Chapter 7a - Accident Analysis Oheads · 2015-01-11 · November 19, 2006 Lecture 10 – Accident Analysis cont'd.ppt Rev. 4 vgs 8 Characteristics of Debris Bed n Top channels collapse

November 19, 2006Lecture 10 – Accident Analysis

cont'd.ppt Rev. 4 vgs 1

Accident Analysis – cont’d

Dr. V.G. Snell



Where We Are

Probabilistic RequirementsDeterministic Requirements

Design BasisAccidents

Plant safetyas operated

Safety GoalsExperience

CredibleAccidents

Plant safetyas designed

Safety CultureGood Operating

Practice

MitigatingSystems

ProbabilisticSafety Analysis

SafetyAnalysis

Chapter 2

Chapter 3

Chapter 1

Chapter 2

Chapter 4

Chapter 5

Chapter 6

Chapters 7 & 8

Chapter 9



Severe Accidents

n Severe accidentsn core geometry is preservedn fuel is damaged but remains inside intact

pressure tubes

n Severe core damage accidentsn fuel channels fail and collapse to the

bottom of the calandria



Examplesn Loss of coolant + Loss of Emergency

Core Coolingn Loss of all secondary side heat sinks +

Loss of shutdown cooling system, moderator available

n Loss of coolant + Loss of Emergency Core Cooling + Loss of moderator heat removal



System ContinuousHeat RemovalCapability (%full power)

Time to heat upand boil off, noheat removal

Moderator 4.4% > 5 hours

Shield Tank 0.4% 10 to 20 hours

Water Near Core



Loss of Core Geometry - 1n E.g., Loss of all heat sinks + inability to

depressurize HTSn Pressure rises to relief valve setpointn Loss of water through relief valvesn Overheating of fuel and pressure-tubes

at high pressuresn Failure of a few pressure tubes (6-10

MPa) – depressurize HTS



Loss of Core Geometry - 2n Remaining pressure

tubes strain to contact calandria tube

n Boil-off of moderatorn Sag & failure of

channels at lowpressure at ~1200C as moderator level falls

n Collapse of channels onto lower neighbours



Characteristics of Debris Bedn Top channels collapse when moderator is half voided,

so they sag into a pool of watern Debris likely to be composed of coarse pieces of

ceramic materialsn Bed will not be molten until all the moderator water

is boiled off - will then dry out and heat up due to decay heat & remaining Zircaloy-steam reaction

n No energetic fuel-coolant interactionn Models for heat transfer from debris bed to calandria

walls developed by T. Rogers et al. for dry debris, and also debris with molten centre



Debris Bed Modelsn Uniform porous mixture

of UO2, ZrO2 and/or Zircaloy

n Fuel decay heat + metal water reaction

n Thermal radiation to inner surface of calandria from top of the bed

n Conduction through bottom of calandria to shield tank water



Debris Bed Heatupn Melting of

debris starts about 7 hours after the event

n Upper & lower surfaces of debris bed stay below melting temperature



Calandria Wall Temperaturesn Outer surface

temperature below 140C

n Stainless steel wall

n Do not expect creep under applied stresses



Heat Flux to Shield Tankn Heat flux to shield tank

15 times less than CHFn Calandria will remain

intact while shield tank water boils off

n Behaviour insensitive to porosity and timing of metal-water reaction

Critical heat flux200 W/cm2



Summary of Time-scalesTime (hr) Event

0 Loss of heat sinks, reactor shutdown

0.75 Steam Generators boil dry, liquid relief valvesopen, fuel cooling degrades

0.83 A few pressure tubes fail and depressurize heattransport system

0.86 High pressure ECC initiated; medium pressureECC assumed to fail

1.1 Heat transport system empty

5 Moderator boiled off, channels sagged to bottomof calandria

25 Vault water boiled off to top of debris bed,calandria fails

Days Interaction of debris with vault floor &penetration to containment basement



Uncertainties

n Mechanical and thermal behaviour of end-shields

n Capability of shield tank to relieve steamn Local effects in molten pool and hot-

spotsn Lack of experimental validation of debris

melting transientn Demonstration of core collapse mode



Tests on Channel Collapsen 1/5 scale studyn Scaling retains full

size stress levels, ratio of bundle size to channel length and channel length to pitch height of assembly



