Insights from the Peach Bottom SOARCA

Uncertainty Analysis

S. Tina Ghosh and Edward L. Fuller, USNRC

Kyle W. Ross and Randall Gauntt, Sandia National Laboratories

IAEA Workshop on Advances in Understanding the

Progression of Severe Accidents in Boiling Water Reactors

Vienna, Austria

17-21 July 2017

Outline

• Goals and approach

• Description of PB SOARCA LTSBO scenario

• Selected parameters and bases

• Results

• Phenomenological insights from MELCOR

analyses

• Conclusions

2

3

Goals of the

Uncertainty Analysis

• Develop insight into overall sensitivity of

SOARCA results to uncertainty in inputs

• Identify most influential input parameters for

releases and consequences

• Demonstrate uncertainty analysis methodology

Approach

• Investigated parameter uncertainty

• Focus on epistemic (state-of-knowledge)

uncertainty in input parameters for MELCOR

and MACCS simulations

• Peach Bottom, unmitigated, long-term station

blackout scenario chosen

• Scenario definition not changed after Fukushima

4

5

Approach (continued)

• Key uncertain input parameters were identified

• Uncertainty in these parameters propagated in

two steps using Monte Carlo and Latin

Hypercube (LHS) sampling: – A set of source terms generated using MELCOR model

– A distribution of consequence results generated using

MACCS model

• Three epistemic sample sets of 300 generated

to complete a corresponding number of

individual code runs (Monte Carlo

“realizations”) to evaluate the influence of the

uncertainty on the estimated outcome

Approach (continued)

6

• Results reported included:

– Analysis of source term releases, in particular, cesium and

iodine release over time

– Distribution of latent cancer fatality risk, with three dose

threshold models

– Description of most influential uncertain parameters in study

• Tools used to analyze results include statistical

regression-based methods as well as scatter plots and

phenomenological investigation of individual realizations

of interest

• Guidance solicited from SOARCA peer reviewers on the

uncertainty analysis plan documenting the approach,

chosen parameters and distributions

Process for Choosing

Parameters and Distributions

• Core team of staff from SNL and NRC with

expertise in probability and statistics, uncertainty

analysis, and MELCOR and MACCS modeling

for SOARCA

• Subject matter experts (SMEs) provided support

in reviews of data and parameters

• Approach is based on a formalized PIRT

(phenomena identification, and ranking table)

process.

7

Process (continued)

• Focus on:

– confirming that the parameter representations

appropriately reflect key sources of

uncertainty, and

– ensuring model parameter representations

(i.e., probability distributions) are reasonable

and have a defensible technical basis.

8

Process (continued)

• Attempt to obtain contribution from uncertainty

across the spectrum of phenomena operative in

the analyses, through a balanced depth and

breadth of coverage

• A subset of possible uncertain parameters is

proposed that cover the range of phenomena

across the stages of a severe accident.

9

10

Probabilistic Uncertainty analysis is an iterative process

Scenario Selection current MELCOR model/software

current WinMACCS model/software

Inputs & Parameter

Selection informal expert elicitation from SMEs

& Peer Review

Uncertainty Analysis

Framework MELCOR Uncertainty Engine

MELCOR

MELMACCS

WinMACCS/LHS

Uncertainty Analysis partial rank regression correlation

step-wise rank regression correlation

scatter plots

Single Realization phenomenological analysis

Stability Analysis # realization stability

random seed stability

alt. modeling cases

Iterate Range of validity

Stability of analysis

Alternative model scenarios

SRV Stochastic

SRV Thermal

MSL Creep Rupture

Sensitivity Analysis

Address Bugs or Modeling Errors

Analysis Complete

Event Times for PB SOARCA

Unmitigated LTSBO (NUREG-CR-7110)

Event Time (hr)

