Disowning Fukushima-Managing the Credibility of

Disowning Fukushima:
How Risk Assessments Erase Nuclear Accidents.
1st INRAG Conference
Vienna
Apr 16-17 2015
John Downer
John.Downer@Bristol.ac.uk
“The ‘Zero-risk mindset’
is a thing of the past.”
(?)
Fukushima’s costs
•
Unknown and contested, but
potentially very severe fallout
hazards. (Health &
environmental).
•
Enormous financial costs.
(Japan Center for Economic
Research, late 2011, estimates
US$ 520-650 billion).
•
Heavy social burden (88,000
people evacuated. Most still in
temporary housing. Many lost
their employment.)
Cost-benefit analyses continue to
ignore accidents…
Site planners continue to ‘cluster’ reactors in
single locations…
Evacuation / Resilience-planning remains
inadequate & unrealistic.
Most nuclear construction
plans continue…
Why?
The 'zero-risk mentality' is justified, institutionally,
by deterministic and probabilistic safety analyses.
The 'zero-risk mentality' is justified, institutionally,
by deterministic and probabilistic safety analyses.
eg: Areva assert
that the likelihood of
a “core damage
incident” in its EPR
is of the order of one
incident (per
reactor) every 1.6
million years.
There are compelling logical
reasons to believe that nuclear
reliability calculations cannot
work…
…but these explanations are
unnecessary, given the
empirical evidence that
nuclear reliability calculations
do not work.
Nuclear plants fail far more often than expert
assessments predict…
Fukushima (2011) (x3)
Three Mile Island (1979)
Historical failure
rate of 1 in
every 1,300 3,600 years (per
reactor)
Windscale (1957)
Chernobyl (1986)
Question:
Why hasn't Fukushima
discredited nuclear risk
assessment (in most
countries)?
(Why does the ‘zero-risk
mindset’ persist?)
Four ‘narratives’ or ‘defenses’:
1. The interpretive defense: (The argument that the
assessments did not actually fail).
2. The relevance defense: (The argument that the failure of
one assessment is not relevant to the credibility of others).
3. The compliance defense: (The argument that the
assessments were sound, but people did not obey the
rules).
4. The redemption defense: (The argument that the
assessments were flawed, but now they are fixed).
1. Interpretive:
Experts can deny that the assessments (or the plant) actually failed.
There are different ways to do this:
1. Interpretive:
Experts can deny that the assessments (or the plant) actually failed.
There are different ways to do this:
•
Be economical with the truth.
•
Parse the definition of ‘failure’. (eg: If accident is ‘contained’, then
it is possible to say that the assessments took the failure into
account.)
•
Argue that the conditions faced by the plant were beyond its
‘design basis’. (ie: say that it wasn’t designed to withstand an
earthquake / tsunami of that magnitude).
1. Interpretive:
Experts can deny that the assessments (or the plant) actually failed.
There are different ways to do this:
•
Be economical with the truth.
•
Argue that the conditions faced by the plant were beyond its
‘design basis’. (ie: say that it wasn’t designed to withstand an
earthquake / tsunami of that magnitude).
“[Fukushima] could actually be considered a
‘success’ given the scale of this natural
disaster that had not been considered in the
original design.”
~American Nuclear Society (April 2011)
1. Interpretive:
But…
•
The fact that Fukushima represents a technical ‘failure’ is now
unambiguous.
•
The design basis of a nuclear plant is an essential element of
any reliability assessment. (To say that an event was
unexpected because it was ‘beyond design basis’, is to say
that an essential element of the reliability calculation was
wrong.)
2. Relevance:
If accounts can show that Fukushima’s (or, more widely, Japan’s) risk
calculations were significantly different from other such calculations,
then the failings of those calculations can be theirs alone.
Thus one can contest the ‘relevance’ of the accident and its assessment
by emphasizing differences between the failing system and others.
eg: by highlighting the plant…
“Using a plant built 40 years ago to
argue against 21st-century power
stations is like using the
Hindenburg disaster to contend
that modern air travel is unsafe.”
(George Monbiot, in The Guardian, 2011)
2. Relevance:
If accounts can show that Fukushima’s (or, more widely, Japan’s) risk
calculations were significantly different from other such calculations,
then the failings of those calculations can be theirs alone.
Thus one can contest the ‘relevance’ of the accident and its assessment
by emphasizing differences between the failing system and others.
…or the regulatory regime:
2. Relevance:
But…
•
•
The fact that the technology is changing does not speak to the risk
assessment.
In many ways Japan’s / TEPCO’s procedural & regulatory failures
mirror those of other regulatory regimes.
3. Compliance:
It is possible to argue that the assessments were sound, but
people did not obey the rules.
• All reliability assessments embody implicit and explicit caveats
about compliance, such as: “given proper maintenance,” or “if
handled correctly.”
• Thus it is possible to exculpate the assessments after a disaster
by highlighting operator error, noncompliance, or malfeasance.
3. Compliance:
But…
•
No complex socio-technical system in history has been able to
eradicate all human failings, or provide adequate rules for every
contingency.
•
An assessment that fails to account adequately for humanerror/malfeasance/non-compliance is missing an essential
variable.
4. Redemption:
Experts can concede the existence of errors in the plant’s
reliability calculations and then argue that experts have identified
the errors, altered the assessments, and remedied the problem.
eg: 2012, NRC announced that US
Nuclear safety assessments "...do
not adequately weigh the risk of a
single event that would knock out
electricity from the grid and from
emergency generators, as a quake
and tsunami recently did in Japan".
…it then promised to reframe its
assessments accordingly.
4. Redemption:
But…
•
Who is to say there are not more errors hiding in the
calculations?
•
If the ‘real’ lesson of Fukushima is that all reliability
assessments can contain hidden flaws, then there can be no
‘fix’.
Recap:
When the credibility of assessment practices is challenged by a
disaster, it is possible to say:
1. “The assessments are fine, this accident wasn’t really a failure.”
(Interpretive defense)
2. “Assessment practices (and plants, etc) differ, so the fact that one
was wrong doesn’t mean all are wrong.” (Relevance defense)
3. “The assessments are fine, but people failed to follow the rules. It
will never happen again. “ (Compliance defense)
4. “The assessments were wrong, but now they are correct.”
(Redemption defense)
Conclusion
In focusing on the close lessons of Fukushima,
public discourse is obscuring a larger lesson: that
the safety of nuclear plants is not a property that
can be predicted with objective rigor.
A wider understanding of this would have significant
ramifications for nuclear policymaking and
scholarship.
Thank You
For more see: Downer, J (2014) “Disowning Fukushima: Managing the Credibility of Nuclear
Reliability Assessment In The Wake Of Disaster” In Regulation and Governance.
Or email: john.Downer@Bristol.ac.uk