INSDECTION AND ACCEPTANCE

/

K R E PRESE NT ATIONS. CE RT sF IC ATIONS AND I

X F

DEtivERIES OR PERFORMANCE 6

OTHE R ST ATEME NTs OF OF F E RORS l

1 X

G CONTR ACT ADMINiSTR ATION DAT A 10 t

INSTR $.CONoS. AND NOTICES TO OF F E R crge::,cm.,-c ;-ap.pices.gs,e 2}

v gy,. :.-.. ;-

- t r e av.:: ;

x s.

CONTRACTING OHICER WilL COMPLETEITEM UOR M AS AMICABLE It b coNTeacTom s NicoTsaTro acattutNT scoar.ecree u ee

@awano reoar cea.

ao

,,..,d te e.ga en dec. mea.

le I

a..wa so,a.s= the dor mear end nevra res.es to seaweas off.ce s fee oa See.otspor Nwmoe.

o'ci.esmo the seeet.oas or sneasse mooe er you wn.ca aseit.ons o. ca-l Coateecio e s

coas.eeret.tas., tooweaisa one e...eer see stems on modeem en tae serences setassene emot.e one on say conteaust.on enests toe tae so on ea, contenwetion sneets Tn..eo.r ece sotee en to the noms hoiee. sea s teeta or e e..

are est sectn sa een soewe. m ne.

i e ees coasummates tae contraci -

ee l

Tne e sats one osi.es!..a the seet.es to ines one po caes ey tae foHow.s o'e oa sietee ace.a coateact saan ce sweisc? tocstet.oa et say. mas ten seca pro sc.umeats:

o'ver, eas to) taes e=ses/coateact, av swetae, eeament a ses.c.te t.c.* ev. *.

l at (a) lais asets of the folsee.at cocumeats ge) tae Go e=searcontract. son tae soe

.s oas. essee contractus socumeat es at.e-to.as., ace nereiacertiracsitoas. saa soeuricelens, en s.ee atteca e se encorporei nsate e av sery.

l me.

(A ttac4= ease em aaered Anm = s asA N A w t A N c.1. ' E a s s e c N t m (73 pe ok pe.a a, zoa. Nawt os coNin Act.NG or a sct s.

v E }g ng Officer

~

19e Naut os coNTeactoa sec.DATt ssGNED pt. velf t ST A T E.s n ec A-roc.paTE sign c' MI

/A 2.----

ft g '

l eY ev ts.snerwn o.e pereea ea rnecues se essa,

.//ettes% tv e of co i,y orp,,, f r

NsN Tseo 03152-so49 34 tos ST A DARDFORM tRt la r:'

• ntveous to TsoN UNusaser

• .evor=ee.Gs*

1. ## c # "I '3 #3 'dI e CPO s 1944 0 - 421-526 iM.

8501090083 841224 PDR CONTrt NRC-33-85-311 PDR DCS

NRC-33-85-311 Page 1A 1

~

The following administrative changes are hereby made:

Paragraph B.2., Cost and Fee Information is hereby deleted.

Paragraph C.4., Level of Effort is hereby deleted.

Paragraph F.2., Place of Delivery is hereby completed as follows:

U.S. Nuclear Regulatory Comission Attention: Gordon Fowler Office of Resource Management Division of Budget and Analysis Mail Stop: mBB 12217 Washington, D.C.

20555 Paragraph G.2.1, Consideration is hereby changed to read as follows:

A.

The total ceiling amount of this cost-plus-fixed-fee task order contract is $258,845.00.

B.

The total funds currently available for issuance of individual task orders under this contract are $258,845.00.

C. The obligation amount may be unilaterally increased from time to time by the Contracting Officer by written notice to the Contractor. Any such increase shall be for performance of Task Orders requirements initiated during the contract period. The Contractor shall, at no time, exceed the obligation amount as specified herein. When and if the amount (s) paid and payable to the Contractor hereunder shall equal the obligation amount, the Contractor shall not be obligated to continue performance of the work unless and until the Contracting Officer shall increase the amount obligated with respect to this contract. Any work undertaken by the Contractor in excess of the obligation amount specified above is done so at the Contractor's sole risk.

Paragraph G.3., Overhead / General and Administrative Rates is completed as follows:

A.