Loss of all onsite and offsite AC power (SBO) 0.0

Low-level 2 and RCIC actuation signal 10 minutes

Operators manually open SRV to depressurize RPV 1.0

RPV pressure drops below LPI setpoint (400 psig) 1.2

Battery depletion and immediate SRV re-closure 4.0

Steam line floods and RCIC flow terminates 5.2

SRV sticks open because of excessive cycling 8.2

Downcomer water level reaches top of active fuel 8.4

First hydrogen production 8.9

First fuel-cladding gap release 9.1

First channel box failure 9.3

Water level reaches bottom of lower core plate 9.3

First localized failure of lower core plate 9.6

First core cell collapse due to time at temperature 9.8

Large-scale relocation of core debris to lower plenum 10.5

Lower head dries out 13.3

Lower head failure 19.7

Drywell head flange leakage begins 19.9

Hydrogen burns initiated in reactor building 20.0

Refueling bay to environment blowout panels open 20.0

Hydrogen burns initiated in refueling bay 20.0

Drywell shell melt-through initiated 20.0

Hydrogen burns initiated in lower reactor building 20.1

Railroad access opens due to overpressure 20.1

Refueling bay roof fails due to overpressure 20.2

Calculation terminated 48.0

11

Table 5-1 Timing of Key Events for LTSBO

RPV Pressure

(PB SOARCA LTSBO)

12

0

200

400

600

800

1000

1200

1400

0 2 4 6 8 10 12 14 16 18 20 22 24

Pre

ssu

re [

psig

]

time [hr]

RPV Pressure

Batteries exhaust

- SRV recloses

Operator manually

opens 1 SRV SRV seizes open

Lower head

failure

Initial debris

relocation intolower head

RPV water level

(PB SOARCA LTSBO)

13

0

100

200

300

400

500

600

700

800

0 2 4 6 8 10 12 14 16 18 20 22 24

Tw

o P

ha

se

Mix

ture

Le

ve

l [i

n]

time (hr)

In-Shroud

Downcomer

RPV Water Level

Automatic RCICactuation

Operator takes manualcontrol of RCIC

RCIC steamline floods

Large scale debrisrelocation into

lower head

+5 to +35"

Batteries exhaust- SRV recloses

Main Steam Line Nozzle

Top of Active Fuel (TAF)

Bottom of Active Fuel (BAF)

SRV sticks open

Containment Pressure

(PB SOARCA LTSBO)

14

0

10

20

30

40

50

60

70

80

90

100

0 10 20 30 40 50

Pre

ssu

re [

psig

]

time [hr]

Containment Pressure

DW liner melt-through(head flange leak recloses)

SRV sticks open

DW head flangeleakage begins

Debris relocationinto lower head

LTSBO Source Term to

the Environment

15

1.E-06

1.E-05

1.E-04

1.E-03

1.E-02

1.E-01

1.E+00

0 6 12 18 24 30 36 42 48

Fra

cti

on

of

Init

ial

Co

re I

nve

nto

ry

time [hr]

Fission Products in Environment

NG

I & Te

Cd

Ba

Sn

Ce

La

Ru

Mo

Containment failure

Cs

see next figure for detail in this range

LTSBO iodine fission

product distribution

16

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0 6 12 18 24 30 36 42 48

Fra

cti

on

of

Init

ial

Co

re I

nven

tory

time [hr]

Iodine Distribution

Release toenv. (2.0%)

Captured in Suppression Pool

Deposited/Airbornewithin RPV

Drywell (negligible)

Releasedfrom fuel

LTSBO cesium fission

product distribution

17

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0 6 12 18 24 30 36 42 48

Fra

cti

on

of

Init

ial C

ore

In

ven

tory

time [hr]

Cesium Distribution

Release toenvironment (0.5%)

Captured in Suppression Pool

Deposited/Airbornewithin RPV and steam lines

Drywell (negligible)

Releasedfrom fuel

18

Speciation of Cesium and

Iodine • Based on Phebus program

findings

– Iodine treated as CsI

– Cs treated as CsI and

Cs2MoO4

• Cs2MoO4 considerably less

volatile than CsOH or CsI

– Affects retention in RCS and

long term revaporization

• Uncertainty analysis

exploring alternative

balance of speciation

– I2, CsOH, CsI and Cs2MoO4

1000 1500 2000 2500 3000 35001 10

4

1 103

0.01

0.1

1

1010

0.0001

P Cs Ti

atm

P CsI Ti

atm

P CsMo Ti

atm

P Mo Ti

atm

3.5 103

1.025 103

Ti

K

Mo

Cs2MoO4

CsI

CsOH

va

po

r p

res

su

re -

atm

Temperature – K

SRVLAM – SRV stochastic failure to

reclose (use PB IPE distribution)