Test Rig



Channel Failure Mode



Channel Collapse – Early Results

n Significant sag occurs only above 800C.n Sag is creep-controlledn PT wall thins at the bundle junctions -

debris may be two to three bundles long

n The end-load is not sufficient to pull out the channel from the end-fitting



Containmentn Containment heat removal (local air coolers)

may or may not be available depending on the accident

n If not available, pressure initially controlled by dousing sprays

n Will eventually rise above design pressuren Structure will remain intact due to leakage

through cracks and pressure relief



Observationsn Severe core damage in CANDU is very different from LWRsn Low power density (16 MW/Mg of fuel at full power)n Long heatup times (hours)n Gradual collapse of the core into a coarse debris bedn Dispersion of the debris in the large calandria

n shallow molten pool about 1 metre deep

n Presence of two large sources of water in or near the coren Potential to stop or slow down the accident at two points:

n channel boundary (moderator)n calandria boundary (shield tank)

n Hydrogen control a necessity in short- and long-term



Coherent vs. IncoherentLWRsn Fuel melts rapidly

(minutes) and “candles” down to the bottom of the vessel

n Vessel fails suddenly ejecting molten fuel

n Potential for steam explosion in vessel and in containment

CANDUn Fuel melts slowly

(hours) and slumps gradually down to the bottom of the vessel

n Vessel fails after a day; shield tank provides further barrier

n Continual steam release - explosion less likely



Severe Accident Conclusionsn Severe accident mitigation requirements for new

reactors stress two design measures:n core debris spreading arean ability to add water to cool debris

n CANDU: calandria spreads the debris, and shield tank provides cooling water

n Long time scales allow for severe accident counter-measures and emergency planning

n Independent makeup to moderator and shield tankfor EC6 and ACR

n Future: backup or passive containment heat removal



Uncertainty Analysisn Disadvantages of conservative

approach:n Overestimates / exaggerates riskn Misleadingly small marginsn Useless as operator guiden Distorts safety resource allocation

n UA useful only with best-estimate methods (BE+UA)



Types of Uncertaintyn Physical models

n Model biasn Experimental scattern Example: WIMS predicts coolant void reactivity for

37-element fuel with a bias of –1.6 mk. and an experimental uncertainty of ±1mk.

n Plant idealizationn Example: How many spatial nodes for

convergence?n Plant data

n Example: Uncertainty in flow measurement



Focus of Uncertainty Analysis

n UA stated only for key safety parameters output by the code & compared to acceptance criterian E.g., peak fuel temperaturen E.g., not internal parameter such as fission

gas pressure



Issuen Assume there are < 8 contributors to a key

parametern E.g., key parameter is fuel temperature in LOCA

power pulsen Contributors are void reactivity, sheath-to-coolant

heat transfer, delayed neutron fraction etc.n Cannot afford to vary 8 simultaneously using

fundamental codesn Surrogate needed (curve fitting or functional

form)



Simplified Methodologyn Generate cases for

variation of ncontributory parameters using fundamental code

n Fit surface with functional form

n Test goodness of fitn Sample surface based

on statistical distribution of each parameter

Probability distribution ofContributory parameter(2 examples)



Graphical ExampleKey output parameter

n1, n2, n3

Key output parameter

n1

Code run

Functional or curve fit

Mean value &95% confidencelevels

Documents

Chapter 7a - Accident Analysis Oheads · 2015-01-11 · November 19, 2006 Lecture 10 – Accident Analysis cont'd.ppt Rev. 4 vgs 8 Characteristics of Debris Bed n Top channels collapse