Risk M anagement and Risk Perceptions

May 3, 2004

5/3/04

Nuclear Energy Economics and

Policy Analysis

1

Elements of Risk Management

• Risk assessment

– What is the probability of an accident? What are the likely consequences?

• Risk management

– Prevention and mitigation

– External regulation vs. self-regulation

• Risk communication

– Informing the public about risk, and responding to expressed concerns

5/3/04

Nuclear Energy Economics and

Policy Analysis

2

Probabilisti c Ris k Assessmen t (PRA)

• A methodology for answering three questions:

– What can go wrong (accident scenario)?

– How likely is this to occur (probability, frequency)?

– What will be the outcome (consequences)?

• Two key tools:

•

•

5/3/04

Event tree analysis

Fault tree analysis

Nuclear Energy Economics and Policy Analysis

3

Definition of Risk :

Risk (consequences/unit time) = frequency (event/unit time) x

magnitude (consequence/event)

PRA: Event Tree Analysis

• “An analytical technique for systematically

identifying potential outcomes of a known initiating event.”

– Select candidate initiating event

– Using inductive reasoning, construct sequences of subsequent events or scenarios that end in a ‘damage state’

– Estimate probability of each event on the pathway leading to the accident

5/3/04

Nuclear Energy Economics and

Policy Analysis

4

LNG Accident Event Tree

5/3/04

Nuclear Energy Economics and

Policy Analysis

5

Probability o f disaster = p 1 x p 2 x p 3 x p 4 Sum of probabilities of all out comes = p 1

Probability o f no consequence given an ac cident = p 2 (1- p 4 ) + (1- p 2 ) Probability o f a small consequence giv en an a ccident = p 2 x (1- p 3 ) x p 4

Source: Adam Markwoski, Oi l an d Ga s Journal , 9 September 2002

Courtesy of Adam Markwoski and Oil and Gas Journal. Used with permission.

5/3/04

Nuclear Energy Economics and

Policy Analysis

6

Light Water Reactor Safety Philosophy: Defense-in-depth

5/3/04

Nuclear Energy Economics and Policy Analysis

7

A

Pipe Break

B

Electric Power

C D E

Emergency Fission- Containment Overall Core-Cooling Product System Probability System Removal of sequence

System

P A

P E 1

P A x P E 1

P A x P D 1

Succeeds

P D 1

P E

P A x P D x P

1 E 2

2

P

E 3

P A x P C 1

P A x P C 1 x P E 3

P C 1

Initiating Event

P A

P A x P C x P D

1

2

P D 2

P E

P A x P C x P D x P

1

2 E 4

4

P A x P B

P E 5

fails

P B

P D 3

P E 6

P A x P B x P E 5

P A x P B x P D 3

P A x P B x P D 3 x P E 6

P A x P B x P C

2

P C 2

P E 7

P A x P B x P C 2 x P E 7

P A x P B x P C x P D

2

4

P D 4

P E

P A x P B x P C x P D x P

2

4 E 8

8

Source: Reactor Safety Study W ASH-1400 analysis of the 1975 Brown's Ferry accident After Lewis, 1980.

5/3/04

Nuclear Energy Economics and Policy Analysis

8

PRA -- Fault Tree Analysis

1) “An analytical technique whereby an undesired state of the system is specified, and thesystem is

then analyzed in the context of its e nvironment and operation to find allcredible ways in which the undesired event can occur.

2) ”Use deductive reasoning to think of all possible ways in which the ‘top event’ couldhave occurred.

3) Then estimate the probabilities (relying on empirical data for the most basic events,and algebra to get combined probabilities.)

Failure to maintain fluid

inventory in reactor vessel during second phase of accident

P = .003

P = .05

P = .06

Failure to restore operation of isolation- cooling system in reactor core within two hours

Failure to restore operation of high-pressure coolant-inj ection system within two hours

Failure to valve in other possible sources of coolant-rod-drive pump within two hours

P = . 4

P = 1

P = .12

Failure of remaining relief val v es and failure to repair at least one of 1 1 such valves within two hours

P = .06

Failure to repair main steam-isolation valve within two hours

P = 1

Failure to valve in flow by-passing control-rod- drive pumps within two hours

Failure to repair stand-by liquid-control system for coolant delivery within two hours

P = . 3

P = .4

5/3/04

Source: Reactor Safety Study W ASH-1400 analysis of the 1975 Brown's Ferry accident After Lewis, 1980.