19

20

BWR SRV Seizure Modeling (from PB SOARCA report)

10

in

6 in

Pilot Valve

Assembly

In severe accident conditions, high

temperature gases well exceed

design conditions

Top ~ 600K

TSA > 800 to 1100K

cycles for hours

Seizure and sticking open eventually

occurs

excessive cycling

thermal deformation

partial or full open

Valve behavior important to accident

progression 0

200

400

600

800

1000

1200

1400

0 2 4 6 8 10 12 14 16 18 20 22 24

Pre

ssu

re [

ps

ia]

time [hr]

RPV Pressure

Batteries exhaust- SRV recloses

Operator manuallyopens 1 SRV SRV sticks open

Lower headfailure

Large scale debrisrelocation intolower head

Modes of Valve

Seizure

• Excessive cycling

• Differential

thermal

expansion

• Material

deformation

SRV Thermal Failure Modeling (from PB SOARCA report)

• In severe accident conditions, high temperature

gases well exceed design conditions – Top ~ 600K

– TSA > 800 to 1100K

– cycles for hours

• SOARCA Best Estimate Boundary

Conditions: – Gas flow through the valve is based on a cycle period of

45 seconds with an open cycle duration of 5 seconds

– Gas velocity during an open cycle is constant during a

single five second cycle, but increases from 420 m/s to

500 m/s as the gas temperature increases from 600 K

to 1100 K

– Gas temperature as a function of time. Approximately

100 K temperature changes between the beginning and

end of a cycle were neglected. The mid-point values

were used

– The steam and hydrogen mole fractions were specified

as a function time 21

Approx. 3-in

6 in

Valve stem sleeve

Valve stemValve disc

Ring piston

600

650

700

750

800

850

900

950

1000

1050

1100

600 620 640 660 680 700

time [min]

Tem

pera

ture

[K

]

Gas Temp

Valve Stem: Centerline of 3-in diam SS cylinder

Valve Body: Outer surface of 1-in thick CS cylinder

Valve Body: Mid-point of 750 lb CS cylinder

~140K~20K

SRVOAFRAC – SRV open area fraction

22

SLCRFRAC – Main steam line creep

rupture area fraction

23

24

Modeling Melt

Progression Stages

U O2

Z r O2

2 9 0 0 K3 1 0 0 K

s o lid

liq u id

2 p h a s e

2 9 0 0 K

U O2

/Z r O2

Q u a s i B in a r y E q u ilib r iu m D ia g r a m

U Zr

O

UO 2 ZrO2(3 1 0 0 K )

(2 8 0 0 K )

ZrO(2 2 0 0 K )

(2 9 0 0 K )

2 2 5 0 K

3 1 0 0 K

2 6 7 3 K

s o lid

liq u id2 p h a s e

Z r ( O )U O 2

Z r ( O ) /U O2 E q u ilib r iu m P h a s e D ia g r a m

2 1 5 0 K

2 9 0 0 K

Z rZ r O2

s o lid

liq u id2 p h a s e

Z r /Z r O 2 Q u a s i B in a r y E q u ilib r iu m P h a s e D ia g r a m

Z r ( O )

T > 1000K

rapid

oxidation

UO2-

ZrO2

liquefac-

tion at

2800K

molten Zr

Breakout

2400K

2800K Zr metal melt at 2150K

UO2 dissolution in Zr metal ~20%

Zr melt breakout ~2400K

Loss of rod geometry ~2600K

UO2-ZrO2 liquefaction ~2800K

Parameters uncertain

oxidizing

Molten Zr

T > 2100K

Cumulative distribution function for flow area

resulting from drywell liner failure for PB

unmitigated LTSBO Replicate 1, with samples

identified that have a surge of water from the

wetwell during depressurization of the drywell

25

Railroad Doors

• Railroad doors opening

creates a chimney effect

leading to increased

releases to the environment;

actual opening area is not as

important as whether they

are open or closed

• This effect was noted in the

SOARCA best estimate and

has been noted in previous

MELCOR studies.

