CANDU Safety

#2 - Risk from Nuclear Power Plants

Dr. V.G. Snell

Director
Safety & Licensing

24/05/01 8:18 AM CANDU Safety - #2 - Risk from Nuclear Power Plants.ppt Rev. 0 vgs 1
What is the Public Hazard?
λ chemical? Chlorine for water treatment as in fossil plants
λ biological? None
λ physical? Nuclear explosion impossible
λ radiological? Small risk of delayed health effects, very small risk of
prompt health effects, even in severe accidents

24/05/01 8:18 AM Nuclear Reactor Safety - An Overview.ppt Rev. 0 vgs 2

The Chernobyl Disaster
λ More than 12,500 of the λ For a population of the age and sex
350,000 people who distribution of the “liquidators” in 1986, the
worked on the Chernobyl normal mortality rate was 3 per 1000 per
cleanup have since died year. Thus the “expected” number of deaths
would be:
350,000 people x 12 years x 3/1000
= 12,600
λ The number should be larger (by 50%)
because the normal rate of 0.3% increases
as the group ages
λ Is reporting inadequate? Does monitoring
improve the life expectancy of the
liquidators?

24/05/01 8:18 AM Nuclear Reactor Safety - An Overview.ppt Rev. 0 vgs 3

Effects of Radiation
λ prompt health effects (deterministic, non-stochastic)
– dose of >1 Sv: illness
– dose of >3 Sv: increasing risk of death (LD 50 is 3 to 10 Sv)
λ delayed health effects (random, stochastic)
– risk of cancer
λ 0.25 Sv gives approx. 0.5% increase in individual risk
– risk of damage to foetus
– risk of genetic damage
λ not observed in humans

24/05/01 8:18 AM Nuclear Reactor Safety - An Overview.ppt Rev. 0 vgs 4

 EXAMPLES OF RADIATION DOSE
0.000001 Max. in Canada, banned food
0.00003 Typical from CANDU
0.00083 Maximum, Three Mile Island
0.002 Natural, in Toronto, /yr.
0.005 Accident limit in CANDU (single failure)
0.01 Natural, in Kerala, /yr.

0.25 Severe Accident limit in CANDU (dual failure)

1 Nausea
Firefighters at Chernobyl 10

0 2 4 6 8 10 Dose (Sv )

24/05/01 8:18 AM Nuclear Reactor Safety - An Overview.ppt Rev. 0 vgs 5

What Is Risk?

Risk = Frequency of an event x consequences of the event

λ Examples of risk:
– annual individual risk of death
– annual nuclear plant risk of core damage
– annual nuclear plant risk of a large release of radioactivity
– risk of psychotic reaction to malaria drug, per dose

24/05/01 8:18 AM CANDU Safety - #2 - Risk from Nuclear Power Plants.ppt Rev. 0 vgs 6
Safest and Most Dangerous Occupations*
Occupation Fatalities
/ 100,000 / year
Administrative support, clerical 1
Executive & Managerial 3
News Vendors 16
Police 17
Truck drivers 26
Farm Workers 30
Construction labourers 39
Miners 78
Pilots & navigators 97
Lumberjacks 101
*US, 1995
Sailors 115
24/05/01 8:18 AM CANDU Safety - #2 - Risk from Nuclear Power Plants.ppt Rev. 0 vgs 7
“Acceptable” (since accepted) Occupational
Risk?
5 per 100,000 per year (5 x 10-5 per year)
to
100 per 100,000 per year (1 x 10-3 per year)

24/05/01 8:18 AM CANDU Safety - #2 - Risk from Nuclear Power Plants.ppt Rev. 0 vgs 8
Non-Occupational Accidental Fatalities*
Accident Fatalities
/ 100,000 / year
Lightning 0.06
Poisoning 1.5
Firearms 1.1
Drowning 3.6
Fires 3.6
Falls 8.6
Motor vehicle 27

*US, 1970

24/05/01 8:18 AM CANDU Safety - #2 - Risk from Nuclear Power Plants.ppt Rev. 0 vgs 9
“Acceptable” (since accepted) Public Risk?
4 per 100,000 per year (4 x 10-5 per year)
to
27 per 100,000 per year (3 x 10-4 per year)

Total risk of accidental death = 4 x 10-4 per year

Note that these are population-average risks

Some groups will be considerably more (or less) at risk than
others.

