Nuke Stupidity Rebuttal
Carl Fago
US reactor plants.
From one who has seen both sides (of the US plants, nuclear and coal), the
nuclear plants are much safer. The welds are of much higher quality. The
pipe has a higher pedigree.
The article was written just so the writer could blow off his mis-informed
opinion and not relate _any_ facts.
BTW, a lot of the "backwards" stuff at that California plant has been
analyzed in its new configuration and accepted for use. Doesn't mean it
isn't safe.
*-=Carl=-* INTERNET - cd...@psuvm.psu.edu | Be wary of strong drink. |
DELPHI - WULFGAR GEnie - C.FAGO1| It can make you shoot at tax |
Carl Fago State College, PA | collectors -- and miss! |
Ken B Kirksey
The follwing column appeared in the Monday, September 21 edition of the
_Charlotte Observer_ (Charlotte, NC).
-----------------------------------------------------------------------------
ACCIDENTS WAITING TO HAPPEN
* Nuclear power plants are a glowing example of American insanity
by Howard Fast
The Greenwich Time
We have a nephew who is a steam engineer, a very good one with an
international reputation. He is called into service in a number of
countries as a consultant for electric utilities, atomic-powered and
otherwise. But he avoids, whenever possible, work on American atomic power
plants.
He says that when he works on a native atomic power plant and sees the
quality of the installation, he is nervous every hour he spends on such a
job, cannot wait to get out, and has nightmares for months afterward. He
holds that the work is so sloppy, pipes and joints so badly put together,
wiring so defective, that every plant he worked in is, for him, a ghastly
accident waiting to happen.
Now, in the current issue of Washington Monthly, in an article on nuclear
power by Christopher Georges, I find the following: "Twenty-four-year-old
Jim Horn an engineer new to Califomia's Diablo Canyon nuclear power
plant, was flipping through a copy of the reactor's blueprints when he
spotted a design flaw so massive it could have been a Cal-Tech marching
band prank. Huge Segments of the plant, which was just weeks away from
full-power operation, had been assembled backward. Many of the cables
and supports for the two reactors, which were mirror images of each other,
had been erected exactly opposite where they should have been. Although
construction of the $2.3 billion Southem California plant was 97%
complete, no one before Horn Q not the thousands of engineers and
construction workers, not the Nuclear Regulatory Commission's team of
inspectors Q had spotted the screw-up."
The incident described above happened 11 years ago, but nothing our
nephew told us leads me to think that the situation has improved since
then. Almost every week, there is an accident or a necessary shut-down in
one of our atomic power plants, and every one of these plants is
potentially destructive enough to take an area the size of Massachusetts
out of human habitation for generations. And in all truth, beyond the fact
that they are enormously profitable to those who build them and supply
the atomic fuel, there is no real necessity for atomic power.
We have enough natural gas, combined with other sources, to provide clean
energy for years to come. New agricultural techniques make it possible to
grow sufficient plant sources of fuel to take care of any petroleum
shortfall. We have vast sources of coal, and scientists are finding new
nonpolluting ways to use it, and the advances in wind-power and sun-power
open a whole new, clean world of power.
When one thinks of the trillions of dollars that have been poured into
creating atomic-energy plants Q all of them doomed by the dangers they
threaten and by their innate shortness of life, as compared to other
sources of power Q one can conclude only that we have an inborn and
unshakable streak of insanity.
Howard Fast is an author, playwright and screenwriter.
-----------------------------------------------------------------------------
Well, there it is. I'd like to write a rebuttal to this, but I don't see
how it would be possible to fit it in their 150 word letters to the editor
limit. :( So I'll probably just write what's necessary, send it in, and
hope they print it.
My copy of _The Health Hazards of not Going Nuclear_ seems to have been
lost in my recent move, so if anyone could give me some facts, figures,
or point by point rebuttals, they would be appreciated. Any suggestions
on the best way to go about this would be welcome as well. Many thanks in
advance for the help.
Ken
+-------------------------------+--------------------------------------------+
| Ken Kirksey | Sorry, currently out of cool .sig |
| kkir...@world.std.com | quotes. |
| The World, Public Access Unix | |
Richard A. Schumacher
standardization, not an argument against nuclear plants.
Steve Chapin
}}
}} Be sure to point out that this horror story is an argument for design
}} standardization, not an argument against nuclear plants.
I think the problem isn't with the design, but with the construction
(implementation vs. design). How do we ensure that construction is up
to spec without undue gov't intervention?
s...@cs.purdue.edu Steve Chapin Today's Grammar Lesson:
Aye, I like her eyes!
"Oh my God! I shot my eye out!" (Ralphie Parker, in "A Christmas Story")
Don T. Borowski
coal fired plant would probably make them die of fright. It is a
wonderment that any electricity is generated at all, by this sort of
logic.
If a faucet springs a leak in the restroom of a nuke plant, we hear about
it on the news. A boiler explosion at a coal fired plant scarcely rates
a mention.
Donald Borowski WA6OMI Hewlett-Packard, Spokane Division
"Angels are able to fly because they take themselves so lightly."
-G.K. Chesterton
Howard Smith
KBK>_Charlotte Observer_ (Charlotte, NC).
KBK>---------------------------------------------------------------------------
KBK>ACCIDENTS WAITING TO HAPPEN
KBK>* Nuclear power plants are a glowing example of American insanity
KBK>by Howard Fast
KBK>The Greenwich Time
KBK>When one thinks of the trillions of dollars that have been poured into
KBK>creating atomic-energy plants Q all of them doomed by the dangers they
KBK>threaten and by their innate shortness of life, as compared to other
KBK>sources of power Q one can conclude only that we have an inborn and
KBK>unshakable streak of insanity.
KBK>Howard Fast is an author, playwright and screenwriter.
KBK>---------------------------------------------------------------------------
KBK>Well, there it is. I'd like to write a rebuttal to this, but I don't see
KBK>how it would be possible to fit it in their 150 word letters to the editor
KBK>limit. :( So I'll probably just write what's necessary, send it in, and
KBK>hope they print it.
KBK>My copy of _The Health Hazards of not Going Nuclear_ seems to have been
KBK>lost in my recent move, so if anyone could give me some facts, figures,
KBK>or point by point rebuttals, they would be appreciated. Any suggestions
KBK>on the best way to go about this would be welcome as well. Many thanks in
KBK>advance for the help.
I for one see Mr. Murphy, like Dr. Malthus, leaning back in his chair,
smiling. They will not be denied.
---
. SLMR 2.1a . Nothing is so smiple that it can't get screwed up.
victor yodaiken
>If the steam section of a nuke plant makes someone nervous, then a
>coal fired plant would probably make them die of fright. It is a
>wonderment that any electricity is generated at all, by this sort of
>logic.
>
>If a faucet springs a leak in the restroom of a nuke plant, we hear about
>it on the news. A boiler explosion at a coal fired plant scarcely rates
>a mention.
Clearly, the worst case failure mode of a nuclear fission plant is
somewhat more spectacular than the worst case failure mode of a coal
plant. The inability of many nuclear advocates to understand this
difference does not damp the skepticism of non-fans.
--
Steinn Sigurdsson
Such failure of the imagination! Clearly you have not considered some
of the worst case failure modes of coal fired plants....
Did you know for example that coal fired plants usually keep huge
quantities of inflammable low level radioactives on site, and that
if a tornado were to strike and pulverise and disperse the materials,
not only would highly carcinogenic particles be spread all over the
place - in fact more than enough to give every human lung cancer
ten times over if distributed uniformly, but a passing jet might
ignite the particles while suspended in air and if the air mixture
happened to be just right the resulting explosion would be comparable
to a medium sized thermonuclear weapon! And that this is more probable
than some worst case scenarios for nuclear plants, especially given
the enormous numbers of large coal plants with coal storage facilities
incapable of withstanding a direct hit by a tornado!!!
* Steinn Sigurdsson Lick Observatory *
* ste...@lick.ucsc.edu "standard disclaimer" *
* Just because there's a reason *
* Doesn't mean it's understood Specials, 1979 *
Kevin Brown
I understand the difference. All I can say is: big deal.
What is your obsession with worst-case failure modes, Victor? You
*know* that worst-case failure modes aren't a good prime determining
factor in comparing alternatives. Rather, *average*-case failure
modes are.
What's the probability of the worst-case failure mode of a coal
plant, and what's the probability of the worst-case failure mode of
a nuclear fission plant?
Indeed, let's be analytical about this, shall we? Compute the sum of
P[i] * C[i], i = 1..n
for both a modern Canadian or American nuclear fission plant and a modern
Canadian or American coal plant, where i is the failure mode index, n is
the maximum failure mode, P[i] is the probability of failure mode i
occurring in a given amount of time, and C[i] is the cost (in whatever
terms you're considering) of failure mode i occurring.
The resulting sums will be the expected costs due to failure of each type
of plant.
Anyone got decent numbers to plug into the above?
--
Kevin Brown ke...@nuchat.sccsi.com
This is your .signature virus: < begin 644 .signature (9V]T8VAA(0K0z end >
This is your .signature virus on drugs: <>
Any questions?
victor yodaiken
> Clearly, the worst case failure mode of a nuclear fission plant is
> somewhat more spectacular than the worst case failure mode of a coal
> plant. The inability of many nuclear advocates to understand this
> difference does not damp the skepticism of non-fans.
>
>Such failure of the imagination! Clearly you have not considered some
>of the worst case failure modes of coal fired plants....
>
>Did you know for example that coal fired plants usually keep huge
>quantities of inflammable low level radioactives on site, and that
>if a tornado were to strike and pulverise and disperse the materials,
>not only would highly carcinogenic particles be spread all over the
>place - in fact more than enough to give every human lung cancer
>ten times over if distributed uniformly, but a passing jet might
>ignite the particles while suspended in air and if the air mixture
>happened to be just right the resulting explosion would be comparable
>to a medium sized thermonuclear weapon! And that this is more probable
>than some worst case scenarios for nuclear plants, especially given
>the enormous numbers of large coal plants with coal storage facilities
>incapable of withstanding a direct hit by a tornado!!!
So, your scenario is "more probable" than, say, a reactor core fire?
Got a reference for this claim?
--
victor yodaiken
>In article <53...@dime.cs.umass.edu> yoda...@chelm.cs.umass.edu (victor yodaiken) writes:
>>Clearly, the worst case failure mode of a nuclear fission plant is
>>somewhat more spectacular than the worst case failure mode of a coal
>>plant. The inability of many nuclear advocates to understand this
>>difference does not damp the skepticism of non-fans.
>
>I understand the difference. All I can say is: big deal.
>
>What is your obsession with worst-case failure modes, Victor? You
>*know* that worst-case failure modes aren't a good prime determining
>factor in comparing alternatives. Rather, *average*-case failure
>modes are.
They are? The simplicity of a mathematical model is not an infallible
indication of its correctness. Low probability/high consequence
events are not necessarily equivalent in riskiness to high probability
low consequence events.
>What's the probability of the worst-case failure mode of a coal
>plant, and what's the probability of the worst-case failure mode of
>a nuclear fission plant?
>
>Indeed, let's be analytical about this, shall we? Compute the sum of
>
> P[i] * C[i], i = 1..n
>
>for both a modern Canadian or American nuclear fission plant and a modern
>Canadian or American coal plant, where i is the failure mode index, n is
>the maximum failure mode, P[i] is the probability of failure mode i
>occurring in a given amount of time, and C[i] is the cost (in whatever
>terms you're considering) of failure mode i occurring.
>
>The resulting sums will be the expected costs due to failure of each type
>of plant.
That's very nice. But, you have to do two things before you can use
this little model. First, you have to show that the linear equation
actually corresponds to our intuitive concept of risk. Then you
have to show that there are some uncontroversial numbers to crunch.
But, as I've pointed out previously, neither of these two conditions is
true.
>Anyone got decent numbers to plug into the above?
>
Of course not. There is not even an exhaustive list of possible failure modes
of PWR/BWR plants.
--
Snide Pucker
#>[...] worst-case failure modes aren't a good prime determining
#>factor in comparing alternatives. Rather, *average*-case failure
#>modes are.
More or less, this is true. Unfortunately, the public isn't quite
as logical about it. Cases involving the word "radioactive" have
a multiplier of 5 (or 10 or 50...).
#>What's the probability of the worst-case failure mode of a coal
#>plant, and what's the probability of the worst-case failure mode of
#>a nuclear fission plant?
#> P[i] * C[i], i = 1..n
#>P[i] is the probability of failure mode i
#>occurring in a given amount of time, and C[i] is the cost (in whatever
#>terms you're considering) of failure mode i occurring.
#>
#>The resulting sums will be the expected costs due to failure of each type
#>of plant.
#>
#>Anyone got decent numbers to plug into the above?
^^^^^^
Good luck.
The whole "problem" with nuclear reactors is that we haven't been
fortunate enough to have a few fail miserably to test the estimated
(or stab-in-the-dark) probabilities. Does a failure probability of
10e-7/year mean once every 10e7 reactor-years or never? Who knows?
If you do, the Nuclear Regulatory Commission (NRC) has a job waiting
for you.
Statistical probability relies on (relatively) large sample
populations. When you don't have them, you are forced to rely on
combining individual failure probabilities to generate a sum
failure mode probability for a particular scenario. To say that
the fault trees become intricate and elaborate is to say that the
sun might be hot on its surface. This is more a case for
standardization and reduction of components than a case for banning
nuclear power plants, by the way.
Oh, FYI, try to stay away from assigning a "cost" to the possibility
of potential human casualties (yes, people die on occasion). People
get a bit gun-shy when numbers are assigned to human "costs." (for
good reason)
---
"Unfortunately, I belong to that satan-worshipping breed of neofascists
who have the gall to support the obviously idiotic, environmentally-
disastrous notion that nuclear power can be used in constructive
ways without converting the planet into an uninhabitable wasteland"
--me
--
---
Jerry W. Miller # "Opinions? Of course they
mil...@gn.ecn.purdue.edu # are mine, do YOU want them?"
Purdue Univ. School of Nuclear Engg//Argonne Nat'l Lab--Reactor Engg
Ken Basye
Victor Yodaiken writes:
>ke...@nuchat.sccsi.com writes:
>> Kevin's linear model of risk deleted
>That's very nice. But, you have to do two things before you can use
>this little model. First, you have to show that the linear equation
>actually corresponds to our intuitive concept of risk. Then you
>have to show that there are some uncontroversial numbers to crunch.
>But, as I've pointed out previously, neither of these two conditions is
>true.
>
>>Anyone got decent numbers to plug into the above?
>
>Of course not. There is not even an exhaustive list of possible failure modes
>of PWR/BWR plants.
>--
>yoda...@chelm.cs.umass.edu
Two comments:
1) The phrase "our intuitive concept of risk" presupposes the
existence of a common conception of risk, which I find (intuitively) a
risky presupposition. If you want to demand that a model correspond
to something, it strikes me as only fair that you show that there
really is something to correspond to, perhaps even say what it is, and
make some argument about why such a correspondence is necessary,
useful or beneficial. A complete version of the latter would also
include an admission of the drawbacks of requiring such a
correspondence, which is this case are significant.
2) In the context of this thread, the restriction of your last claim
to PWR/BWR plants is either an assertion that an exhaustive list of
failure modes for coal plants actually exists, or sophistry. And if
there is such a list, how do we know it corresponds to our intuitive
list of coal plant failure modes? :->
Internet/CSnet k...@cs.brown.edu U.S. MAIL Ken Basye
UUCP uunet!brunix!kjb Box 1910
Dept. of Computer Science
Brown University
Providence, RI 02912
Snide Pucker
#>So, your scenario is "more probable" than, say, a reactor core fire?