26

CV412(Refueling Bay)

CV411

CV409

(195' SE

Quadrant)CV407

(195' NE

Quadrant)

CV405

(165' SE

Quadrant)

CV404

(165' North Half)

CV403

(135' South Half)

CV402

(135' North Half)

FL407

(open hatch)

FL423

FL902(DW liner shear)

FL422

(open hatch)

CV401 (Torus Rm)

FL425 (blowout panels)

CV

90

3

(En

viro

nm

en

t)

CV401

FL414

(open hatch)

FL017(DW nom leakage)

FL401 (open floor grating)FL402 (open grating)

FL404 (E)

FL408 (E)

FL409 (W)

CV410

(195' SW

Quadrant)

FL416 FL417

Dryer - Separator Storage Pit

divides the SE/SW Quadrants

CV408

(195' NW

Quadrant)

FL415

Spent Fuel Pool Volume

included in CV412

which divides

NE/NW Quadrants of 195'

FL403 (W)

Section A-A

CV570

(Equip Access

Airlock)

FL445FL446

N

A

A

B

B

CV

90

1

(En

viro

nm

en

t)

FL424 (nom. leakage)

FL421 (roof failure)

FL

90

3

CV406

(165' SW

Quadrant)

FL410

FL904

Drywell Liner

Melt-

Through

MELCOR Uncertain Parameters

27

Sequence Issues

• Battery duration

• SRV stochastic failure rate,

thermal seizure criteria, and

open area fraction

In-Vessel Accident Progression

• Main steam line (MSL) creep

rupture open area fraction

• Zircaloy melt breakout

temperature

• Molten clad drainage rate

• Fuel failure criterion

• Debris radial relocation time

constants

Ex-vessel Accident Progression

• Debris lateral relocation time

constants

Containment & building behavior

• Drywell liner failure flow area

• Drywell head flange leakage

parameters

• Hydrogen ignition criteria

(where flammable)

• Railroad doors open fraction

Fission Product release,

transport, and deposition

• Cesium and Iodine chemical

forms

• Aerosol deposition parameters

28

Cesium Release Fractions

for the Three Categories of

SRV Failures Observed (one

horse hair for each individual

realization)

• Stochastic stuck-open SRV failures

• SRV seizures at high temperature

• SRV seizures leading to main

steam line creep rupture

• Some individual realizations singled

out for closer examination in the

report

Influences strongly affecting timing and

magnitude of environmental releases

• Whether an SRV sticks open before or after the onset of core

damage, with higher releases if SRV sticks open after core

damage

• Whether a MSL creep rupture occurs, with higher releases if MSL

rupture occurs (partial suppression pool bypass)

• The amount of cesium chemisorbed as CsOH onto the stainless

steel of RPV internals (more chemisorption means less cesium

release to the environment)

• Whether core debris relocates from the RPV to the reactor cavity

all at once or over an extended period of time (relocation all at

once leads to lower releases to the environment,

• The degree of oxidation, primarily fuel-cladding oxidation,

occurring in-vessel with greater oxidation resulting in larger

releases, and

• Whether a surge of water from the wetwell up onto the drywell

floor occurs at drywell liner melt-through (draining of surge water

into reactor building leads to larger releases).

29

Sampled parameters strongly

influencing the magnitude of the

fission product releases

• The expected number of cycles an SRV can undergo before

failing to reclose (i.e., remain in the fully open position),

• The chemical form of cesium (i.e., the amount of cesium as CsOH

as opposed to Cs2MoO4),

• The size of the breach in the drywell liner resulting from core

debris contacting and melting through the liner,

• The fractional open area of an SRV after it has failed to reseat

because of overheating,

• The time-at-temperature criterion specified for loss of “intact” fuel

rod geometry, and

• The temperature at which oxidized cladding mechanically fails.