Nuclear Energy Economics and Policy Analysis

9

Failure to restrict pressure in reactor cooling system to less than 350 p.s.i. by steam relief

Failure to provide high-pressure delivery of coolant to augment control- rod-drive pump during two-hour period when pressure in reactor-cooling system was high

Image removed due to copyright considerations.

5/3/04

Nuclear Energy Economics and Policy Analysis

10

• The value of PRA:

– Forces systematic attention to accident scenarios

– Structures debate about differences in scenario definition or parameter estimation

– Identifies ‘most bang for the buck’ components, subsystems

– Provides quantitative estimates of failure probabilities and risks

• Problems:

– Is the list of initiating events exhaustive?

– Can the probability of events and failures be estimated?

– Common mode failures

– Lack of alignment with public risk perceptions?

5/3/04

Nuclear Energy Economics and

Policy Analysis

11

A ssessing PRA

Risk perceptions

 People often express great anxiety about hazards that technical analyses indicate pose very low risks, yet are indifferent to other hazards about which experts are much more concerned.

 The experts measure risk as the product of probability and consequence.

– No difference between activities with a high likelihood of causing a small number of fatalities and those with a low likelihood of causing a large number of fatalities.

– If the expected number of fatalities is the same, the risk, according to this measure, is also the same.

– Yet many people seem to be much more concerned about low-probability accidents with high consequences.

• How are people’s perceptions and beliefs about risks formed? What causes these perceptions to change?

5/3/04

Nuclear Energy Economics and

Policy Analysis

12

“Some people say that the prime responsibility

for reducing exposure of workers to dangerous substances rests with the workers themselves, and that all substances in the workplace should be clearly labeled as to their levels of danger and workers then encouraged or forced to be careful with these substances. Do you agree or disagree?”

From a survey of public attitudes towards the chemical industry

5/3/04

Nuclear Energy Economics and

Policy Analysis

13

M easuring risk perceptions is not straightforw ard:

E.g. , it depends on how you ask the question.

5/3/04

Nuclear Energy Economics and

Policy Analysis

14

Question. Rank the risk of death from the following activities:

Source: Paul Slovic, Baruch Fischoff and Sarah Lichtenstein,

"Facts and Fears: Understanding Perceived Risk", in R. Schwing and W.A> Albers, Jr (eds), Societal Risk Assessment: How Safe is Safe Enough?, New York, Plenum (1980): 181-214

5/3/04

Nuclear Energy Economics and

Policy Analysis

15

Two possible explanations for the divergence

between lay people’ s perceptions of risk and the actual fatality data:

1. Members of the public base their judgments of

risk on factors other than expectations of annual fatalities

2. Public risk perceptions are based on

expectations of fatalities, but these expectations are inaccurate.

5/3/04

Nuclear Energy Economics and

Policy Analysis

16

Question: How many people are likely to die in a typical

year from these activities?

Source: Paul Slovic, Baruch Fischoff and Sarah

Lichtenstein, "Facts and Fears: Understanding Perceived Risk", in R. Schwing and W.A> Albers, Jr (eds), Societal Risk Assessment: How Safe is Safe Enough?, New York, Plenum (1980): 181-214

5/3/04

Nuclear Energy Economics and

Policy Analysis

17

Question: How many people are likely to die in a typical

year from these activities?

Source: Paul Slovic, Baruch Fischoff and Sarah

Lichtenstein, "Facts and Fears: Understanding Perceived Risk", in R. Schwing and W.A> Albers, Jr (eds), Societal Risk Assessment: How Safe is Safe Enough?, New York, Plenum (1980): 181-214

5/3/04

Nuclear Energy Economics and

Policy Analysis

18

O ther qualitative factors of risks that affect

public perceptions

• Controllability

• To what degree can people exposed to the risk avoid death by their own skill or diligence

• Immediacy

• Is the risk of death immediate, or more likely to occur at a later time

• Severity

• How likely is it that the consequences of an accident will be fatal

• Knowledge of risk

• To what extent is the nature of the risk understood by those exposed to it, and by the scientific community?

• Dread

• Is the risk one that people have learned to live with, or is it one that inspires feelings of dread?

5/3/04

Nuclear Energy Economics and

Policy Analysis

19

Source: Paul Slovic, Baruch Fischoff and Sarah Lichtenstein, "Facts and Fears:

Understanding Perceived Risk", in R. Schwing and W.A> Albers, Jr ( eds), Societal Risk Assessment: How Safe is Safe Enough?, New York, Plenum (1980): 181-214

5/3/04

Nuclear Energy Economics and

Policy Analysis

20

Source: Paul Slovic, Baruch Fischoff and Sarah Lichtenstein, "Facts and Fears:

Understanding Perceived Risk", in R. Schwing and W.A> Albers, Jr ( eds), Societal Risk Assessment: How Safe is Safe Enough?, New York, Plenum (1980): 181-214

5/3/04

Nuclear Energy Economics and

Policy Analysis

21

Two Alternative Responses to Findings on

Risk Perceptions

• Positio n 1 ( ‘ Rationalist ’ view) : Quantitative evidence and estimates on fatalities, injuries and damage are the only basis on which to make design and technology selection decisions. Augment quantitative risk estimation methods with more effective risk communication strategies.