Pending the establishmer.t of final overhead rates which shall be negotiated based or, audit of actual costs, the contractor i

shall be reimbursed for allowable indirect costs hereunder at the provisional rate of S percent of total direct labor. [PqerQ(TAV) l S.

Pending the establishment of final general and adr,inistrative rates which shall be negotiated based on audit of actual costs, the contractor shall be reimbursed for allowable indirect costs

~

hereunder at the provisonal rate of S percent of total

[pgop lff4G) q direct cost.

C.

Notwithstanding A. and 6. of this Section, said provisional overhead and G&A rates may be adjusted as appropriate during the tem of the contract upon the acceptance of such revised rates by the Contracting Officer.

l l

l

m l

- a.

NRC-33 85-311 Page 18 Paragraph G.4., Payment of Fixed Fee is hereby changed to read as follows:

At the time of each payment to the Contractor on account of allowable cost for each Task Order issued hereunder, the Contractor shall be paid an amount which is in the same ratio to the total fixed fee under that Task Order as the related payment being made on account of allowable cost is to the total estimated cost of performance of the Task Order; provided, however, that after payment of eighty-five percent (851) of the total fixed fee, the provisions of paragraph (b) of the Clause 52.216-8 entitled Fixed Fee (April 1984) shall be followed for each individual Task Order.

The fixed fee applicable to each task order issued pursuant to this contract shall amount to 7 percent of total estima ted cost negotiated on each task order, Paragraph G.6.B., Project Officer is completed as follows:

Name and Mail Code: Gordon Fowler, Mail Stop MNBB 12217 Office Address:

Office of Resource Management Washington, D. C.

20555 Telephone:

(301) 492-9861 Paragraph H.l., Key Personnel is hereby completed as follows:

J. H. Crowley J. B. Mulligan E. J. Ziegler A. Shinnar Paragraph M.2., 52.252-3 Alterations in Solicitation is hereby deleted I

l l

l

~

--.=. - -

'~

' ~

O' -

J V2.c o, ty

?.

1. CERTIFIED FOR NATIONAL DEFENSE

. soUCITATION, OFFER AND AWARD UNDER BOSA REG. 2 ANO/OR DMS REG.1 >

1 l103,,,,,

L4;gygTnM;T NO.

J. 50LACITATION NO.

4. TYPE OF 50L4CATATsON 5. DAE 455uEO

6. R Equi 3l TION /Pu RCHASE t

1 AOvERTISED(IFBI "O

,a

/,

y RFP-RS-0RM-84-409 l NEGOTI ATED (RFPI 6/21 /84 ORM-84-409

i. i35uto av CODEI e d M e[s'*b'oM 'O N dlfr Nsed as indicated in U.S. Nuclear Regulatory Commi'ssion Block 7, however, handcarried offers. (this Division of Contracts includes Express Mail and all comercial delivere Washington, DC 20555 g c g must be delivered to the address in NOTE: In advertised soaicitations " offer" and " offeror" mean "hid" and "bsdder" SOUCITATION f

T four ( 4 Ic $ses for furnishing the supohes or services in the Schedule wi fM'/N 8.or b

9. Siseine offers m original and p

I 222, 4550 Montgomery Avenue L

handes,ried. in me deposiiory sisted in on

, ti y y.g Bethesda, MD 20814 N'8 7/JU)1TJug-CAUTION - LATE Sutwmsesons. Modifications and withdraweis: See Section 1. Provision No. 52.214 7 or 52.21510. All offers are sutgect to all terms and condstions conramed in this sohcitation.

i io. FoR INFoRMATioN k ^ * "

CALu F

Joyce P. Bazin 301/492-7182 or 492-4800

11. TASLE OF CONTENTS Ul lSEC. l OESCRIPTION lPAGE(S)

WIl SEC. l OESCRIPTION lPAGE(S)

PART I - THE SCHEOULE PA RT 18 - CONTR ACT CLAUSES y

A SOLICITATION /CONTR ACT FORM l

Xl 1 l CONTRACT CLAUSES I 29 y

8 SUPPLIES OR SERVICES AND PP'CES/ COSTS 2

PART esi-LIST oP OoCuMENTS, EXHISITS AND OTHER ATTACM.