30

Sampled parameters strongly

influencing the timing of the fission

product releases

• When the RCIC system fails

• When a SRV fails to reseat,

• The open fraction of a SRV after it fails to reseat given a

thermally-induced failure.

• The time to station battery failure

• The number of cycles an SRV can undergo before failing to

reclose (i.e., remain in the fully open position)

• The fractional open area of a SRV after it has failed to

reseat because of overheating

31

Conclusions

• The use of more advanced non-linear regression techniques

proved to be advantageous because they capture interaction

effects and non-monotonic effects missed by the linear rank

regression technique.

• Interaction effects among variables and non-monotonic effects

are common in complex systems, such as nuclear power plant

systems and environmental factors during and after a severe

accident.

• The use of select single-realization analyses in this

uncertainty analysis proved useful in validating the results of

the statistical regression analyses through phenomenological

explanations.

• A major determinant of source term magnitude is whether the

SRV sticks open before or after the onset of core damage.

– Compounding this effect is whether or not MSL creep rupture occurs,

which leads to higher environmental releases and consequences.

32

Questions?

33

Acronyms

BWR Boiling water reactor

DW Dry well

LCF Latent cancer fatality

LHS Latin Hypercube Sampling

LNT Linear no threshold

LTSBO Long-term station blackout

MSL Main steam line

ORNL Oak Ridge National Laboratory

PIRT Phenomena Identification and Ranking Table

RN Radionuclide

RPV Reactor Pressure Vessel

SME Subject matter expert

SNL Sandia National Laboratories

SOARCA State-of-the-Art Reactor Consequence Analyses

SRV Safety relief valve

SST1 Siting study source term 1

UA Uncertainty Analysis

34

BACKUP SLIDES (PARAMETER DISTRIBUTIONS)

35

BATTDUR – Battery Duration

36

SC1141(2) – Molten clad drainage rate

37

FFC – Fuel failure criterion

38

FFC – Fuel failure criterion (continued)

39

Radial debris relocation time

constants – RDSTC (solid)

40

Radial debris relocation time constants – RDMTC

(liquid)

41

Drywell head flange gap

area versus pressure

42

FL904A – Drywell liner failure flow area

43

Water level in containment for Realization 170

of the Peach Bottom unmitigated LTSBO

Replicate 1, an SRV Stochastic failure (water

drains into reactor building from DW)

44

Chemical form of Cesium

• Chemical form of Cs as CsOH (present in bins

1, 2, and 4) can have an important effect

limiting the mass available for release to the

environment due to chemisorption for high

temperature scenarios like MSL creep rupture

conditions

• CsOH can slightly increase releases for lower

temperature accident progression events like

stochastic failure of SRV

45

CHEMFORM – Iodine and cesium

fraction

46

Parameter Distribution

CHEMFORM: Five alternative combinations of RN

classes 2, 4, 16, and 17 (CsOH, I2, CsI, and Cs2MoO4)

Note the fraction cesium below represents the

distribution of 'residual' cesium which is the mass of

cesium remaining after first reacting with the amount of

iodine assumed to form CsI.

Discrete distribution

Combination #1 = 0.125

Combination #2 = 0.125

Combination #3 = 0.125

Combination #4 = 0.125

Combination #5 = 0.500

Five Alternatives Species (MELCOR RN Class)

CsOH (2) I2 (4) CsI (16) Cs2MO4 (17)

Combination #1 fraction iodine -- 0.03 0.97 --

fraction cesium 1 -- -- 0

Combination #2 fraction iodine -- 0.002 0.998

fraction cesium 0.5 -- -- 0.5

Combination #3 fraction iodine -- 0.00298 0.99702 --

fraction cesium 0 -- -- 1

Combination #4 fraction iodine -- 0.0757 0.9243 --

fraction cesium 0.5 -- -- 0.5

Combination #5 fraction iodine -- 0.0277 0.9723 --

fraction cesium 0 -- -- 1

SOARCA estimate Fraction iodine -- 0.0 1.0 --

Fraction cesium 0.0 -- -- 1.0

RRIDRFRAC, RODRFRAC – Railroad

door open fraction

47

H2IGNC – Hydrogen ignition criteria

48

RHONOM – Particle density

49