#>Got a reference for this claim?
Let's say that reactor core fires are so improbable that utilities
feed hydrogen gas into a core to reduce oxidation and _never_ worry
about a fire of any type (hydrogen being as reactive as it is).
In addition, there is considerably more concern over what the
hydrogen outside of the core will do as opposed to zero concern
(zero being defined as none, zip, nada) about any hydrogen in
the core. I'd say that that pretty much rules out an in-core
fire. In all my research in the matter (disclaimer: I guess I
could still be called a "spring chicken" =), I have yet to run
across this as a possibility.
I guess more importantly, could you ever start a fire in a core?
For a change, I'd have to say absolutely not. The fuel melts first,
then ruptures the vessel, _then_ builds enough heat to combust _if_
you can maintain its criticality (read as: no you can't).
Why not? Moderator, again. Water boils, beryllium melts. Now,
_carbon_ (a.k.a. graphite) burns (cf. Chernobyl). Since most U.S.
reactors don't use graphite (none on the grid, at least), an
ex-core, much less an in-core, _fuel_ fire cannot occur (you can
always burn steel and aluminum and wood ...).
Besides, fuel fires aren't the self-perpetuators that graphite is,
not to say that the fuel itself isn't producing (decay) heat.
Steinn Sigurdsson
What reactor type are you talking about?
For LWRs and PWRs I'd say the above scenario is comparable
to a core fire - certainly as probable as a "China Syndrome"
which you so adroitly pointed out has not been disproven... (sic).
But, in any case we are after all comparing worst case scenarios are
we not, and with my limited imagination it is clear that there are
failure modes for coal plants that neither I nor anyone else has
allowed for - and obviously the plant is not designed to cope with...
hence we must conclude that the coal plants are an intolerable risk,
or do you contest the basic physics that coal dust is a low level
radioactive, is carcinogenic and generally toxic, and that the energy
involved is easily enough to devastate a city if released in an
appropriate uncontrolled manner? Do you have any evidence that the
unforeseen failure modes of coal plants are "better" than those of
nuclear plants? Certainly the _known_ failure modes kill, sometimes
large groups of people, and at all stages of production, including
transport and waste storage - can you say the same of nuclear plants?
* Steinn Sigurdsson Lick Observatory *
* ste...@lick.ucsc.edu "standard disclaimer" *
* Some people think they're really clever *
* Smash your head against the wall Specials, 1979 *
victor yodaiken
>In article <1992Sep23....@nuchat.sccsi.com> ke...@nuchat.sccsi.com (Kevin Brown) writes:
>#>
>#>[...] worst-case failure modes aren't a good prime determining
>#>factor in comparing alternatives. Rather, *average*-case failure
>#>modes are.
>
> More or less, this is true. Unfortunately, the public isn't quite
> as logical about it. Cases involving the word "radioactive" have
> a multiplier of 5 (or 10 or 50...).
What's so logical about an average case analysis?
--------start quote
The criteria recommended in this article have no fundamental
basis. Indeed, there is no fundamental approach to this issue
and no way of proving whether any proposed criteria are right or
wrong except by using them over a period of time and discovering
whether the costs, risk, and other consequences of their use meet the
requirements of society (185)
SOURCE: D.J. Higson Nuclear Safety Assessment Criteria
Nuclear Safety 31-32 April-June 1990 193-185
> Statistical probability relies on (relatively) large sample
> populations. When you don't have them, you are forced to rely on
> combining individual failure probabilities to generate a sum
> failure mode probability for a particular scenario. To say that
> the fault trees become intricate and elaborate is to say that the
> sun might be hot on its surface. This is more a case for
> standardization and reduction of components than a case for banning
> nuclear power plants, by the way.
Sure. And how about for spending the money required to do some extensive
experiments? The recent fracas over measuring water levels in BWR reactors
was a good example of the nuclear regulatory/engineering mess.
--
Steinn Sigurdsson
What's so logical about an average case analysis?
--------start quote
The criteria recommended in this article have no fundamental
basis. Indeed, there is no fundamental approach to this issue
and no way of proving whether any proposed criteria are right or
wrong except by using them over a period of time and discovering
whether the costs, risk, and other consequences of their use meet the
requirements of society (185)
SOURCE: D.J. Higson Nuclear Safety Assessment Criteria
Nuclear Safety 31-32 April-June 1990 193-185
So you advocate building lots of nuclear reactors and operating them
under varying conditions for long periods until we have enough
failure samples to do a rigorous estimate of all failure modes
and associated risk?
victor yodaiken
>In article <53...@dime.cs.umass.edu> yoda...@chelm.cs.umass.edu (victor yodaiken) writes:
>#>>In article <1992Sep23....@nuchat.sccsi.com> ke...@nuchat.sccsi.com (Kevin Brown) writes:
>#>>#>
>#>>#>[...] worst-case failure modes aren't a good prime determining
>#>>#>modes are.
>#>>
>#>> More or less, this is true. Unfortunately, the public isn't quite
>#>> as logical about it. Cases involving the word "radioactive" have
>#>> a multiplier of 5 (or 10 or 50...).
>#>
>#>What's so logical about an average case analysis?
>
> I would consider likely failure modes to be a better comparator than
> worst-case failure modes (which tend to be highly improbable). Would
> you be horrified of living downstream of a hydro dam because it might
> fail or would you rather breathe the fumes of a coal plant and the
> occasional coal fire?
Nice choice. Depends on the dam and the coal plant, does it not?
And it depends on a lot of imponderables and ethical choices. Do
I believe that the dam is being maintained, do I value clear skies
more than trout fishing, do I believe that new technologies will
make the coal plant safer, do I believe that averting the possibility
of a total wipeout from the dam is worth the higher chance of late life
lung disease from the coal plant plus a low chance of an earlier coal
induced problem .... So, returning to nuclear, it is by no means
more "scientific" or logical to use average case or expected deaths
rather than an exponential scale which gives more weight to larger
scale events. A Chernobyl class accident is not identical to 5000
automobile accidents, and if your mathematical model of accidents obscures
this difference, the difference remains. For me, the possibility
of a major containment breach at Indian Point is an enormous risk, one
which is not worth taking. And if you can show that the expected deaths
from Indian Point are far fewer than the expected deaths from Con Eds
poorly maintained diesel generators, it will not reconcile me to the
risk (even if I believed the probability estimates and believed it
was an either/or choice). It's perfectly legit to argue that
the average case estimate bolsters the nuclear case, or that even that
average case estimates should be used over other costings, but it
is another thing entirely to claim that average case represents THE
"analytical " approach or that it offers definitive answers.
>#> The criteria recommended in this article have no fundamental
>#> basis. Indeed, there is no fundamental approach to this issue
>#> and no way of proving whether any proposed criteria are right or
>#> wrong except by using them over a period of time and discovering
>#> whether the costs, risk, and other consequences of their use meet the
>#> requirements of society (185)
>#>SOURCE: D.J. Higson Nuclear Safety Assessment Criteria
>#>Nuclear Safety 31-32 April-June 1990 193-185
>
> I'd have to question just what this article is trying to do (well,
> we really don't know anything so just give it up?). If dollars or lives is
> the expected result, the criteria (for evaluating consequences) is
> obvious.
I have to disagree. Suppose that Indian Point had to scram and
the reactor vessel cracked open because ductility had been lost
faster than estimated and the old containment structure
collapsed. Even if Jeremy Whitlock is correct and the core just
safely buried itself in the bedrock rather than sliding into the
Hudson it is imposible to figure out the consequences of such
an incident in any reliable manner. We can make educated guesses:
maybe the stock market would go up in survival euphoria, maybe
there would be a blackout and a mega riot spiced by fear of what
the core was really going to do, maybe evey other nuke near a US city
would be forced to shut due to panic, maybe the city water system
would have to be shut down as Cesium leaked into the Hudson ....
What Higson is arguing, as far as I can tell, is that we should not
decieve ourselves about what the numbers in a risk assessment mean. They
are possibly helpful indicators, but they are not definitive measures
even if they are correctly calculated.
> The criteria for deterimining probability of various failure
> modes is also fairly sound; it's the _determination_ of those
> probabilities that is lacking. What is really
Just the massive human factors element involved makes the probability
estimates suspect. I have zero confidence in the ability of the NRC
and the nuclear industry to adopt the very high engineering standards
required in fission power plants. What's the probability of an
accident in a plant where the batteries to start the backup control
system power generators are not even inspected for a year or two?
--
Steinn Sigurdsson
In article <1992Sep23....@cs.brown.edu> k...@cs.brown.edu (Ken Basye) writes:
>Victor Yodaiken writes:
>>ke...@nuchat.sccsi.com writes:
>>> Kevin's linear model of risk deleted
>>That's very nice. But, you have to do two things before you can use
>>this little model. First, you have to show that the linear equation
>>actually corresponds to our intuitive concept of risk. Then you
>>have to show that there are some uncontroversial numbers to crunch.
>>But, as I've pointed out previously, neither of these two conditions is
>>true.
>>>Anyone got decent numbers to plug into the above?
>>Of course not. There is not even an exhaustive list of possible failure modes
>>of PWR/BWR plants.
>Two comments:
>1) The phrase "our intuitive concept of risk" presupposes the
>existence of a common conception of risk, which I find (intuitively) a
>risky presupposition. If you want to demand that a model correspond
>to something, it strikes me as only fair that you show that there
>really is something to correspond to, perhaps even say what it is, and
>make some argument about why such a correspondence is necessary,
>useful or beneficial. A complete version of the latter would also
>include an admission of the drawbacks of requiring such a
>correspondence, which is this case are significant.
You have placed the burden of proof on the wrong party. Kevin advances
??? The formula defines risk, if you do not agree that this
corresponds with the intuitive notion of risk - and it is not clear
that it does - then you must propose a better estimator.
a formula which he claims will help to provide an analytical basis for
determining relative risks of power production methods. It then seems
incumbent on Kevin to show that his formula corresponds to what is
generally considered to be "risk". If there is no informed consensus on what
constitutes "risk" then his claim is necessarily false: he has a formula
but it does not measure what he says it does.
>2) In the context of this thread, the restriction of your last claim
>to PWR/BWR plants is either an assertion that an exhaustive list of
>failure modes for coal plants actually exists, or sophistry. And if
>there is such a list, how do we know it corresponds to our intuitive
>list of coal plant failure modes? :->
Again, the sense of your logic eludes me. Kevin advances a formula
which involves summing the possible failure modes of PWR/BWR plants.
I argue that this summation is not calculable because the number
of terms is unknown. If other aspects of the formula are also
not calculable it only strengthens my objections.
Tut, tut, there is an assumption that the missing terms have P(i)
small enough that they contribute negligibly to the sum, ie it is
assumed that the sum of the remaining terms is small compared to
the finite sum used - it is possible to disprove this by demonstrating
that some of the missing terms are large enough.
That a part of the formula is missing _weakens_ your claim if you
are doing comparative risks, which IMHO are the only ones that make
sense, thus the fact that _no_ terms are presented for coal plants
does not imply the risk is zero, it may be large and you must show
otherwise - eg by estimating what you consider the largest term to be...
* Steinn Sigurdsson Lick Observatory *
* ste...@lick.ucsc.edu "standard disclaimer" *
Steinn Sigurdsson
Sorry for my imprecision, but I had in mind any type of fire which
would disperse core material not just core _fuel_ fires.
Please remember the original claim by Siugurdsson was quite strong:
This is the bit I wrote:
>Did you know for example that coal fired plants usually keep huge
>quantities of inflammable low level radioactives on site, and that
>if a tornado were to strike and pulverise and disperse the materials,
>not only would highly carcinogenic particles be spread all over the
>place - in fact more than enough to give every human lung cancer
>ten times over if distributed uniformly, but a passing jet might
>ignite the particles while suspended in air and if the air mixture
>happened to be just right the resulting explosion would be comparable
>to a medium sized thermonuclear weapon! And that this is more probable
>than some worst case scenarios for nuclear plants, especially given
>the enormous numbers of large coal plants with coal storage facilities
>incapable of withstanding a direct hit by a tornado!!!
Please note complete absence of any reference to core fires,
I was specifically vague as to what constituted "some worst
case scenarios" because I wrote the estimate in Greenpeace-ese,
eg the estimation of cancer risk due to well enough dispersed
coal dust more commonly seen in the context of plutonium.
The FAE scenario was a little dig at a post from last month
and the energy estimate is valid - now think about why this
is dismissed as a ridicilous risk estimate yet some nuclear
accident scenarios are discussed with grave concern...
* Steinn Sigurdsson Lick Observatory *
* ste...@lick.ucsc.edu "standard disclaimer" *
victor yodaiken
>predictions to real-world events.
>
>So: how many worst-case failures, of the kind you refer to, have happened
>in the history of the world to reactors of the kind used in the U.S. and
>Canada? I dare say none have.
But there have been several severe failure modes which were not forseen.
True, none of these failures has resulted in a major off site release
but the fact that's a pretty slim reed to hold onto. The sample space
is small, extrapolations are dubious.
>Since more average-case failures have occurred than worst-case failures,
>it seems ludicrous to suggest that a worst-case analysis is a better
>means of comparison than an average-case analysis.
>
Lost me here. By analogy, do you believe that because the brakes have
never failed in my car, yet the headlights burnt out twice, I should
ignore the possibility of brake failure? We pay more attention to
accidents with terrible consequences than to accidents with
minor consequences.
>And so I reiterate my question: what is this obsession you have with
>worst-case failure modes?
>
Well, I find the worst-case failure consequences of nuclear power to
be worrisome.
>>Low probability/high consequence
>>events are not necessarily equivalent in riskiness to high probability
>>low consequence events.
>
>By what measure is *this* true? I was under the impression that the
>very *definition* of statistical risk was probability times cost.
>
Well, your impression is incorrect. See for example:
--------start quote:
In this section, we introduce commonly used measures of it
[risk]. There is, of course, no definitive measure, as we indicate
later. An often used definition of risk for a particular set of events is
(summed over all events):
Total Average = Sum_{events} (frequency of occurrence)x magnitude of effect
...
This measure of risk is a simple measure; however it may not be
appropriate for analyses. For example, if the probability of harm is 10^-7
and the magnitude is 10^7 persons killed, then the expected value of
risk, E(R), is:
E(R) = 10^-7 * 10^7
and the expected value is unity. Using the expected value in this case
masks the fact that 10^7 people are killed, admittedly with very low
probability.
-----end quote
SOURCE Brookhaven/EPRI workshop on "Health and Environmental Risk Assessment"
(Pergamon Press, 1985). Introduction by Paolo F. Ricci.
and see:
--------start quote
The formulation of societal risk as an expectation value runs into
difficulties when the probability of the event is low, but the
consequence is high if it occurs. In this case, there would be no
consequence or a very large consequence. Therefore the use of
expectation value does not adequately reflect the real societal risk
because the numerical value does not reflect a consequence that would
actually occur.
A second point is that other aspects of societal risk which are not
relevant in the case of controlled exposures to radiation, may become
important for accidents,particularly the following:
The possibility for individual doses above the threshold for
acute health problems.
Economic losses.
Social disruption.
Risk aversion for higher consequence accidents.