y C

OESCRIPTION/SPECSJWORK STATEMENT 2

Xl J l LIST OF ATTACHMENTS I 80 y

0 PACKAGING AND MARKING 7

PART IV - REPRESENTATIONS ANO INSTRUCTIONS y

E INSPECTION AND ACCEPTANCE 7

RE8RESENTATIONS. CERTIFICATIONS AND g

y F

DELIVERIES OR PERFORMANCE g

X OTHER STATEMENTS OF OFFERORS 81 y

G CONTRACT ADMINISTRATION

• DATA 10 X

L INSTR $.. CONOS.. AND NOTICES TO OFFER 87 y

M SPECIAL CONTR ACT REQUIREMENTS 2]

X M

EVALUATION FACTORS FOR AWARo 100 OFPER Westbe /t4 comPdered W onkrer/

NOTE: Item 12 does not apply if the soiicitation inesudes the crowiesons as 52.21416. M6aenum Sid Assestance Period.

i

12. In comohance with the above, the underssened agrees. if this offor is acceosed withm calender devs (so calender deve unases e derresent pened is inserted by the e/ferers frorn the date for receipt of offers specified aeove. to furnish any or all isome upon which prices are offered at the price set

(

ocoosite each item delivered at the dessenesed Domtts). withm the time specified in tt's schedule.

to CALENDAR OAY5 2o CALENOAA DAY 5 Jo CALENDAR OAY5 CALENDAR DAY 5

13. DISCOUNT FOR PROMPT PAYMENT k

tsn seetten t. csemee se. s2 222.no V

O 0

0 0

14 ACKNOWLEDGMENT OF AMENOMENTS AMENOMENT NO.

DATE AMENOMENT NO.

DATE iThe offerer mennoistedses reeempt of amend.

1 77gjg4 ments to the SOLICITATION for offerere and redefed documente nam 6ered and dated:

CODE l l FACILITV l

16. g g A LE OF PER50N Al ATHolttIED TO 56GN B

i 1S A. NAME ONO United Engineers & Constructors Inc.

p f, r__

dooa*55 30 South 17th Street T. H. Zarges la P.O. box 8223 Vice President Phila., PA 19101 ISS. TELEPMON [ NO (fneded* *'**

ISC. CHECK IF REMITTANCE AOORESS 17.siG uRE -

as.OrrER OATE (2 Y 422-4400 SuSNERYSS7N SC 8E 7/26/84 AnARD ITo be comoerted W Genernment)

/)

I

19. ACCEPTED AS 70 eTEMS NUMSEREO

20. AMOUNT

21. ACCOUNTING AND APPROPW1 ATION l

\\

l I

&3. SUBMIT INVOICES TO ADDRESS SHOWN IN l

(O cepees undees otherwsse spect/ fed; 10 U.S.C. 2304(a) (

)

41 U.S.C. 252(c) (

)

24. AOMsNsSTERED SY tIf other then treni 7)

CODE l

25. PAYMENT WILL BE MADE GV CODE l l

t

26. NAME OF CONTR ACTsNG OFFsCER (Type or pnnt)

27. UNITED STATES OF AMERICA 2s. AWARD OATE (sdeneture of concrettfas Officerf I

l IMPORTANT - Award well be enade on this Form, or on Standard Form 26. or tw other authorised official untten notice.

8sSN 7940 41 192-8504 33 832 opo: sees 0-eas-eer STANOARO POAM 33 ("EV' I8*833

• REVsOUS EOeflON NOT USAGLE P'**E**8 l'Y OSA F AR (48 CF R) 53.214(C)

._,1--- -r---,,-pv.m7-,

w -, e y ry vr -

,ww--,-y--,.-m p,,,y-.,-wv.-c--~

.->-w-

.,ewm.--..

--c.

.m..+.-,,,.,---y

---~,.r--_

NRC-33-85-311 Page 2

~

Section B - Supplies or Services and Prices / Costs 3.1 Brief Description of Work The objective of this cost analysis is to identify and analyze the economic costs of proposed changes in NRC regulatory requirements. The resulting costs analyses will be an input to decisions on whether or not changes should be required.