24/05/01 8:18 AM CANDU Safety - #2 - Risk from Nuclear Power Plants.ppt Rev. 0 vgs 10
Many Factors Determine “Acceptability”
λ occupational risk vs. public risk
λ presence of offsetting benefit
λ voluntary vs. involuntary risk
– can one really eliminate risk from motor vehicles by not
driving??
λ “dread” factor (cancer vs. automobile accident)
λ perceived ability to control risk
λ knowledge and familiarity (coal mining vs. operating nuclear
plant)

24/05/01 8:18 AM CANDU Safety - #2 - Risk from Nuclear Power Plants.ppt Rev. 0 vgs 11
Safety Goals for Nuclear Power Plants
λ Safety goal - an acceptable value of risk
– risk from NPPs chosen to be very small in comparison to
comparable activities
λ Risk of prompt fatality from NPP should be << risk of prompt

fatality from all other causes

λ Risk of fatal cancer from NPP should be << risk of cancer from

all other causes

Risk of fatal cancer just from “natural” radiation in Canada =
0.002Sv/year x 0.02 cancers/Sv = 4 x 10-5 per year
(according to linear dose-effect hypothesis)

24/05/01 8:18 AM CANDU Safety - #2 - Risk from Nuclear Power Plants.ppt Rev. 0 vgs 12
Risk Goals
The only significant health effects from a nuclear
power plant are from a large release

A large release can only occur if:

1) There is severe core damage, and
2) The containment does not work or is damaged

Nuclear safety goals therefore focus on:

1) preventing a large release
2) preventing severe core damage

24/05/01 8:18 AM CANDU Safety - #2 - Risk from Nuclear Power Plants.ppt Rev. 0 vgs 13
Example #1
λ Three Mile Island
– severe core damage (~20 tons of molten fuel)
– the pressure vessel was thinned but did not fail
– the containment was not damaged but some liquids and
gases escaped through lines which bypassed the
containment
– public health effects were minor: ~1 additional (statistical)
cancer case in the surrounding population

24/05/01 8:18 AM CANDU Safety - #2 - Risk from Nuclear Power Plants.ppt Rev. 0 vgs 14
Example #2
λ Chernobyl
– the core was severely damaged due to a reactivity increase
which was made worse by the shutdown systems
– the containment was ineffective as the steam explosion
blew off the top cover of the reactor & exposed the core
– about 32 prompt fatalities among station staff
– most volatile fission products were released to atmosphere
– public health effects: predict several thousand (additional)
cancer cases in the surrounding area
– an increase in thyroid cancers in children has been
observed (mostly curable)
24/05/01 8:18 AM CANDU Safety - #2 - Risk from Nuclear Power Plants.ppt Rev. 0 vgs 15
Numerical Safety Goals for Nuclear Power Plants
λ For existing nuclear power plants:
– risk of a severe core damage accident must be < 10-4 per
plant per year
– risk of a large release must be < 10-5 per plant per year
λ For new nuclear power plants:
– factor of 10 lower on both counts
λ the factor of 10 must therefore come from:
– severe accident management & mitigation procedures
– residual containment effectiveness

24/05/01 8:18 AM CANDU Safety - #2 - Risk from Nuclear Power Plants.ppt Rev. 0 vgs 16
How is Risk Calculated?
λ For frequent events - easy - just collect the observed statistics
λ For rare events - build up from combinations of more frequent
components
λ e.g., risk / year of plane crash on Shanghai University =
risk of a plane crash per kilometer of steady flight
x number of flights / year landing or taking off from Shanghai
airport
x fraction of flights which fly over the University
x diameter of University in km.
– does not account for evasive action, skyjacking

24/05/01 8:18 AM CANDU Safety - #2 - Risk from Nuclear Power Plants.ppt Rev. 0 vgs 17
Fault trees and Event trees
λ to determine the risk from rare events:
– calculate frequency or probability of a system failure (fault
tree)
– calculate consequences of the system failure (event tree)
– in the event tree, assume each mitigating system either
works or fails; if it fails, account for the probability of failure
λ end result is the frequency or probability and consequences of
a family of events

24/05/01 8:18 AM CANDU Safety - #2 - Risk from Nuclear Power Plants.ppt Rev. 0 vgs 18
Douglas Point
λ an early risk assessment in Canada in the 1960s for the first
prototype CANDU
λ goal: risk from nuclear power plant must be 5× less than coal
λ only prompt effects well known then, so compared prompt
fatalities from mining and nuclear power
λ e.g., large release frequency = initiating event frequency ×
unavailability of shutdown × unavailability of containment
λ must set targets for & measure:
– frequency of initiating events (process system failures)
– unavailability of each safety system

24/05/01 8:18 AM CANDU Safety - #2 - Risk from Nuclear Power Plants.ppt Rev. 0 vgs 19
Frequency and Reliability Targets
λ process system failures:
– must be less than 0.3 events / year
– deliberately chosen high so it could be confirmed
λ safety system unavailability:
– each must be less than 10-3 years / year (8 hours / year or 1
failure in 1000 tries)
λ can one multiply the numbers?
– e.g., small LOCA + LOECC + containment failure to isolate
= 10-2 / year × 10-3 years/year × 10-3 years / year
= 10-8 / year ???
λ only if there are no cross-links
24/05/01 8:18 AM CANDU Safety - #2 - Risk from Nuclear Power Plants.ppt Rev. 0 vgs 20