These aspects may be represented by quantities that are not
linearly related to dose or to the release of radioctivity
SOURCE: D.J. Higson Nuclear Safety Assessment Criteria
Nuclear Safety 31-32 April-June 1990 193-185
>>That's very nice. But, you have to do two things before you can use
>>this little model. First, you have to show that the linear equation
>>actually corresponds to our intuitive concept of risk.
>
>In terms of demonstrating objective correctness, correspondence to human
>intuition of risk is irrelevent. All that matters is that the predictions
>match real-world observation.
You claim that n deaths distributed over 10 years are equivalent to n
deaths over 3 days embedded in 10 death free years.
I disagree. Who's being objective?
>>Then you
>>have to show that there are some uncontroversial numbers to crunch.
>
>This is the most difficult bit. Except for failure cases that have actually
>happened, the numbers will come from theoretical considerations, and
>unfortunately there is some disagreement over which theories are most
>correct.
no kidding.
>>Of course not. There is not even an exhaustive list of possible failure modes
>>of PWR/BWR plants.
>
>No doubt this is also true of coal plants. If so, then upon what is the
>notion that nuclear plants (of the U.S./Canadian variety) are more dangerous
>than coal plants based on?
There is no analytical model which will allow us to make an unbiased and
objective judgement.
--
victor yodaiken
You have placed the burden of proof on the wrong party. Kevin advances
a formula which he claims will help to provide an analytical basis for
determining relative risks of power production methods. It then seems
incumbent on Kevin to show that his formula corresponds to what is
generally considered to be "risk". If there is no informed consensus on what
constitutes "risk" then his claim is necessarily false: he has a formula
but it does not measure what he says it does.
>
>2) In the context of this thread, the restriction of your last claim
>to PWR/BWR plants is either an assertion that an exhaustive list of
>failure modes for coal plants actually exists, or sophistry. And if
>there is such a list, how do we know it corresponds to our intuitive
>list of coal plant failure modes? :->
Again, the sense of your logic eludes me. Kevin advances a formula
which involves summing the possible failure modes of PWR/BWR plants.
I argue that this summation is not calculable because the number
of terms is unknown. If other aspects of the formula are also
not calculable it only strengthens my objections.
--
victor yodaiken
>#>
>#>So, your scenario is "more probable" than, say, a reactor core fire?
>#>Got a reference for this claim?
>
> Let's say that reactor core fires are so improbable that utilities
> feed hydrogen gas into a core to reduce oxidation and _never_ worry
> about a fire of any type (hydrogen being as reactive as it is).
Note that the claim was that a tornado creating a coal dust cloud
which would then detonate was "more probable" than a catastrophic
reactor accident. In view of Mr. Sigurdsson's oft stated concern for
scientific standrds, I merely wanted to elicit the, no doubt sound,
factual basis of his assertion.
> I guess more importantly, could you ever start a fire in a core?
> For a change, I'd have to say absolutely not. The fuel melts first,
> then ruptures the vessel, _then_ builds enough heat to combust _if_
> you can maintain its criticality (read as: no you can't).
> Why not? Moderator, again. Water boils, beryllium melts. Now,
> _carbon_ (a.k.a. graphite) burns (cf. Chernobyl). Since most U.S.
> reactors don't use graphite (none on the grid, at least), an
> ex-core, much less an in-core, _fuel_ fire cannot occur (you can
> always burn steel and aluminum and wood ...).
Sorry for my imprecision, but I had in mind any type of fire which
would disperse core material not just core _fuel_ fires.
Please remember the original claim by Siugurdsson was quite strong:
>Did you know for example that coal fired plants usually keep huge
>quantities of inflammable low level radioactives on site, and that
>if a tornado were to strike and pulverise and disperse the materials,
>not only would highly carcinogenic particles be spread all over the
>place - in fact more than enough to give every human lung cancer
>ten times over if distributed uniformly, but a passing jet might
>ignite the particles while suspended in air and if the air mixture
>happened to be just right the resulting explosion would be comparable
>to a medium sized thermonuclear weapon! And that this is more probable
>than some worst case scenarios for nuclear plants, especially given
>the enormous numbers of large coal plants with coal storage facilities
>incapable of withstanding a direct hit by a tornado!!!
--
Snide Pucker
#>>In article <1992Sep23....@nuchat.sccsi.com> ke...@nuchat.sccsi.com (Kevin Brown) writes:
#>>#>
#>>#>[...] worst-case failure modes aren't a good prime determining
#>>#>modes are.
#>>
#>> More or less, this is true. Unfortunately, the public isn't quite
#>> as logical about it. Cases involving the word "radioactive" have
#>> a multiplier of 5 (or 10 or 50...).
#>
#>What's so logical about an average case analysis?
I would consider likely failure modes to be a better comparator than
worst-case failure modes (which tend to be highly improbable). Would
you be horrified of living downstream of a hydro dam because it might
fail or would you rather breathe the fumes of a coal plant and the
occasional coal fire?
The "problem" with worst-case scenarios is that they tend to be glorified
by the media thus heightening their perceived risks by the public. The
public is (whether justified or not) horrified of the announced presence
of radioactivity (else they would be continually horrified =). This
does not increase the _risks_ of radioactivity, but it does increase
the public's attention to it. I consider this human, but illogical.
#> The criteria recommended in this article have no fundamental
#> basis. Indeed, there is no fundamental approach to this issue
#> and no way of proving whether any proposed criteria are right or
#> wrong except by using them over a period of time and discovering
#> whether the costs, risk, and other consequences of their use meet the
#> requirements of society (185)
#>SOURCE: D.J. Higson Nuclear Safety Assessment Criteria
#>Nuclear Safety 31-32 April-June 1990 193-185
I'd have to question just what this article is trying to do (well,
we really don't know anything so just give it up?). If dollars or lives is
the expected result, the criteria (for evaluating consequences) is
obvious. The criteria for deterimining probability of various failure
modes is also fairly sound; it's the _determination_ of those
probabilities that is lacking. What is really
missing is the public's acceptance of said measure and the reliable
inclusion of a public perception into the model. I don't think anyone
would argue that the latter is never going to happen accurately.
#>Sure. And how about for spending the money required to do some extensive
#>experiments? The recent fracas over measuring water levels in BWR reactors
#>was a good example of the nuclear regulatory/engineering mess.
We are (at least till the budget gets hacked again). Worst case scenarios
are one of the few active area of LWR research, and by worst case, I
mean the high-pressure melt ejection, direct containment heating kind
of improbabilities. There's big bucks going into figuring out whether
or not the containment will be safe. Core migration (via melting) is
the stuff for computer codes (that means we understand most of it) now.
The BWR fracas is too applied to warrant research money. That's in
the hands of the utilities/vendors/etc., which doesn't make it any
less important.
victor yodaiken
By your "reasoning" the risks involved in manufacturing
hydrogen bombs are equivalent to those involved in manufacturing
ice cream as one never knows when a tornado might send lethal fragments
of ice cream flying through the air at high speeds.
--
Steinn Sigurdsson
It's your "reasoning" Victor, I'm only using the same "logic"
for coal plants as is used against nuclear plants.
Btw you're wrong about the ice cream, unless you know
of a carcinogen in ice cream the rest of us don't,
- the physics is quite different. Even then I'm sure
people have died in the production of ice cream, I know
people who died in dairy production, and cheese is postively
lethal - far more frenchmen die from cheese industry than
the nuclear industry for example, every year.
Tino
>>worst-case failure modes?
>>
>
>Well, I find the worst-case failure consequences of nuclear power to
>be worrisome.
Geez, do you stay up all night trembling over the consequences of a large
meteor hitting the earth? That's a pretty big worst-case failure mode.
This is getting pretty tiresome.
Tino
--
---------------------------------------------------------------------------
"The nuclear scientist will prepare the bed on which mankind must lie; and
if mankind doesn't fit - well, that will be just too bad for mankind."
- Aldous Huxley
victor yodaiken
>
>
> What's so logical about an average case analysis?
>
> --------start quote
> The criteria recommended in this article have no fundamental
> basis. Indeed, there is no fundamental approach to this issue
> and no way of proving whether any proposed criteria are right or
> wrong except by using them over a period of time and discovering
> whether the costs, risk, and other consequences of their use meet the
> requirements of society (185)
> SOURCE: D.J. Higson Nuclear Safety Assessment Criteria
> Nuclear Safety 31-32 April-June 1990 193-185
>
>So you advocate building lots of nuclear reactors and operating them
>under varying conditions for long periods until we have enough
>failure samples to do a rigorous estimate of all failure modes
>and associated risk?
No. Please excuse me if I opt out of further elaboration of this issue with
you, as you seem to lack a committment to rational discussion.
--
Kevin Brown
>In article <1992Sep23....@nuchat.sccsi.com> ke...@nuchat.sccsi.com (Kevin Brown) writes:
>>In article <53...@dime.cs.umass.edu> yoda...@chelm.cs.umass.edu (victor yodaiken) writes:
>>>Clearly, the worst case failure mode of a nuclear fission plant is
>>>somewhat more spectacular than the worst case failure mode of a coal
>>>plant. The inability of many nuclear advocates to understand this
>>>difference does not damp the skepticism of non-fans.
>>
>>I understand the difference. All I can say is: big deal.
>>
>>What is your obsession with worst-case failure modes, Victor? You
>>*know* that worst-case failure modes aren't a good prime determining
>>factor in comparing alternatives. Rather, *average*-case failure
>>modes are.
I really should have said that average-case failure modes are a *better*
prime determining factor in comparing alternatives than worst-case failure
modes.
>They are? The simplicity of a mathematical model is not an infallible
>indication of its correctness.
This is true. So?
The only *really* good test of correctness of a model is comparison of its
predictions to real-world events.
So: how many worst-case failures, of the kind you refer to, have happened
in the history of the world to reactors of the kind used in the U.S. and
Canada? I dare say none have.
How many average-case failures have occurred? I dare say more than zero.
Since more average-case failures have occurred than worst-case failures,
it seems ludicrous to suggest that a worst-case analysis is a better
means of comparison than an average-case analysis.
And so I reiterate my question: what is this obsession you have with
worst-case failure modes?
>Low probability/high consequence
>events are not necessarily equivalent in riskiness to high probability
>low consequence events.
By what measure is *this* true? I was under the impression that the
very *definition* of statistical risk was probability times cost.
>>What's the probability of the worst-case failure mode of a coal
>>plant, and what's the probability of the worst-case failure mode of
>>a nuclear fission plant?
>>
>>Indeed, let's be analytical about this, shall we? Compute the sum of
>>
>> P[i] * C[i], i = 1..n
>>
>>for both a modern Canadian or American nuclear fission plant and a modern
>>Canadian or American coal plant, where i is the failure mode index, n is
>>the maximum failure mode, P[i] is the probability of failure mode i
>>occurring in a given amount of time, and C[i] is the cost (in whatever
>>terms you're considering) of failure mode i occurring.
>>
>>The resulting sums will be the expected costs due to failure of each type
>>of plant.
>
>That's very nice. But, you have to do two things before you can use
>this little model. First, you have to show that the linear equation
>actually corresponds to our intuitive concept of risk.
In terms of demonstrating objective correctness, correspondence to human
intuition of risk is irrelevent. All that matters is that the predictions
match real-world observation.
In terms of convincing people who are unable or unwilling to put aside
their intuition in favor of an objective model, I agree that one must
show the correspondence between the equation and human intuition.
>Then you
>have to show that there are some uncontroversial numbers to crunch.
This is the most difficult bit. Except for failure cases that have actually
happened, the numbers will come from theoretical considerations, and
unfortunately there is some disagreement over which theories are most
correct.
>But, as I've pointed out previously, neither of these two conditions is
>true.
>
>>Anyone got decent numbers to plug into the above?
>>
>
>Of course not. There is not even an exhaustive list of possible failure modes
>of PWR/BWR plants.
No doubt this is also true of coal plants. If so, then upon what is the
notion that nuclear plants (of the U.S./Canadian variety) are more dangerous
than coal plants based on?
Snide Pucker
#>would disperse core material not just core _fuel_ fires.
Same difference.
To disperse core material, the fire would have to be in the core
(unless you have some burning desire to combust concrete) else the
fire wouldn't make any difference. Fires also require large amounts
of oxygen, which is not very plentiful in reactor cores (even if
you hydrolize the steam). Should a fire start, it wouldn't be long-
lived enough to matter to anyone but the person picking up the tab
for cleaning up the mess.
#>Please remember the original claim by Siugurdsson was quite strong:
I do. I thought it was very creative, even illustrative, maybe,
but then I kinda get used to the But-what-if-all-the-air-molecules-
in-this-room-concentrate-in-my-hand-and-we-suffocate?-type scenarios. =)
#> >Did you know for example that coal fired plants usually keep huge
#> >quantities of inflammable low level radioactives on site, and that
#> >if a tornado were to strike and pulverise and disperse the materials,
#> >not only would highly carcinogenic particles be spread all over the
#> >place - in fact more than enough to give every human lung cancer
#> >ten times over if distributed uniformly, but a passing jet might
#> >ignite the particles while suspended in air and if the air mixture
#> >happened to be just right the resulting explosion would be comparable
#> >to a medium sized thermonuclear weapon! And that this is more probable
#> >than some worst case scenarios for nuclear plants, especially given
#> >the enormous numbers of large coal plants with coal storage facilities
#> >incapable of withstanding a direct hit by a tornado!!!
Since all this scenario requires is a passing (strong) wind, an open
flame, and a stoichiometric particle suspension, I'd have to say it's
quite a bit more probable than virtually all worst-case nuclear accident
scenarios. (okay, so the flame is a bit far-fetched ;)
victor yodaiken
>??? The formula defines risk, if you do not agree that this
>corresponds with the intuitive notion of risk - and it is not clear
>that it does - then you must propose a better estimator.
>
What a peculiar idea. K proposes that formula F measures quantity R
Y argues that F does not measure R because F assigns a low
value to certain R which quite arguably should have a high value.
S rebuts Y by pointing out that Y has not advanced an alternate formula.
Sorry, but mathematical models cannot be validated in this way.
> Again, the sense of your logic eludes me. Kevin advances a formula
> which involves summing the possible failure modes of PWR/BWR plants.
> I argue that this summation is not calculable because the number
> of terms is unknown. If other aspects of the formula are also
> not calculable it only strengthens my objections.
>
>Tut, tut, there is an assumption that the missing terms have P(i)
>small enough that they contribute negligibly to the sum, ie it is
>assumed that the sum of the remaining terms is small compared to
>the finite sum used - it is possible to disprove this by demonstrating
>that some of the missing terms are large enough.
K claims that formula F provides an analytic method for comparing N and
C. Y points out that F = sum x_1 ... x_n where n is not known. S rebuts
Y by announcing that there is an "assumption" that the missing terms
are not significant. Sorry, but an unverified and contentious "assumption"
does not an analytical method make.
>That a part of the formula is missing _weakens_ your claim if you
>are doing comparative risks, which IMHO are the only ones that make
>sense, thus the fact that _no_ terms are presented for coal plants
>does not imply the risk is zero, it may be large and you must show
>otherwise - eg by estimating what you consider the largest term to be...
Ridiculous. I deny that a definitive comparison can be made using these
tinny models. Please look at the risk assessment literature for an
indication as to the difficulty of making such comparisons and the
non-utility of facile use of oversimplified models. While
representing a complex poorly understood situation with high school
algebra may serve to disguise an opinion as a scientific judgement,
it serves little purpose otherwise.