The work under the contract will be issued as task orders, about one task each month. It is entirely possible that two or three tasks could be on-going simultaneously. Each task should take approximately 5 chronological weeks and consume approximately 11 man-weeks of effort, although this estimate could very substantially.

B.2 (Offer should provide Cost and Fee informa' tion)

Total Estimated Cost $

Fixed Fee $

Total Estimated Cost Plus Fixed Fee 5 Section C - Description / Specifications / Work 8

C.1 Scope of Work l

C.l.1

Background

The U.S. Nuclear Regulatory Commission (NRC) is responsible for the 3

protection of the public health and safety in the civilian use of i

nuclear power and nuclear materials. In the pursuit of this mission, the NRC imposes regulatory requirements from time to time on the nuclear industry to improve the safety of licensed activities.

Because new regulatory requirements may be expensive for licensees to implement, the NRC analyzes the costs of possible new regulatory requirements. The results of the analyses have an important bearing on the c,hances that a regulatory requirement will be issued.

The NRC staff office which proposes a new regulatory requirement is responsible for its cost and benefit analysis. However, the independent Office of Resource Management also either performs or evaluates the cost elements for selected proposed regulatory requirements either in-house l

or by contract.

t i

C.1.2.

Objective l

The objective of this cost analysis contract is to identify and analyze the economic costs of proposed changes in regulatory require-ments in accordance with specific task orders and the provisions of this contract.

' ~.... -

.,,,,_,_.m.

4

~ '.

gRC-3g-85-311 C.1.3.

Statement of Work The contractor shall supply the personnel, materials, transportation, and facilities necessary to provide analyses ordered by the NRC under individual task assignments to estimate costs resulting from proposed changes in NRC requirements.

The tasks will be to identify and estimate all costs associated with the proposed regulatory action. Primary emphasis will be on the costs to the licensee ~ of replacement power and the economic impact on society of proposed regulatory changes affecting any existing or potential licensed activity. Overall costs to the

' licensee for engineering and construction shall also be included.

Task orders will describe the proposed regulatory change in sufficient detail to enable the contractor to make the costs analysis. Any special instructions which may be required because of the nature of the regulatory change, or limitations which should apply to the analysis will be included. Task orders will also specify the date or time for submission of the analysis or parts thereof.

C.I.4.

Approach and Method The tasks may be assigned in any of the following fonns:

- pre-existing cost analyses to be validated;

- Actual alternatives for satisfying a regulatory change with technical specifications provided; or

- perfonnance objectives, without technical specifications.

The method to be used by the contractor shall include anclysis of tac extent and durcti:n of the inpact of the regulatory change, wh:r applicable, on such elements as:

- Individual plants or classes of plants or other regulated l'

activity including:

l o amount of calendar time required for the repair; and o downtime of the plant if modifications cannot be made during scheduled outages; i

- In the case of power reactors, the power systems of the i

utilities affected and the inter-utility grid to which the utility belongs; I

- Surrounding populations (percent affected, type and extent of effect);

t

NRC-33-85-311 Page 4

- Other fims (vendors, licensee suppliers, contractors, including any significant impact on small businesses. The latter is required to comply with the Regulatory Flexibility Act - See Appendix B to ).

- Government agencies (impact on resources and services or NRC, other Federal State and local Government);

- The extent and probability of radiation exposure of workers.

C.1.4.

Elements of Cost for the Analyses After the type, extent, and duration of the impacts have been detemined by the contractor, cost estimates shall be made by the contractor, when applicable, of:

1) Replacement power, in the case of nuclear power plants, based for example on:

available replacement power under the control of the affected utility or utilities;'

available replacement power within the inter-utility grid to which the utility or utilities belong; replacement power by region of the country *

2) Impact on the socio-economic condition of the consnunity affected by a shutdown, based for example on:

employment;-

property values; other business activity

3) The initial (capital) and long-tem (operations) costs (direct and indirect) of materials and labor to the licensee, based, for example on the need for:

engineering design ~ and testing of components and systems to be installed equipment)(i.e., structures, pipes, and valves, electrical procurement, installation, operation and maintenance of components and systems; integration with other propose'd requirements; operating procedures, training. curricula and documentation necessary to operate newly-installed components and systems; 7

_.,,,,,,