• Positio n 2 ( ‘ Populist ’ view) : Technical choices should

reflect the full spectrum of society’s risk preferences, including qualitative as well as quantitative risk attributes.

5/3/04

Nuclear Energy Economics and

Policy Analysis

22

An engineering-based risk communication

strategy

Measure

‘actual’ risks (PRA)

Measure risk

perceptions

Deviation is a

measure of ‘ignorance’

Focus risk

communication and education wh ere the deviation i s highest

5/3/04

Nuclear Energy Economics and

Policy Analysis

23

Compare risk perceptions with ac tual risks from best available technology

Sunstein on ‘probability neglect’ -- rationalist view

People often focus on the goodness or badness of outcomes, and pay too little attention to the probability that a good or bad outcom e will occur.

In such cases, people fall prey to ‘ probability neglect’.

Probability neglect is especially large when people focus on the worst possible case or otherwise are subject to strong emotions.

When such emotions are at work, people do not give sufficient consideration to the likelihood whether the worst case will occur.

Experts are mostly concerned with the number of lives at stake, and are thus closely attuned, as ordinary people are not, to the issue of probability.

When ordinary people suffer from probability neglect, they are exhibiting behavior that is not fully rational. It is not true to say that they have a kind of ‘ richer rationality’, that is superior to that of the experts .

-- Cass Sunstein, “Probability Neglect: Emotions, Worst Cases, and Law”, U.. Pa L. Rev.

5/3/04

Nuclear Energy Economics and

Policy Analysis

24

But still, what should the government do?

Example: Suppose that p eople are af raid of arsenic in drinking water, and that the y demand steps to provide assurance that arsenic levels won’t be hazardous. Even if the r isks at existing levels are infinitesimal, should the government refuse to do what people want it to do ? What if the costs of not acting – the costs of continuing public fear – a re very high? Shouldn’t the go vernment a ct to reduce public f ear?

“At first glance, the government should not respond if the public is d emanding attention to a statistically miniscule risk, and d oing so simply because people are visualizing the worst that can happen. The best response is information and edu cation. But public fear is itself an independent concern, and it can represent a h igh cost in itself and lead to serious associated costs, often in the form of ‘ripple effects’. If public fear cannot be alleviated without r isk reduction, then government should engage in risk reduction, at least if the relevant steps are justified by an assessment of c osts and benefits.”

-- Cass Sunstein,

5/3/04

Nuclear Energy Economics and

Policy Analysis

25

The future of nuclear power:

Passive safety vs. defense-in-depth

Ligh t wate r reactors : defense-in-depth

5/3/04

Nuclear Energy Economics and

Policy Analysis

26

PR A result

Frequency of core-damage accidents:

10 -4 per year

i.e. , 1 in 10,000 reactor -years of LWR operation

Helium-cooled, graphite-moderated, modular pebble-bed reactor

• Hundreds of thousands of tennis-ball-sized graphite pebbles

• Each pebble contains thousands of uranium-oxide particles, ~0.5 mm in diameter, coated with layers of carbon and silicon carbide to prevent fission product escape

• Ordinarily, core is cooled by high pressure helium gas, which either drives gas turbine directly or generates steam to drive steam turbine

• Loss of coolant accident: helium has v. low heat capacity, so all heat initially absorbed by graphite pebbles.

• Pebbles are thermally stable and retain integrity even at very high temperatures.

• Even in worst-case scenario -- withdrawal of control rods, depressurization of core, and complete loss of coolant, fuel remains intact with no requirement for active cooling (passive heat removal by thermal conduction and radiation sufficient to remove decay heat)

5/3/04

Nuclear Energy Economics and

Policy Analysis

27

28

PBMR Design Certification

How much is ‘passive safety’ worth?

• Reduced need for engineered emergency safety systems

• No need for massive containment? (Aircraft/missile strikes)

• Safety more transparent: physically demonstrate safe shut-down in worst case conditions vs. reliance on complex simulations and PRA calculations.

5/3/04

Nuclear Energy Economics and

Policy Analysis

29