--
Russ Brown
In principle, there is nothing wrong with developing an _estimate_ of
risk, particularly the relative risks of similar systems, by a summation
of probability-hazard sub-products. In practice, we can never know all
possible failure modes or the exact probabilities associated with them.
But nonetheless, these types of estimates would appear to be superior to
the classical (and still popular) method developed by C. Little, who
took a fundamentally conservative position on the descent of the upper
atmosphere and that which lies beyond.
Basing our social, political, and economic judgments on worst-case
assessments could place us in situations with demonstrably higher risks,
however fuzzy the accuracy of the estimates.
The thoughts of Samuel Butler are worth consideration,
"Though analogy is often misleading, it is the least misleading thing
we have."
Risk analysis is, and will remain, an uncertain science. But we live
with a vast array of uncertainties. This should not immobilize us.
Consider a choice of home protective devices:
What would you spend your money on, a meteor detector or a circuit
breaker for your electrical system?
This is a simple risk-based decision. Just to make it easier, assume
that the cost of implementation is the same. Few of us know the exact
risks of meteor strikes or electrical overloads, but we do know that a
meteor strike could be expected to have a somewhat higher hazard. But
how many of us would opt for the meteor detector on that basis?
In real life, the cost of protection against worst-case situations is
often orders of magnitude higher than protection against the quotidian
hazards of lesser impact. The practical impact of investing in
worst-case avoidance is that we may be getting very little
bang-avoidance for the buck. This doesn't make much sense from either
an environmental or an economic perspective.
-------------------------------------------------------------------------
"There is no worse lie than a truth misunderstood by those who hear it."
-William James-
E. Michael Smith
>They are? The simplicity of a mathematical model is not an infallible
>indication of its correctness. Low probability/high consequence
>events are not necessarily equivalent in riskiness to high probability
^^^^^^^^^^^^
>low consequence events.
Since the whole question turns on this point, I'd suggest sharpening
the point by saying 'in the human valuation placed on the riskiness'.
That correctly puts the focus on the values of the folks being exposed
to the risk and takes it off the engineering definition of calculated
abstract 'risk'.
The alternative is a long border war of the form "Engineers don't do
risk stuff right" vs. "Non-Nukers are innumerate flakes since they don't
accept our mathematical model of risk". The reality is more along the
lines that pro-nukers do have good models, but have some limits on
quality of input data, and that anti-nukers don't value the probable
outcomes in the same way that engineers do.
I think the engineers are quite right in how the calculate risk, I just
don't think their premises are fully proven ... I'll take a 10% risk of
loosing one member of my staff over a 1% risk of loosing them all (for
a staff of 10) where the product of #folks x probability of loss is
the same for both events. Why? The consequences hit a threshold that
is just not acceptable ...
The issue of 'can a real catastrophy on the scale of Chernobyl happen
in the US?' is a different question (I think it "can't", but that doesn't
really matter...). It does, though, illustrate an interesting symetry
of thresholding... If an event is implausable 'enough' for me, it
doesn't 'matter' any more. A .000000001 percent chance that my whole
group will be hit by a P3 Orion on final approach to Moffett is just not
a concern, since it drops below my 'threshold'.
So, IMHO, there are two competing threshold effects. First is the
threshold of consequences. If these are high enough, the VALUE placed
on a given probability of loss is raised. Second is likelyhood. If
this is remote enough, the consequences don't matter. (i.e. are VALUED
as low). The problem? It's a race condition ... If you don't trust
the low probability estimate or if it is 'not low enough' for you, then
the large consequences dominate. If you don't think the consequences
are really that big (i.e. no China Syndrome is really possible, only some
local contamination, so consequence is overstated.) then the low
probability dominates the thresholding and you 'write if off as a non-risk'.
I think this accurate describes the conflict between the two camps. The
Nukers think the consequences are overstated (or are more comfortable
with the idea of extra rads - no surprise ...) and the likelyhood numbers
are accurate and very low. The Anti-nukers think consequences are under
stated (or are less comfortabel with extra cancers in their whole region)
and don't trust the likelyhood numbers due to stories about poor
construction et.al. Unfortunately, rather than having a liniear
difference of n%, they have a non-linear difference ... (Yes, I think
both sides go non-linear ;-)
I hope this makes some sense ...
--
E. Michael Smith e...@apple.COM
'Whatever you can do, or dream you can, begin it. Boldness has
genius, power and magic in it.' - Goethe
I am not responsible nor is anyone else. Everything is disclaimed.
victor yodaiken
>In article <53...@dime.cs.umass.edu> yoda...@chelm.cs.umass.edu (victor yodaiken) writes:
>
> In article <STEINLY.92...@topaz.ucsc.edu> ste...@topaz.ucsc.edu (Steinn Sigurdsson) writes:
> >But, in any case we are after all comparing worst case scenarios are
> >we not, and with my limited imagination it is clear that there are
> >failure modes for coal plants that neither I nor anyone else has
> >allowed for - and obviously the plant is not designed to cope with...
> >hence we must conclude that the coal plants are an intolerable risk,
> >or do you contest the basic physics that coal dust is a low level
> >radioactive, is carcinogenic and generally toxic, and that the energy
> >involved is easily enough to devastate a city if released in an
> >appropriate uncontrolled manner? Do you have any evidence that the
> >unforeseen failure modes of coal plants are "better" than those of
> >nuclear plants? Certainly the _known_ failure modes kill, sometimes
> >large groups of people, and at all stages of production, including
> >transport and waste storage - can you say the same of nuclear plants?
>
> By your "reasoning" the risks involved in manufacturing
> hydrogen bombs are equivalent to those involved in manufacturing
> ice cream as one never knows when a tornado might send lethal fragments
> of ice cream flying through the air at high speeds.
>
>It's your "reasoning" Victor, I'm only using the same "logic"
>for coal plants as is used against nuclear plants.
No. I'm afraid that you've confused yourself. I argued that worst case
nuclear accidents were important considerations. You argued that as a
known improbability could turn a coal plant into an instrument of mass
destruction, worst case incidents were dismissable. I simply noted that
your argument could be used to "show" that production of atom bombs was
not more dangerous than production of ice cream.
--
Michael Zika
In article <1992Sep24....@michael.apple.com> you write:
>In article <53...@dime.cs.umass.edu> yoda...@chelm.cs.umass.edu (victor yodaiken) writes:
>
>>They are? The simplicity of a mathematical model is not an infallible
>>indication of its correctness. Low probability/high consequence
>>events are not necessarily equivalent in riskiness to high probability
> ^^^^^^^^^^^^
>>low consequence events.
>
>Since the whole question turns on this point, I'd suggest sharpening
>the point by saying 'in the human valuation placed on the riskiness'.
>
>That correctly puts the focus on the values of the folks being exposed
>to the risk and takes it off the engineering definition of calculated
>abstract 'risk'.
At the risk of beating a dead horse (kick, kick, kick)... In mathematical
terms this seems to take the form of a delta function. Where:
R = P(i)C(i) {1 + alpha*delta(i) }
In this model the delta function takes into account the indiviuals's
perception/acceptance of event "i" (where alpha is some large number that
Significantly increases cost). However, the delta function is "tripped"
at a different threshold for every individual. So, _my_ delta function
for a severe core melt is zero, whereas Victor's is one...
Hmm... Using the same model, we come to very different estimates
of the _Risk_ involved...
> I hope this makes some sense...
ditto
>
>E. Michael Smith e...@apple.COM
--
|
Michael Zika (zi...@ecn.purdue.edu) | Hey don't ask me "why?",
Purdue University | I'm still working on "how?" !
School of Nuclear Engineering |
Steinn Sigurdsson
In article <STEINLY.92...@topaz.ucsc.edu> ste...@topaz.ucsc.edu (Steinn Sigurdsson) writes:
>In article <53...@dime.cs.umass.edu> yoda...@chelm.cs.umass.edu (victor yodaiken) writes:
>??? The formula defines risk, if you do not agree that this
>corresponds with the intuitive notion of risk - and it is not clear
>that it does - then you must propose a better estimator.
>
What a peculiar idea. K proposes that formula F measures quantity R
Y argues that F does not measure R because F assigns a low
value to certain R which quite arguably should have a high value.
S rebuts Y by pointing out that Y has not advanced an alternate formula.
Sorry, but mathematical models cannot be validated in this way.
Victor, _why_ should the certain R be assigned higher values,
if you answer that you are proposing an alternative model
of the risk assessment, if it is to be useful it will be
quantifiable, ergo you submit an alternative model.
How precisely do you validate numerical models without
showing the numbers come out wrong - other than showing
inconsistencies of course - and how do you demonstrate the
number is incorrect without proposing an alternative (range) of
values?
> Again, the sense of your logic eludes me. Kevin advances a formula
> which involves summing the possible failure modes of PWR/BWR plants.
> I argue that this summation is not calculable because the number
> of terms is unknown. If other aspects of the formula are also
> not calculable it only strengthens my objections.
>
>Tut, tut, there is an assumption that the missing terms have P(i)
>small enough that they contribute negligibly to the sum, ie it is
>assumed that the sum of the remaining terms is small compared to
>the finite sum used - it is possible to disprove this by demonstrating
>that some of the missing terms are large enough.
K claims that formula F provides an analytic method for comparing N and
C. Y points out that F = sum x_1 ... x_n where n is not known. S rebuts
Y by announcing that there is an "assumption" that the missing terms
are not significant. Sorry, but an unverified and contentious "assumption"
does not an analytical method make.
They most certainly do, show me any quantified system of analysis that
does not make unverified assumptions. At least with the assumptions
explicit they can compared with reality.
Do you seriosuly suggest that numerical analysis using finite
term sums is inherently improper? Or do you require analytic proof
that the discarded terms are small? In which case you can never model
any aspect of physical reality.
>That a part of the formula is missing _weakens_ your claim if you
>are doing comparative risks, which IMHO are the only ones that make
>sense, thus the fact that _no_ terms are presented for coal plants
>does not imply the risk is zero, it may be large and you must show
>otherwise - eg by estimating what you consider the largest term to be...
Ridiculous. I deny that a definitive comparison can be made using these
tinny models. Please look at the risk assessment literature for an
indication as to the difficulty of making such comparisons and the
non-utility of facile use of oversimplified models. While
representing a complex poorly understood situation with high school
algebra may serve to disguise an opinion as a scientific judgement,
it serves little purpose otherwise.
Victor, we're dealing with reality here, there are no definitive
comparisons, if you have a better way please contribute it to the
literature, or at least make constructive suggestions as to what is
wrong with the approach suggested, you cannot both demand a risk
assessment and deny that any risk assessment is possible.
* Steinn Sigurdsson Lick Observatory *
* ste...@lick.ucsc.edu "standard disclaimer" *
victor yodaiken
>In article <53...@dime.cs.umass.edu> yoda...@chelm.cs.umass.edu (victor yodaiken) writes:
>>
>>>And so I reiterate my question: what is this obsession you have with
>>>worst-case failure modes?
>>>
>>
>>Well, I find the worst-case failure consequences of nuclear power to
>>be worrisome.
>
>Geez, do you stay up all night trembling over the consequences of a large
>meteor hitting the earth? That's a pretty big worst-case failure mode.
Are we currently considering construction of meteors?
>This is getting pretty tiresome.
I think it's interesting that the idea that "risk" *is* average case
or expected deaths is so entrenched. The error I wish to expose
is the error of believing that one's subjective, emotional, ethical
estimation of the relative costs of a system failure are somehow
transformed into objective, scientific, and unemotional pure facts, when
generated by a calculation. My ethical viewpoint is such that I find the
mere possibility of, say, dumping Indian Point's core into the Hudson
to be a strong argument against nuclear power. Your ethical viewpoint may
be that the risks of this possibility are outweighed by the benefits.
That's fine. But please don't argue that your ethics is somehow "objective"
while mine is less so.
--
Snide Pucker
#>>
#>> I would consider likely failure modes to be a better comparator than
#>> worst-case failure modes (which tend to be highly improbable). Would
#>> you be horrified of living downstream of a hydro dam because it might
#>> fail or would you rather breathe the fumes of a coal plant and the
#>> occasional coal fire?
#>
#>Nice choice. Depends on the dam and the coal plant, does it not?
#>And it depends on a lot of imponderables and ethical choices.
Correct. The question was a sham because it involved an infinite
number of variances in just the utilities operating the plants.
It brings up a very valid point about the individual values each
person places on an infinite number of variables. No model could
possibly hope to encompass every person's perceptions about any
consequence (I'll even disregard strange phobias, schizophrenics,
emotionally traumatized, etc.).
#>So, returning to nuclear, it is by no means
#>more "scientific" or logical to use average case or expected deaths
#>rather than an exponential scale which gives more weight to larger
#>scale events.
This is true. If you haven't already, do read E. Michael Smith's
article in this thread on the incorporation of a perceptual component
to risk analysis. Since, as he writes and I concur, individual
tolerance limits will vary (and are discontinuous), a highly non-
linear, discontinuous scale must necessarily be used. Naturally,
this would require an inordinate number of reference points that,
should the nuclear industry ever produce those points, the whole
discussion over nuclear power will have been effectively nullified.
#>For me, the possibility
#>of a major containment breach at Indian Point is an enormous risk, one
#>which is not worth taking. And if you can show that the expected deaths
#>from Indian Point are far fewer than the expected deaths from Con Eds
#>poorly maintained diesel generators, it will not reconcile me to the
#>risk (even if I believed the probability estimates and believed it
#>was an either/or choice).
You admit that under your current perceptions and ideologies, (current?)
nuclear plants will never be acceptable to you. There is nothing wrong with
that opinion (at least you know _why_ you think that way), but it
invalidates any model that intended to assess your weighted costs
associated to nuclear accidents.
#>It's perfectly legit to argue that
#>the average case estimate bolsters the nuclear case, or that even that
#>average case estimates should be used over other costings, but it
#>is another thing entirely to claim that average case represents THE
#>"analytical " approach or that it offers definitive answers.
I never claimed _any_ approach was definitive; that would be
necessarily idiotic and condemn me as a fanatic. I just opined
that the average (unusual) cases gave a better representation as
to the consequential costs than the worst-case (improbable) events.
Naturally, this does not mean that I would feel safe living next
to a major airport even though few jets actually crash nor that I
would not contemplate my brakes going out on my car, etc.; I would
just worry _more_ about the incessant jet wash, the nut-cake
drivers, etc.
I take it that, with certain processes (say, that produces a potent
carcinogen, e.g.), you would not ever be adequately reassured of
the safety standards adhered to?
#>We can make educated guesses:
#>maybe the stock market would go up in survival euphoria, maybe
#>there would be a blackout and a mega riot spiced by fear of what
#>the core was really going to do, maybe evey other nuke near a US city
#>would be forced to shut due to panic, maybe the city water system
#>would have to be shut down as Cesium leaked into the Hudson ....
Or what about the possibility that an operator could push the
worst possible button at the worst possible time, or that an
axe-murderer takes a liking to the coolant system, or maybe an
ex-munitions sociopath decides to smear plastique on all the fuel
rods, or what about aliens disintegrating the containment, ....
It isn't just difficult, but impossible, to anticipate the effects
of _every_ action by _every_body to _every_ event. _Some_where
_some_bodies must decide where to draw the line. Perhaps the
real dispute is where to draw the line between contemplated
scenarios and not-contemplated scenarios?
#>Just the massive human factors element involved makes the probability
#>estimates suspect.
Well, for better or worse, there's big money to be made in that
area. (Pseudo)scientific studies attempting to address individual
factors (number of spoons of sugar in the morning coffee, current
relationship with wife and kids, score of last night's big game, ...)
are being developed to specifically qua(nt/l)ify the human factors
element.
Quite frankly, I will never fully believe that any model can accurately
predict things like which button an operator will push by mistake,
what words an operator will hear incorrectly, etc.
#>I have zero confidence in the ability of the NRC
#>and the nuclear industry to adopt the very high engineering standards
#>required in fission power plants.
That's institutional. I don't trust them either and believe that
corruption plagues the industry just as it does everywhere else.
I do, however, believe that the safety factors incorporated into
the designs make them tolerably unsafe, if you wish to say it that
way. I have much more confidence in newer designs since they rely
less on material limits, but that will have to wait to be tested.
#>[What's the probability of]
#>accident in a plant where the batteries to start the backup control
#>system power generators are not even inspected for a year or two?
Much, much higher than the probability that the backup generators
will start at all (I didn't say that =).
Steinn Sigurdsson
In article <1992Sep24.0...@gn.ecn.purdue.edu> cons...@gn.ecn.purdue.edu (Tino) writes:
>In article <53...@dime.cs.umass.edu> yoda...@chelm.cs.umass.edu (victor yodaiken) writes:
>>>And so I reiterate my question: what is this obsession you have with
>>>worst-case failure modes?
>>Well, I find the worst-case failure consequences of nuclear power to
>>be worrisome.
>Geez, do you stay up all night trembling over the consequences of a large
>meteor hitting the earth? That's a pretty big worst-case failure mode.
Are we currently considering construction of meteors?
Actually, yes. Not only considering, we have.
This raises a curious point - you consider risk in human
constructed activity to be inherently different from "natural"
risks? So you worry about human constructed meteors but not
the more common and riskier natural meteors?
I take it then you rank "worst case nuclear plant accidents"
higher than volcanoes, earthqueakes and weather disasters?
* Steinn Sigurdsson Lick Observatory *
* ste...@lick.ucsc.edu "standard disclaimer" *
Snide Pucker
#>>No doubt this is also true of coal plants. If so, then upon what is the
#>>notion that nuclear plants (of the U.S./Canadian variety) are more dangerous
#>>than coal plants based on?
#>
#>There is no analytical model which will allow us to make an unbiased and
#>objective judgement.
Isn't this the goal of risk analysis? Granted, it will be imperfect,
but it attempts to quantify the "danger" of a particular device.
It is meant to be _one_ of anumber of factors in analyzing the costs
and benefits, not the _sole_ criterion. Aren't we trying to do a bit
too much with analytical models?
Steinn Sigurdsson
You consider worst case nuclear scenarios to be important, yet you
dismiss equally destructive worst case coal plant scenarios as
irrelevant, this would suggest that you are not consistent in your
risk evaluation, you dismiss certain scenarios of comparably low
probability and high cost while going non-linear over others.
Your ice cream analogy is false. It involves materials with
intrinsically different properties, both physical and biological,
than energy production. It still served as a useful reminder that
dairy production has an experimentally measured risk, more people
die from it yearly than from nuclear power generation. Do you really
want me to work up some "worst case" scenarios for dairy? Do you
know the mutation probability of listeria? Can you trust the
biologists estimates?
* Steinn Sigurdsson Lick Observatory *
* ste...@lick.ucsc.edu "standard disclaimer" *
Kevin Brown
>In article <STEINLY.92...@topaz.ucsc.edu> ste...@topaz.ucsc.edu (Steinn Sigurdsson) writes:
>>In article <53...@dime.cs.umass.edu> yoda...@chelm.cs.umass.edu (victor yodaiken) writes:
>>
>>
>> What's so logical about an average case analysis?
>>
>> --------start quote
>> The criteria recommended in this article have no fundamental
>> basis. Indeed, there is no fundamental approach to this issue
>> and no way of proving whether any proposed criteria are right or
>> wrong except by using them over a period of time and discovering
>> whether the costs, risk, and other consequences of their use meet the
>> requirements of society (185)
>> SOURCE: D.J. Higson Nuclear Safety Assessment Criteria
>> Nuclear Safety 31-32 April-June 1990 193-185
>>
>>So you advocate building lots of nuclear reactors and operating them
>>under varying conditions for long periods until we have enough
>>failure samples to do a rigorous estimate of all failure modes
>>and associated risk?
That's strange. I thought we'd already done this. How many nuclear
reactors have we, as a race, been operating, under varying conditions,
in military sea vessels and on land-based power stations, since the
mid-fifties? Surely this is a considerable amount of data?
>No. Please excuse me if I opt out of further elaboration of this issue with
>you, as you seem to lack a committment to rational discussion.
Inasmuch as you've admitted yourself that the risk analysis is basically a
subjective thing, I find it rather humourous that you're bailing out
because the discussion isn't "rational"...
Kevin Brown
>>The only *really* good test of correctness of a model is comparison of its
>>predictions to real-world events.
>>
>>So: how many worst-case failures, of the kind you refer to, have happened
>>in the history of the world to reactors of the kind used in the U.S. and
>>Canada? I dare say none have.
>
>But there have been several severe failure modes which were not forseen.
>True, none of these failures has resulted in a major off site release
>but the fact that's a pretty slim reed to hold onto. The sample space
>is small, extrapolations are dubious.
While this may be true, I was under the impression that the only types
of severe failure that would be of any concern at all involve breaches
of containment. I would think that this would simplify the problem
space considerably.
>>Since more average-case failures have occurred than worst-case failures,
>>it seems ludicrous to suggest that a worst-case analysis is a better
>>means of comparison than an average-case analysis.
>
>Lost me here. By analogy, do you believe that because the brakes have
>never failed in my car, yet the headlights burnt out twice, I should
>ignore the possibility of brake failure? We pay more attention to
>accidents with terrible consequences than to accidents with
>minor consequences.
Yes, we *do* pay more attention to accidents with terrible consequences
than those with minor consequences. I do not question that this is so.
I question that this is wise.
As for your car, what *I'm* saying is that you're probably better off
using the reliability of your headlight system than the reliability of
your brakes as the prime determining factor in the decision of what
form of transportation to use (note, however, that *neither* is probably
a good factor to use in the decision. It's a relative thing).
>>And so I reiterate my question: what is this obsession you have with
>>worst-case failure modes?
>
>Well, I find the worst-case failure consequences of nuclear power to
>be worrisome.
So? Unless you have some personal grudge against nuclear power, you should
find the worst-case failure of *most* power systems to be worrisome. That
being the case, what makes nuclear so special that you obviously devote so
much extra attention to its worst-case failure mode?
It wouldn't be because you happen not to *like* nuclear power, would it?
>>>Low probability/high consequence
>>>events are not necessarily equivalent in riskiness to high probability
>>>low consequence events.
>>
>>By what measure is *this* true? I was under the impression that the
>>very *definition* of statistical risk was probability times cost.
>
>Well, your impression is incorrect. See for example:
[interesting examples deleted]
Interesting stuff. I would ask, then: what is the proper metric to use for
deciding which risk model to use (note: there may not be a good or simple
answer to this one, in which case references would be appreciated)?
>>>That's very nice. But, you have to do two things before you can use
>>>this little model. First, you have to show that the linear equation
>>>actually corresponds to our intuitive concept of risk.
>>
>>In terms of demonstrating objective correctness, correspondence to human
>>intuition of risk is irrelevent. All that matters is that the predictions
>>match real-world observation.
>
>You claim that n deaths distributed over 10 years are equivalent to n
>deaths over 3 days embedded in 10 death free years.
>I disagree. Who's being objective?
Whether or not these are "equivalent" depends on how you're measuring them.
In terms of the numbers alone, they are quite obviously equivalent. Dead
is dead, no matter how it's concentrated.
What you are saying is that there is a *subjective* difference, a *value*
difference. But this depends on what your values are, i.e. how you value
things. It turns out, of course, that most people consider n deaths over
3 days embedded in 10 death-free years to be subjectively worse than n
deaths distributed evenly over 10 years. All this means is that your
(implied) subjective assessment is a popular one.
So: you ask who is being objective. Well, the only *objective* difference
between the two scenarios is the distribution, and they are otherwise
equivalent, at least in objective terms. And that is as far as I'm willing
to go in terms of damage assessment, because I realize that any further
meaning that is assigned is purely subjective. Inasmuch as you brought
this point up, presumably to demonstrate a case in which the two options
are equivalent in terms of the numbers but are "really" much different, I
would have to say it is myself who is being objective, and you who is
being subjective.
>>>Of course not. There is not even an exhaustive list of possible failure modes
>>>of PWR/BWR plants.
>>
>>No doubt this is also true of coal plants. If so, then upon what is the
>>notion that nuclear plants (of the U.S./Canadian variety) are more dangerous
>>than coal plants based on?
>
>There is no analytical model which will allow us to make an unbiased and
>objective judgement.
THANK YOU. You have put into words something that I have observed to be
the case for quite some time.
Since the danger ("risk") presented by each type of plant cannot be assessed
in an unbiased and objective fashion, but the benefits and costs *can* be
assessed in an objective and unbiased fashion (reasoning: we have large
amounts of data on the amount of power produced by each type of plant, the
operating expenses for each, the building costs for each, and the pollutants
produced by each), and since the real world doesn't give a damn about how we
feel about things, I argue that we would be better off, in real-world terms,
making the decision *primarily* based on the objective cost/benefit
evaluation and only *secondarily* based on the subjective risk evaluation.
If the judgement of the danger presented by the various power-plant options
is a subjective one, why is it that the anti-nuclear people insist on
concentrating on that judgement?
Kevin Brown
>>In article <1992Sep23....@nuchat.sccsi.com> ke...@nuchat.sccsi.com (Kevin Brown) writes:
>>#>[...] worst-case failure modes aren't a good prime determining
>>#>factor in comparing alternatives. Rather, *average*-case failure
>>#>modes are.
>>
>> More or less, this is true. Unfortunately, the public isn't quite
>
By definition, an average case analysis will provide you with a better idea
of what will *probably* happen in the real world. Since we're living in a
world where the average case happens more often (by definition) than less
probable (but possibly more spectacular) cases, in general it makes more
sense to pay more attention to the average case. This is because you will
have to deal with the average case more often (again, by definition).
All of this is basically irrelevent, however, with respect to the statement
about the public. The problem is that the public treats nuclear power
*differently* than it does other forms of power. The public's view of
nuclear power seems to be much more negative than their view of other forms
of power, even though nuclear power isn't *objectively* much worse (if any)
than any other form of power generation (except *possibly* in the worst
case, and even *that* is debatable, inasmuch as the worst case scenario
for other forms of power generation is rarely discussed, and thus not
clearly defined). Thus, the public's largely negative view of nuclear
power is illogical, because it is subjective and highly biased.
I would argue that most people use the average case as the primary metric
of comparison of alternatives for most of the decisions they make. That
they seem not to do this with respect to nuclear power indicates to me an
illogical bias.
>--------start quote
> The criteria recommended in this article have no fundamental
> basis. Indeed, there is no fundamental approach to this issue
> and no way of proving whether any proposed criteria are right or
> wrong except by using them over a period of time and discovering
> whether the costs, risk, and other consequences of their use meet the
> requirements of society (185)
Of course, if this is the case, then any criteria is as good as any other,
until you have data, and thus a positive, glowing assessment of the risks
of nuclear power are just as good as a negative, foreboding assessment.
What, then, causes you to use the latter assessment rather than the former
assessment?
>> Statistical probability relies on (relatively) large sample
>> populations. When you don't have them, you are forced to rely on
>> combining individual failure probabilities to generate a sum
>> failure mode probability for a particular scenario. To say that
>> the fault trees become intricate and elaborate is to say that the
>> sun might be hot on its surface. This is more a case for
>> standardization and reduction of components than a case for banning
>> nuclear power plants, by the way.
>
>Sure. And how about for spending the money required to do some extensive
>experiments? The recent fracas over measuring water levels in BWR reactors
>was a good example of the nuclear regulatory/engineering mess.
Since you are very quick to point out that a number of failure modes of
nuclear power plants that were not anticipated have occurred, I doubt
that such experiments, no matter what they may indicate, would do anything
at all to sway you from your opinion that nuclear power is a Bad Idea,
since such experiments must be carefully controlled by their very nature,
and as such *cannot* account for everything that can happen in the real
world (indeed, they are designed to carefully *limit* what can happen).
Since you have admitted that the risk assessment of a power-generating
method (be it nuclear, coal, or whatever) is a subjective thing (inasmuch
as it's not an *objective* thing), and since your objections to nuclear
power seem to be based *solely* on your view of the risks of using nuclear
power (for some reason, the worst-case scenario with respect to nuclear
power causes you great concern, yet you clearly do not treat the worst-case
scenario with respect to your car with the same concern, else you would
reach the same conclusion about driving that you reach about nuclear
power), it follows that your rejection of nuclear power is entirely
subjective in nature, i.e. a mere opinion with no objective grounding.
That being the case, why should we take your opinion seriously?
Kevin Brown
>>1) The phrase "our intuitive concept of risk" presupposes the
>>existence of a common conception of risk, which I find (intuitively) a
>>risky presupposition. If you want to demand that a model correspond
>>to something, it strikes me as only fair that you show that there
>>really is something to correspond to, perhaps even say what it is, and
>>make some argument about why such a correspondence is necessary,
>>useful or beneficial. A complete version of the latter would also
>>include an admission of the drawbacks of requiring such a
>>correspondence, which is this case are significant.
>
>You have placed the burden of proof on the wrong party. Kevin advances
>a formula which he claims will help to provide an analytical basis for
>determining relative risks of power production methods.
Not quite. The formula I advance is supposed to help provide an analytical
basis for determining relative *costs* (in dollars, lives, etc.) of power
production methods. My mistake, if any, was thinking that risks and costs
are basically equivalent. As you so quite correctly pointed out, however,
there are many ways of defining "risk", and so I retract my idea that risk
and cost are the same (I would argue that, at the very least, one *possible*
way of defining risk is to equate it with cost).
Kevin Brown
Sheesh. Don't you know how to take things to their logical conclusion?
If worst case nuclear accidents are important considerations, then worst
case coal plant accidents are *also* important considerations, unless you
have some good argument that worst case coal plant accidents are somehow
less worthy of consideration. I have not seen such an argument from you.
Since the nuclear plant events that you are concerned with are due to
known improbabilities, and (by Stein's argument) a known improbability
(which Stein argues is an event which is at least as probable as the
worst case scenarios you seem so fond of) can turn a coal plant into an
instrument of mass destruction of the same scale (if not greater) as the
worst-case scenarios you like to talk about, it follows that there should
be equal concern on your part about both cases.
Yet, there is *not* equal concern (on your part, at any rate) about both
cases. Why is that?
I think the reason is that you're biased against nuclear power.
Kevin Brown
[On Victor's comparison of an ice cream disaster with a coal plant disaster]
>Your ice cream analogy is false. It involves materials with
>intrinsically different properties, both physical and biological,
>than energy production. It still served as a useful reminder that
>dairy production has an experimentally measured risk, more people
>die from it yearly than from nuclear power generation. Do you really
>want me to work up some "worst case" scenarios for dairy? Do you
>know the mutation probability of listeria? Can you trust the
>biologists estimates?
Heh. You know, with Victor's apparent obsession for worrying about worst-
case situations, I have to wonder whether or not we can cause him to become
truly paranoid if we describe the various worst-case situations that apply
to the things he uses every day. :-) :-) :-)
>* Steinn Sigurdsson Lick Observatory *
victor yodaiken
>In article <53...@dime.cs.umass.edu> yoda...@chelm.cs.umass.edu (victor yodaiken) writes:
>
>>They are? The simplicity of a mathematical model is not an infallible
>>indication of its correctness. Low probability/high consequence
>>events are not necessarily equivalent in riskiness to high probability
> ^^^^^^^^^^^^
>>low consequence events.
>
>Since the whole question turns on this point, I'd suggest sharpening
>the point by saying 'in the human valuation placed on the riskiness'.
>
>That correctly puts the focus on the values of the folks being exposed
>to the risk and takes it off the engineering definition of calculated
>abstract 'risk'.
>
>The alternative is a long border war of the form "Engineers don't do
>risk stuff right" vs. "Non-Nukers are innumerate flakes since they don't
>accept our mathematical model of risk". The reality is more along the
>lines that pro-nukers do have good models, but have some limits on
>quality of input data, and that anti-nukers don't value the probable
>outcomes in the same way that engineers do.
I'm not willing to accept that "engineers" must use linear models. If one
looks at the energy risk scientific literature, one will find that non-linear
models are sometimes employed and that the issues of how to weight evidence
and how much confidence to place in risk estimates are subject to
extensive and wide ranging debate. My concern is that some engineers and
some non-engineers are asserting that the linear measure *is* risk and
that risk analysis essentially proves that the anti-nuclear side is
anti-rational. If you don't understand the limitations of your mathematical
models, you cannot properly understand the issues at hand.
[lucid discussion of differences in perception elided]
--
Carl J Lydick
>}} In article <schumach....@convex.convex.com> schu...@convex.com (Richard A. Schumacher) writes:
>}}
>}} Be sure to point out that this horror story is an argument for design
>}} standardization, not an argument against nuclear plants.
>
>I think the problem isn't with the design, but with the construction
>(implementation vs. design). How do we ensure that construction is up
>to spec without undue gov't intervention?
Well, having a set of standard designs rather than every unit a custom design
has been shown to allow better quality control in many other industries. Seems
like it ought to help in the nuclear power industry, too.
--------------------------------------------------------------------------------
Carl J Lydick | INTERnet: CA...@SOL1.GPS.CALTECH.EDU | NSI/HEPnet: SOL1::CARL
Disclaimer: Hey, I understand VAXen and VMS. That's what I get paid for. My
understanding of astronomy is purely at the amateur level (or below). So
unless what I'm saying is directly related to VAX/VMS, don't hold me or my
organization responsible for it. If it IS related to VAX/VMS, you can try to
hold me responsible for it, but my organization had nothing to do with it.
Carl J Lydick
=In article <191...@hpspkla.spk.hp.com= boro...@hpspkla.spk.hp.com (Don T. Borowski) writes:
==If the steam section of a nuke plant makes someone nervous, then a
==coal fired plant would probably make them die of fright. It is a
==wonderment that any electricity is generated at all, by this sort of
==logic.
==
==If a faucet springs a leak in the restroom of a nuke plant, we hear about
==it on the news. A boiler explosion at a coal fired plant scarcely rates
==a mention.
=
=Clearly, the worst case failure mode of a nuclear fission plant is
=somewhat more spectacular than the worst case failure mode of a coal
=plant. The inability of many nuclear advocates to understand this
=difference does not damp the skepticism of non-fans.
Nor do the non-fans seem capable of understanding that the day-to-day, normal
operation of a coal-fired plant involves emissions that would be considered a
major failure for a nuclear plant.
Carl J Lydick
=In article <53...@dime.cs.umass.edu> yoda...@chelm.cs.umass.edu (victor yodaiken) writes:
=
=
= What's so logical about an average case analysis?
=
= --------start quote
= The criteria recommended in this article have no fundamental
= basis. Indeed, there is no fundamental approach to this issue
= and no way of proving whether any proposed criteria are right or
= wrong except by using them over a period of time and discovering
= whether the costs, risk, and other consequences of their use meet the
= requirements of society (185)
= SOURCE: D.J. Higson Nuclear Safety Assessment Criteria
= Nuclear Safety 31-32 April-June 1990 193-185
=
=So you advocate building lots of nuclear reactors and operating them
=under varying conditions for long periods until we have enough
=failure samples to do a rigorous estimate of all failure modes
=and associated risk?
Isn't that what the Russians did at Chernobyl?
Carl J Lydick
>which involves summing the possible failure modes of PWR/BWR plants.
>I argue that this summation is not calculable because the number
>of terms is unknown. If other aspects of the formula are also
>not calculable it only strengthens my objections.
Ok, then Yackadamn, how about YOU support YOUR claim that comparison of
worst-case scenarios is the valid way to deal with risk. Please note that such
a comparison rules out solar power (a nova's a pretty serious worst-case
scenario).
Carl J Lydick
>hydrogen bombs are equivalent to those involved in manufacturing
>ice cream as one never knows when a tornado might send lethal fragments
>of ice cream flying through the air at high speeds.
No, Victor, it's YOUR reasoning (the way to compare risks is to compare
worst-case scenarios) that yields the above conclusion.
Carl J Lydick
>I think the engineers are quite right in how the calculate risk, I just
>don't think their premises are fully proven ... I'll take a 10% risk of
>loosing one member of my staff over a 1% risk of loosing them all (for
>a staff of 10) where the product of #folks x probability of loss is
>the same for both events. Why? The consequences hit a threshold that
>is just not acceptable ...
Fine. In that case, use a non-linear cost function, such that C(2n) > 2C(n).
Please note that in the original presentation of the formula for risk, the
poster specifically pointed out that you get to pick the cost function that you
think is appropriate. Victor's cost function seems to be a function of one
variable: Whether the term "nuclear" is used in the name of the technology.
If it is, risk is infinite; if it isn't, risk is zero.
Carl J Lydick
==But there have been several severe failure modes which were not forseen.
==True, none of these failures has resulted in a major off site release
==but the fact that's a pretty slim reed to hold onto. The sample space
==is small, extrapolations are dubious.
=
=While this may be true, I was under the impression that the only types
=of severe failure that would be of any concern at all involve breaches
=of containment. I would think that this would simplify the problem
=space considerably.
Not to Victor, it doesn't. His argument is basically: Suppos the fault tree
in a risk analysis only included having a particular valve stick and didn't
include having the operator disable the valve. Then if a failure actually
occurs because an operator disables a valve, this is a complete failure of the
fault tree, rendering every aspect of the analysis totally invalid. It's
really impossible to hold a rational discussion with someone who holds that
attitude.
victor yodaiken
>In principle, there is nothing wrong with developing an _estimate_ of
>risk, particularly the relative risks of similar systems, by a summation
>of probability-hazard sub-products. In practice, we can never know all
>possible failure modes or the exact probabilities associated with them.
As long as we know the difference between "estimation" and measurement.
And as long as we don't fall into the trap of believing that a model with
known and obvious limitations is definitive. Fortunately or unfortunately,
many of our decisions about technology and energy must be decided using
both educated guesswork and politics --- and both of these are known to be
error prone and subject to controversy.
>Basing our social, political, and economic judgments on worst-case
>assessments could place us in situations with demonstrably higher risks,
>however fuzzy the accuracy of the estimates.
I do not suggest that we base all of our social and economic judgements
on worst-case assessments. I do however, note that averaging loses
information, and that basing our engineering judgements on average case
analysis can often lead to very poor decisions. Again, the models are
approximations of reality, blindly following any model is a recipe for
failure.
>Risk analysis is, and will remain, an uncertain science. But we live
>with a vast array of uncertainties. This should not immobilize us.
>Consider a choice of home protective devices:
>
> What would you spend your money on, a meteor detector or a circuit
> breaker for your electrical system?
But would you rather spend your money on a circuit breaker or on high
traction floors for the bathrooms? Perhaps (and I'm making this up) your
chance of slipping in the shower is higher than your chance of overloading
your electrical circuits, and perhaps the expected result of a circuit
overload is a minor heating of some wires, especially if the wiring has
a built in margin of error. But the worst case behavior of an overloaded
circuit is the destruction of your house. Now, Mr. Sigurdsson might point
out that it is statistically possible for Brownian motion to make you
explode after slipping in the shower. Gotta use your best judgement here.
And, with nuclear, the question of the dividing line between
credible and incredible accidents is contentious.
--
victor yodaiken
>In article <53...@dime.cs.umass.edu> yoda...@chelm.cs.umass.edu (victor yodaiken) writes:
>#>In article <1992Sep23....@nuchat.sccsi.com> ke...@nuchat.sccsi.com (Kevin Brown) writes:
>
>#>>No doubt this is also true of coal plants. If so, then upon what is the
>#>>notion that nuclear plants (of the U.S./Canadian variety) are more dangerous
>#>>than coal plants based on?
>#>
>#>There is no analytical model which will allow us to make an unbiased and
>#>objective judgement.
>
> Isn't this the goal of risk analysis? Granted, it will be imperfect,
> but it attempts to quantify the "danger" of a particular device.
> It is meant to be _one_ of anumber of factors in analyzing the costs
> and benefits, not the _sole_ criterion. Aren't we trying to do a bit
> too much with analytical models?
There is a common belief among some technical workers that real world
problems should be as amenable to simple calculations as were the problems
in their textbooks. And bureaucracies like nice simple numerical
answers. The combination is rather dangerous.
"It should be noted that the qualititative
aggregate results of PSAs, e.g. probability for core melt, for releases of
radioactive materials or for health effects on the public should not be
interpreted as frequencies in a statistical sense, although they are
expressed in like units. Rather, probability is a numeral measure of a state
of knowledge, a degree of belief, a state of confidence."
SOURCE: L.V. Konstantinov "On the Safety of Nuclear Power Plants"
Nuclear Engineering and Design 114 (1989) 2 Page 183
--
victor yodaiken
>Victor, _why_ should the certain R be assigned higher values,
>if you answer that you are proposing an alternative model
>of the risk assessment, if it is to be useful it will be
>quantifiable, ergo you submit an alternative model.
> How precisely do you validate numerical models without
>showing the numbers come out wrong - other than showing
>inconsistencies of course - and how do you demonstrate the
>number is incorrect without proposing an alternative (range) of
>values?
Look at E. M. Smith's example in this thread.
> K claims that formula F provides an analytic method for comparing N and
> C. Y points out that F = sum x_1 ... x_n where n is not known. S rebuts
> Y by announcing that there is an "assumption" that the missing terms
> are not significant. Sorry, but an unverified and contentious "assumption"
> does not an analytical method make.
>
>They most certainly do, show me any quantified system of analysis that
>does not make unverified assumptions. At least with the assumptions
>explicit they can compared with reality.
> Do you seriosuly suggest that numerical analysis using finite
>term sums is inherently improper? Or do you require analytic proof
>that the discarded terms are small? In which case you can never model
>any aspect of physical reality.
We calculate weight or velocity using approximations which ignore
terms. But, measurements show that these calculations are "good enough"
for our purposes. Now we seek to calculate "failure probability of nuclear
power" from a set of terms P(failure i). But, in contrast to previous
cases, we do not have a either an empirical or theoretical proof that
the terms we ignore are dominated by the terms we evaluate. In fact,
if you look the critiques of the Rassmussen report from both inside and
outside the AEC, you will find that much of the debate centers on what
terms to include and what terms to exclude. Many critics argued that
common mode failures, fires, and other posible terms had been either
ignored or calculated incorrectly. These critics arrived at aggregate
failure probabilities which differed significantly from those of the
AEC report.
>Victor, we're dealing with reality here, there are no definitive
>comparisons, if you have a better way please contribute it to the
>literature, or at least make constructive suggestions as to what is
>wrong with the approach suggested, you cannot both demand a risk
>assessment and deny that any risk assessment is possible.
I'm perfectly happy with risk assessment. Risk studies can provide us with
a great deal of insight into a problem. But, one must understand
the limitations of the methods, and must make allowances for the
many biases and oversights which can enter into the calculation.
In this case, to believe that one can answer the question of
"nuclear, yes or no" by comparing the results of a simple
calculation is to deceive oneself.
--
victor yodaiken
>>But there have been several severe failure modes which were not forseen.
>>True, none of these failures has resulted in a major off site release
>>but the fact that's a pretty slim reed to hold onto. The sample space
>>is small, extrapolations are dubious.
>
>While this may be true, I was under the impression that the only types
>of severe failure that would be of any concern at all involve breaches
>of containment. I would think that this would simplify the problem
>space considerably.
Not really, small failures can add up to big failures. For example,
suppose that a backup power system fails during a SCRAM, the loss
of a battery or the improper care of a diesel or the misinstallation
of lightning surge protectors .... could all impact on probabilities
of big failures.
>Yes, we *do* pay more attention to accidents with terrible consequences
>than those with minor consequences. I do not question that this is so.
>I question that this is wise.
>
>As for your car, what *I'm* saying is that you're probably better off
>using the reliability of your headlight system than the reliability of
>your brakes as the prime determining factor in the decision of what
>form of transportation to use (note, however, that *neither* is probably
>a good factor to use in the decision. It's a relative thing).
Did you learn to drive in Boston?
>>>>Low probability/high consequence
>>>>events are not necessarily equivalent in riskiness to high probability
>>>>low consequence events.
>>>
>>>By what measure is *this* true? I was under the impression that the
>>>very *definition* of statistical risk was probability times cost.
>>
>>Well, your impression is incorrect. See for example:
>
>[interesting examples deleted]
>
>Interesting stuff. I would ask, then: what is the proper metric to use for
>deciding which risk model to use (note: there may not be a good or simple
>answer to this one, in which case references would be appreciated)?
Look at the EPRI book previously referenced. Also see Brittle Power by
Lovins for another view of risk.
>>You claim that n deaths distributed over 10 years are equivalent to n
>>deaths over 3 days embedded in 10 death free years.
>>I disagree. Who's being objective?
>
>Whether or not these are "equivalent" depends on how you're measuring them.
>In terms of the numbers alone, they are quite obviously equivalent. Dead
>is dead, no matter how it's concentrated.
Right, but we can easily tag on other numbers which expose the difference.
Suppose that there are 100 people living in my village and that we
chose between the risk of losing 20 people in one disaster and losing
20 people over 10 years. Clearly, the village itself may not be able to
survive a sudden loss of 1/5 of the population, but it may be able to
cope with a longer term loss.
>What you are saying is that there is a *subjective* difference, a *value*
>difference. But this depends on what your values are, i.e. how you value
>things. It turns out, of course, that most people consider n deaths over
>3 days embedded in 10 death-free years to be subjectively worse than n
>deaths distributed evenly over 10 years. All this means is that your
>(implied) subjective assessment is a popular one.
Yes.
>
>So: you ask who is being objective. Well, the only *objective* difference
>between the two scenarios is the distribution, and they are otherwise
>equivalent, at least in objective terms. And that is as far as I'm willing
>to go in terms of damage assessment, because I realize that any further
>meaning that is assigned is purely subjective. Inasmuch as you brought
>this point up, presumably to demonstrate a case in which the two options
>are equivalent in terms of the numbers but are "really" much different, I
>would have to say it is myself who is being objective, and you who is
>being subjective.
No. You are chosing to ignore factors which are not accounted for in your
model. A more complete model might have terms for economic disruption,
emotional trauma, .... The real world is very complex, and we cannot
hope to capture it all in any formula. Your choice to value what your
model can calculate over the subjective, but real, phenonemna which your
model ignores, is not an objective choice.
>>There is no analytical model which will allow us to make an unbiased and
>>objective judgement.
>
>THANK YOU. You have put into words something that I have observed to be
>the case for quite some time.
>
>Since the danger ("risk") presented by each type of plant cannot be assessed
>in an unbiased and objective fashion, but the benefits and costs *can* be
>assessed in an objective and unbiased fashion (reasoning: we have large
>amounts of data on the amount of power produced by each type of plant, the
>operating expenses for each, the building costs for each, and the pollutants
>produced by each), and since the real world doesn't give a damn about how we
>feel about things, I argue that we would be better off, in real-world terms,
>making the decision *primarily* based on the objective cost/benefit
Presumably, we make decisions in order to make ourselves happy, not
to maximize some arbitrary indicator. The "real world" may not give
a damn about how we feel, but we do (or should). We invent
mathematical models of the world to help us understand it, and to
help us understand the consequences of our decisions. We do not, or
should not, come to believe that these models should dictate our
decisions.
>If the judgement of the danger presented by the various power-plant options
>is a subjective one, why is it that the anti-nuclear people insist on
>concentrating on that judgement?
"Subjective" judgments are not necessarily content free. In any case, it's
the best we can do for many issues.
--
Gary Coffman
>>
>> I would consider likely failure modes to be a better comparator than
>
>Nice choice. Depends on the dam and the coal plant, does it not?
>I believe that the dam is being maintained, do I value clear skies
>more than trout fishing, do I believe that new technologies will
>make the coal plant safer, do I believe that averting the possibility
>of a total wipeout from the dam is worth the higher chance of late life
>lung disease from the coal plant plus a low chance of an earlier coal
>induced problem ....
There are two completely different issues here. On one side we have
rationalists talking about actuarial risk, and on the other we have
Victor talking about lifestyle esthetics and hysterical risk. Victor
presents arguments like a person, who when shown that his chance of dying
of heart disease is 1 in 2 and his chance of dying of cancer is 1 in 4
and his chance of drowning in his own bathtub is 1 in 10,000
and his chance of being mugged on the street by a stranger is 1 in
30,000 (*mugged*, not killed mind you), huddles in his home afraid to
venture out onto the street. The rational person would conclude that
his chances are better on the street than in his bathroom and venture
forth and live his life.
Gary
Snide Pucker
#>problems should be as amenable to simple calculations as were the problems
#>in their textbooks. And bureaucracies like nice simple numerical
#>answers. The combination is rather dangerous.
Unfortunately, we do not have an infinite amount of time on this planet,
and condensation of facts and opinions is required. It is exactly this
forcing of a "readily-digestable" answer that got the nuclear industry
into the bind it is fighting its way out of. It also forces people to
overstate their claims in order to prove their points quickly (if
watching this newsgroup makes this believable).
#> [...] Rather, probability is a numeral measure of a state
#> of knowledge, a degree of belief, a state of confidence."
#> SOURCE: L.V. Konstantinov "On the Safety of Nuclear Power Plants"
#> Nuclear Engineering and Design 114 (1989) 2 Page 183
I absolutely (cf. previous post) disagree with this definition (i.e.
this is swill).
Uncertainty in confidence levels is _one_ factor of probability
assessment. Since a containment has never collapsed due to concrete
shearing, should we assign this a high probability if we aren't sure
whether or not this can actually happen and we don't have any
quantitative data on this phenomenon?
Just because the diesel backup generators have a high probability of
failure means that we don't know any better? B---shit.
Michael Zika
>In article <1992Sep24.2...@gn.ecn.purdue.edu> mil...@gn.ecn.purdue.edu (Snide Pucker) writes:
>
>There is a common belief among some technical workers that real world
>problems should be as amenable to simple calculations as were the problems
>in their textbooks.
And the tacit assumption is that those engineering doing PRA are only looking
for a solution as amenable as those as simple (? not the problems in _my_
textbooks) as "textbook" answers. Give them a little credit here -- there
_are_ engineers that act on a different motivation than simplification
(like an attempt to model reality!)
>And bureaucracies like nice simple numerical
>answers. The combination is rather dangerous.
>
> "It should be noted that the qualititative
> aggregate results of PSAs, e.g. probability for core melt, for releases of
> radioactive materials or for health effects on the public should not be
> interpreted as frequencies in a statistical sense, although they are
> expressed in like units. Rather, probability is a numeral measure of a state
> of knowledge, a degree of belief, a state of confidence."
> SOURCE: L.V. Konstantinov "On the Safety of Nuclear Power Plants"
> Nuclear Engineering and Design 114 (1989) 2 Page 183
So, do you suggest that a more complicated model based on the same
_imperfect_ probabilities is better? What is this model? I find your
criticism difficult to swallow when no better solution is presented (or at
least a suggestion on the direction that PRA _should_ be going)
>
>yoda...@chelm.cs.umass.edu
--
|
Michael Zika (zi...@ecn.purdue.edu) | Hey don't ask me "why?",
Purdue University | I'm still working on "how?" !
School of Nuclear Engineering |
victor yodaiken
>In article <53...@dime.cs.umass.edu>, yoda...@chelm.cs.umass.edu (victor yodaiken) writes:
>>By your "reasoning" the risks involved in manufacturing
>>hydrogen bombs are equivalent to those involved in manufacturing
>>ice cream as one never knows when a tornado might send lethal fragments
>>of ice cream flying through the air at high speeds.
>
>No, Victor, it's YOUR reasoning (the way to compare risks is to compare
>worst-case scenarios) that yields the above conclusion.
This is a gross distortion of my actual argument.
--
victor yodaiken
>In article <53...@dime.cs.umass.edu> yoda...@chelm.cs.umass.edu (victor yodaiken) writes:
>>In article <1992Sep23.2...@gn.ecn.purdue.edu> mil...@gn.ecn.purdue.edu (Snide Pucker) writes:
>>>
>>> I would consider likely failure modes to be a better comparator than
>>> worst-case failure modes (which tend to be highly improbable). Would
>>> you be horrified of living downstream of a hydro dam because it might
>>> fail or would you rather breathe the fumes of a coal plant and the
>>> occasional coal fire?
>>
>>Nice choice. Depends on the dam and the coal plant, does it not?
>>And it depends on a lot of imponderables and ethical choices. Do
>>I believe that the dam is being maintained, do I value clear skies
>>more than trout fishing, do I believe that new technologies will
>>make the coal plant safer, do I believe that averting the possibility
>>of a total wipeout from the dam is worth the higher chance of late life
>>lung disease from the coal plant plus a low chance of an earlier coal
>>induced problem ....
>
>There are two completely different issues here. On one side we have
>rationalists talking about actuarial risk, and on the other we have
>Victor talking about lifestyle esthetics and hysterical risk. Victor
No. On one side we have people who apparently have a religious faith in
a model of risk which is explicitly disavowed by the actual scientists
in the field, and on the other hand we have my skepticism.
--
victor yodaiken
>In article <53...@dime.cs.umass.edu> yoda...@chelm.cs.umass.edu (victor yodaiken) writes:
>>In article <1992Sep23.1...@gn.ecn.purdue.edu> mil...@gn.ecn.purdue.edu (Snide Pucker) writes:
>>>In article <1992Sep23....@nuchat.sccsi.com> ke...@nuchat.sccsi.com (Kevin Brown) writes:
>>>#>
>>>#>[...] worst-case failure modes aren't a good prime determining
>>>#>factor in comparing alternatives. Rather, *average*-case failure
>>>#>modes are.
>>>
>>> More or less, this is true. Unfortunately, the public isn't quite
>>> as logical about it. Cases involving the word "radioactive" have
>>> a multiplier of 5 (or 10 or 50...).
>>
>>What's so logical about an average case analysis?
>
>By definition, an average case analysis will provide you with a better idea
>of what will *probably* happen in the real world. Since we're living in a
Is that the definition of average case? Please reference.
>All of this is basically irrelevent, however, with respect to the statement
>about the public. The problem is that the public treats nuclear power
>*differently* than it does other forms of power. The public's view of
>nuclear power seems to be much more negative than their view of other forms
>of power, even though nuclear power isn't *objectively* much worse (if any)
>than any other form of power generation (except *possibly* in the worst
>case, and even *that* is debatable, inasmuch as the worst case scenario
>for other forms of power generation is rarely discussed, and thus not
>clearly defined). Thus, the public's largely negative view of nuclear
>power is illogical, because it is subjective and highly biased.
You define "objective" in very limited, and unobjective manner.
>>--------start quote
>> The criteria recommended in this article have no fundamental
>> basis. Indeed, there is no fundamental approach to this issue
>> and no way of proving whether any proposed criteria are right or
>> wrong except by using them over a period of time and discovering
>> whether the costs, risk, and other consequences of their use meet the
>> requirements of society (185)
>
>Of course, if this is the case, then any criteria is as good as any other,
>until you have data, and thus a positive, glowing assessment of the risks
>of nuclear power are just as good as a negative, foreboding assessment.
Look, you claim a scientific and rational basis for your opinion. Cite
a respectable source from the field of risk analysis. The sources I read
disagree with your opinions.
>Since you are very quick to point out that a number of failure modes of
>nuclear power plants that were not anticipated have occurred, I doubt
>that such experiments, no matter what they may indicate, would do anything
>at all to sway you from your opinion that nuclear power is a Bad Idea,
>since such experiments must be carefully controlled by their very nature,
>and as such *cannot* account for everything that can happen in the real
>world (indeed, they are designed to carefully *limit* what can happen).
Your doubt is not well founded.
>Since you have admitted that the risk assessment of a power-generating
>method (be it nuclear, coal, or whatever) is a subjective thing (inasmuch
>as it's not an *objective* thing), and since your objections to nuclear
>power seem to be based *solely* on your view of the risks of using nuclear
>power (for some reason, the worst-case scenario with respect to nuclear
>power causes you great concern, yet you clearly do not treat the worst-case
>scenario with respect to your car with the same concern, else you would
>reach the same conclusion about driving that you reach about nuclear
>power), it follows that your rejection of nuclear power is entirely
>subjective in nature, i.e. a mere opinion with no objective grounding.
Your reasoning is flawed. But since you previously argued that the
condition of an automobile's headlights were a better indication
of its reliability than the condition of its brakes, I am not surprised.
>That being the case, why should we take your opinion seriously?
Why should I care?
--
victor yodaiken
>In article <1992Sep24....@nuchat.sccsi.com>, ke...@nuchat.sccsi.com (Kevin Brown) writes:
>==But there have been several severe failure modes which were not forseen.
>==True, none of these failures has resulted in a major off site release
>==but the fact that's a pretty slim reed to hold onto. The sample space
>==is small, extrapolations are dubious.
>=
>=While this may be true, I was under the impression that the only types
>=of severe failure that would be of any concern at all involve breaches
>=of containment. I would think that this would simplify the problem
>=space considerably.
>
>Not to Victor, it doesn't. His argument is basically: Suppos the fault tree
>in a risk analysis only included having a particular valve stick and didn't
>include having the operator disable the valve. Then if a failure actually
>occurs because an operator disables a valve, this is a complete failure of the
>fault tree, rendering every aspect of the analysis totally invalid. It's
>really impossible to hold a rational discussion with someone who holds that
>attitude.
If you are going to misrepresent my position so grossly, the least you could
do is provide quotes so that the exact point at which your misunderstanding
became orbital could be identified.
--
victor yodaiken
>In article <1992Sep24....@michael.apple.com>, e...@michael.apple.com (E. Michael Smith) writes:
>>I think the engineers are quite right in how the calculate risk, I just
>>don't think their premises are fully proven ... I'll take a 10% risk of
>>loosing one member of my staff over a 1% risk of loosing them all (for
>>a staff of 10) where the product of #folks x probability of loss is
>>the same for both events. Why? The consequences hit a threshold that
>>is just not acceptable ...
>
>Fine. In that case, use a non-linear cost function, such that C(2n) > 2C(n).
>Please note that in the original presentation of the formula for risk, the
>poster specifically pointed out that you get to pick the cost function that you
>think is appropriate. Victor's cost function seems to be a function of one
>variable: Whether the term "nuclear" is used in the name of the technology.
>If it is, risk is infinite; if it isn't, risk is zero.
Stick to Vaxes, Carl. The subtelties of non-binary reasoning are beyond you.
--
victor yodaiken
>In article <53...@dime.cs.umass.edu> yoda...@chelm.cs.umass.edu (victor yodaiken) writes:
>>>In article <53...@dime.cs.umass.edu> yoda...@chelm.cs.umass.edu (victor yodaiken) writes:
>>>
>>>
>>> What's so logical about an average case analysis?
>>>
>>> The criteria recommended in this article have no fundamental
>>> basis. Indeed, there is no fundamental approach to this issue
>>> and no way of proving whether any proposed criteria are right or
>>> wrong except by using them over a period of time and discovering
>>> whether the costs, risk, and other consequences of their use meet the
>>> requirements of society (185)
>>>
>
>That's strange. I thought we'd already done this. How many nuclear
>reactors have we, as a race, been operating, under varying conditions,
>in military sea vessels and on land-based power stations, since the
>mid-fifties? Surely this is a considerable amount of data?
>
>>No. Please excuse me if I opt out of further elaboration of this issue with
>>you, as you seem to lack a committment to rational discussion.
>
>Inasmuch as you've admitted yourself that the risk analysis is basically a
>subjective thing, I find it rather humourous that you're bailing out
>because the discussion isn't "rational"...
"Rational" and "quantifiable" are not always the same.
--
victor yodaiken
>In article <53...@dime.cs.umass.edu> yoda...@chelm.cs.umass.edu (victor yodaiken) writes:
>>In article <1992Sep24.2...@gn.ecn.purdue.edu> mil...@gn.ecn.purdue.edu (Snide Pucker) writes:
>>
>>There is a common belief among some technical workers that real world
>>problems should be as amenable to simple calculations as were the problems
>>in their textbooks.
>
>And the tacit assumption is that those engineering doing PRA are only looking
>for a solution as amenable as those as simple (? not the problems in _my_
>textbooks) as "textbook" answers. Give them a little credit here -- there
>_are_ engineers that act on a different motivation than simplification
>(like an attempt to model reality!)
Not my assumption, tacit or otherwise. I've been arguing that people who
actually know something about this problem are not as enamored of
over-simplified models as are some of our friends here on the net.
>
>>And bureaucracies like nice simple numerical
>>answers. The combination is rather dangerous.
>>
>> "It should be noted that the qualititative
>> aggregate results of PSAs, e.g. probability for core melt, for releases of
>> radioactive materials or for health effects on the public should not be
>> interpreted as frequencies in a statistical sense, although they are
>> expressed in like units. Rather, probability is a numeral measure of a state
>> of knowledge, a degree of belief, a state of confidence."
>> SOURCE: L.V. Konstantinov "On the Safety of Nuclear Power Plants"
>> Nuclear Engineering and Design 114 (1989) 2 Page 183
>
>So, do you suggest that a more complicated model based on the same
>_imperfect_ probabilities is better? What is this model? I find your
>criticism difficult to swallow when no better solution is presented (or at
>least a suggestion on the direction that PRA _should_ be going)
From my, limited, readings of the PRA literature, I have the impression
that there is a lot of reasonably clever research going on and that
people are experimenting with models. I don't claim to have any special
insight into what models should be used. I do claim that, contrary to
what some people have claimed here, PRAs are limited tools which should be
regarded as data points, but not as definitive answers. And I claim that
it is a grave error to imagine that a linear risk model can be used to
obtain an objective answer to the question of whether or not it is a good
thing to develop nuclear power. This debate features repeated assertions that
we are in the midst of a battle betweenationalists who are convinced
by the results of certain nuclear risk estimates, and anti-rationalists
who are not persuaded by these magical numbers. These arguments rest on
a fundamental ignorance of the limits of PRA methodologies and a naive
faith in numerically presented estimates.
--
Bo Curry
|That correctly puts the focus on the values of the folks being exposed
|to the risk and takes it off the engineering definition of calculated
|abstract 'risk'.
|
|The alternative is a long border war of the form "Engineers don't do
|risk stuff right" vs. "Non-Nukers are innumerate flakes since they don't
|accept our mathematical model of risk". The reality is more along the
|lines that pro-nukers do have good models, but have some limits on
|quality of input data, and that anti-nukers don't value the probable
|outcomes in the same way that engineers do.
|
Good post. I think this analysis sums up the situation quite accurately.
Bo Curry
victor yodaiken
>In article <53...@dime.cs.umass.edu> yoda...@chelm.cs.umass.edu (victor yodaiken) writes:
>#>
>#>There is a common belief among some technical workers that real world
>#>problems should be as amenable to simple calculations as were the problems
>#>in their textbooks. And bureaucracies like nice simple numerical
>#>answers. The combination is rather dangerous.
>
> Unfortunately, we do not have an infinite amount of time on this planet,
> and condensation of facts and opinions is required. It is exactly this
> forcing of a "readily-digestable" answer that got the nuclear industry
> into the bind it is fighting its way out of. It also forces people to
> overstate their claims in order to prove their points quickly (if
> watching this newsgroup makes this believable).
I don't think that there is a good excuse for overstating claims. It
rapidly leads to rather dull shouting matches and leads the public to
a well founded distrust of everyone. Feynman was right, scientists
have a responsibility to lean over backwards to point out the limitations
of their knowledge and the possibility of error in their work. BTW, please
don't take this as a criticism, in fact you have been very responsible in
this discussion. If nuclear power ever regains some public trust, it
will be due to well reasoned and calm advocacy, and will be despite the
shrill efforts of some supposed nuclear advocates.
>
>#> [...] Rather, probability is a numeral measure of a state
>#> of knowledge, a degree of belief, a state of confidence."
>#> SOURCE: L.V. Konstantinov "On the Safety of Nuclear Power Plants"
>#> Nuclear Engineering and Design 114 (1989) 2 Page 183
>
> I absolutely (cf. previous post) disagree with this definition (i.e.
> this is swill).
> Uncertainty in confidence levels is _one_ factor of probability
> assessment. Since a containment has never collapsed due to concrete
> shearing, should we assign this a high probability if we aren't sure
> whether or not this can actually happen and we don't have any
> quantitative data on this phenomenon?
No. But we should remember that our analysis depends on a great deal
of unknowns and not confuse the resulting number with, say, the
probability that a pair of dice will fall with the number 7 on top.
> Just because the diesel backup generators have a high probability of
> failure means that we don't know any better? B---shit.
Not sure what you are getting at here.
--
Kevin Brown
>>While this may be true, I was under the impression that the only types
>>of severe failure that would be of any concern at all involve breaches
>>of containment. I would think that this would simplify the problem
>>space considerably.
>
>Not really, small failures can add up to big failures. For example,
>suppose that a backup power system fails during a SCRAM, the loss
>of a battery or the improper care of a diesel or the misinstallation
>of lightning surge protectors .... could all impact on probabilities
>of big failures.
Okay. Well, let's suppose that the things you're talking about happen.
What sort of "big failure" are we talking here?
My point is this: I don't see how the kind of scenario you're so concerned
with can occur without some kind of release of highly radioactive material
from the reactor. And as far as I know (this could be sheer ignorance on
my part, of course), the only way such a release can occur is when the
containment is breached. After all, the *job* of the containment is to
prevent such a release, yes?
>>Yes, we *do* pay more attention to accidents with terrible consequences
>>than those with minor consequences. I do not question that this is so.
>>I question that this is wise.
I wanted to reiterate this.
We humans have a fascination with the spectacular or the unusual. That
we have a tendency to emphasize such things in no way indicates that we
should base our decisions on those tendencies. For some reason, however,
you seem unable to keep yourself from doing just that, at least with
respect to nuclear power. Why is it that you seem inconsistent in this
regard? Why is it that your decisions with respect to transportation
(to name one example) are not similarly based primarily on the spectacular?
>>As for your car, what *I'm* saying is that you're probably better off
>>using the reliability of your headlight system than the reliability of
>>your brakes as the prime determining factor in the decision of what
>>form of transportation to use (note, however, that *neither* is probably
>>a good factor to use in the decision. It's a relative thing).
>
>Did you learn to drive in Boston?
No. I learned to drive in Houston. Is there a significant difference?
By the way, I have experienced electrical system failure that took the
headlights with it before. The danger such a failure presents should not
be underestimated. It is quite possible that the danger is very close to
the danger presented by a brake system failure.
>>Interesting stuff. I would ask, then: what is the proper metric to use for
>>deciding which risk model to use (note: there may not be a good or simple
>>answer to this one, in which case references would be appreciated)?
>
>Look at the EPRI book previously referenced. Also see Brittle Power by
>Lovins for another view of risk.
Okay. These are views, though, correct? Are they considered definitive?
If not, then please direct me to something that *is* considered definitive
(by the scientific community). If no such thing exists, then I must
conclude that there *is* no single proper metric to use.
>>>You claim that n deaths distributed over 10 years are equivalent to n
>>>deaths over 3 days embedded in 10 death free years.
>>>I disagree. Who's being objective?
>>
>>Whether or not these are "equivalent" depends on how you're measuring them.
>>In terms of the numbers alone, they are quite obviously equivalent. Dead
>>is dead, no matter how it's concentrated.
>
>Right, but we can easily tag on other numbers which expose the difference.
>Suppose that there are 100 people living in my village and that we
>chose between the risk of losing 20 people in one disaster and losing
>20 people over 10 years. Clearly, the village itself may not be able to
>survive a sudden loss of 1/5 of the population, but it may be able to
>cope with a longer term loss.
Change the data the model has to work with, and you change the model. What
you've done above is provide additional information to work with. So, with
the additional information, I agree that the two cases (losing 20 people in
one disaster, with no deaths during the remaiing time and losing 20 people
over a period of 10 years) are not equivalent.
But this is only because you've provided a more detailed context.
If you're trying to say that our decisions must account for the specific
circumstances surrounding them, you'll get no argument from me. However,
the discussion of nuclear power versus coal power has thus far been one
without such detail. It has thus far been about the "general case". As
such, there is no such detail of the type you provide above, and as such
the models will necessarily seem somewhat simplistic.
>>So: you ask who is being objective. Well, the only *objective* difference
>>between the two scenarios is the distribution, and they are otherwise
>>equivalent, at least in objective terms. And that is as far as I'm willing
>>to go in terms of damage assessment, because I realize that any further
>>meaning that is assigned is purely subjective. Inasmuch as you brought
>>this point up, presumably to demonstrate a case in which the two options
>>are equivalent in terms of the numbers but are "really" much different, I
>>would have to say it is myself who is being objective, and you who is
>>being subjective.
>
>No. You are chosing to ignore factors which are not accounted for in your
>model. A more complete model might have terms for economic disruption,
>emotional trauma, ....
Yes, yes, I realize this. But note that these factors occur as *consequences*
of the difference in distribution, which the objective includes implicitly.
>The real world is very complex, and we cannot
>hope to capture it all in any formula.
Ah, but this was not my intent. Rather, my intent is to capture the basic
essence of the real world, and to do so with as little reliance on my own
value set as possible.
>Your choice to value what your
>model can calculate over the subjective, but real, phenonemna which your
>model ignores, is not an objective choice.
Come again? You think it's not an objective decision to place less
significance on that which you cannot determine objectively?
This may be. Ultimately, the desire to attempt to model the real world as
accurately as possible is a value judgement. However, I would argue that
at the very least it is a *useful* value judgement (which, one may argue,
is in itself a subjective term).
[on risk]
>>>There is no analytical model which will allow us to make an unbiased and
>>>objective judgement.
>>
>>THANK YOU. You have put into words something that I have observed to be
>>the case for quite some time.
>>
>>Since the danger ("risk") presented by each type of plant cannot be assessed
>>in an unbiased and objective fashion, but the benefits and costs *can* be
>>assessed in an objective and unbiased fashion (reasoning: we have large
>>amounts of data on the amount of power produced by each type of plant, the
>>operating expenses for each, the building costs for each, and the pollutants
>>produced by each), and since the real world doesn't give a damn about how we
>>feel about things, I argue that we would be better off, in real-world terms,
>>making the decision *primarily* based on the objective cost/benefit
>
>Presumably, we make decisions in order to make ourselves happy, not
>to maximize some arbitrary indicator. The "real world" may not give
>a damn about how we feel, but we do (or should). We invent
>mathematical models of the world to help us understand it, and to
>help us understand the consequences of our decisions. We do not, or
>should not, come to believe that these models should dictate our
>decisions.
Right. But note that this doesn't contradict what I'm saying.
That we care about our own happiness is precisely why we attempt to model
the real world, just as you say. The object is to control our own actions
such that the feedback we receive from the real world maximizes our
happiness. The assumption is that the more accurate our models are, the
closer we can get to this maximum (i.e., the happier we'll be). Inasmuch
as the models we use are basically our best guess about how the real world
actually works, it seems obvious to me that if you care so much about our
happiness, you'll rely on those models as the primary determining factor
in your decisions (reasoning: the models predict real world behavior based
on the decisions you make. While no model is 100% accurate, most are
better than chance. Not using them is, I would argue, equivalent to relying
on chance. You might get lucky, but you probably won't).
Should the models dictate (i.e., be the primary determining factor in) our
decisions? If you care about happiness, then I think the answer is a
resounding "yes". Unless, of course, you happen to have a better (i.e.,
more accurate) model. But in *that* case, your model should replace the
one currently in use, and thus become one of "the models".
>>If the judgement of the danger presented by the various power-plant options
>>is a subjective one, why is it that the anti-nuclear people insist on
>>concentrating on that judgement?
>
>"Subjective" judgments are not necessarily content free.
True. But because they depend primarily on your value set, they are, in
essence, random, and thus ill-suited to the task of making decisions about
the real world.
>In any case, it's >the best we can do for many issues.
Yes, it *is* the best we can do for many issues. The problem is that, at
least with respect to the issue of nuclear power, it *isn't* the best we
can do. In the case of nuclear power as an alternative to coal power, we
have behavioral models and cost/benefit models which have real-world data
to support them, and that means that, if we choose to base our decision on
them, we can do better than chance in getting the results we want. Basing
our decision purely on subjective judgements, however, is effectively the
same thing as relying on chance to make the decision.
Most decisions are a combination of objective and subjective judgements. I
think that decisions about power-generation technology are no exception. My
primary argument is that these decisions should be based *primarily* on the
objective judgements and *secondarily* on the subjective judgements. At
least then it's not as likely that you'll be fighting the real world (a
fight which you're guaranteed to lose, inasmuch as you are subject to its
laws and not vice-versa).
Kevin Brown
>In article <1992Sep25.2...@gn.ecn.purdue.edu> zi...@gn.ecn.purdue.edu (Michael Zika) writes:
>>In article <53...@dime.cs.umass.edu> yoda...@chelm.cs.umass.edu (victor yodaiken) writes:
>>>In article <1992Sep24.2...@gn.ecn.purdue.edu> mil...@gn.ecn.purdue.edu (Snide Pucker) writes:
>>>
>>>There is a common belief among some technical workers that real world
>>>problems should be as amenable to simple calculations as were the problems
>>>in their textbooks.
>>
>>And the tacit assumption is that those engineering doing PRA are only looking
>>for a solution as amenable as those as simple (? not the problems in _my_
>>textbooks) as "textbook" answers. Give them a little credit here -- there
>>_are_ engineers that act on a different motivation than simplification
>>(like an attempt to model reality!)
>
>Not my assumption, tacit or otherwise. I've been arguing that people who
>actually know something about this problem are not as enamored of
>over-simplified models as are some of our friends here on the net.
This may be the case, but are they as unwilling to use these models as you
seem to be?
Are they as insistent on concentrating on the worst-case scenario as you are?
I doubt it, but that is mere speculation on my part.
>>So, do you suggest that a more complicated model based on the same
>>_imperfect_ probabilities is better? What is this model? I find your
>>criticism difficult to swallow when no better solution is presented (or at
>>least a suggestion on the direction that PRA _should_ be going)
>
>From my, limited, readings of the PRA literature, I have the impression
>that there is a lot of reasonably clever research going on and that
>people are experimenting with models. I don't claim to have any special
>insight into what models should be used. I do claim that, contrary to
>what some people have claimed here, PRAs are limited tools which should be
>regarded as data points, but not as definitive answers. And I claim that
>it is a grave error to imagine that a linear risk model can be used to
>obtain an objective answer to the question of whether or not it is a good
>thing to develop nuclear power.
You have made a stronger claim than this. You have claimed that *no* risk
model can be used to obtain an objective answer, since you have claimed that
no objective risk model exists.
>This debate features repeated assertions that
>we are in the midst of a battle betweenationalists who are convinced
>by the results of certain nuclear risk estimates, and anti-rationalists
>who are not persuaded by these magical numbers. These arguments rest on
>a fundamental ignorance of the limits of PRA methodologies and a naive
>faith in numerically presented estimates.
This is incorrect.
It is not a question of whether we should be using nuclear power. It
is a question of how nuclear power fares when compared with viable current
alternatives, e.g. coal power. The need for power generation is there.
*Something* has to be used to generate that power.
You can't have your cake and eat it too. If you *insist* that linear
models don't apply to nuclear power, then you must also insist that they do
not apply to the alternatives, unless you have a good, solid argument to
back you up for the coal power case (one which *cannot* apply to the nuclear
power case, i.e. the two must be somehow fundamentally different in a
significant, with respect to the choice of risk models, way).
We are debating the decision of which form of power generation meets our
needs best. That decision has to be based on *something*. If you insist
that a linear cost/benefit analysis is inappropriate to the discussion,
then don't just stand on your hands, SUGGEST SOMETHING TO USE THAT IS
BETTER SUITED. If you cannot do this, then you have NO BUSINESS complaining
that the models being used are somehow inadequate.
Steinn Sigurdsson
Change the data the model has to work with, and you change the model. What
you've done above is provide additional information to work with. So, with
the additional information, I agree that the two cases (losing 20 people in
one disaster, with no deaths during the remaiing time and losing 20 people
over a period of 10 years) are not equivalent.
Funnily enough that argument strengthens the case for nuclear power!
If you argue losing 20 people in a single incident is worse than
losing 20 people statistically over a number of years, then almost all
nuclear plant accident scenarios are preferred as even the worst cases
involve mostly statistical risk to a population - I can think of
no credible scenario where a large population would receive a lethal
dose from a nuclear plant. In contrast, fossil fuel plant accidents typically
involve 10-100 deaths on an occasional basis, including transport
of fuel, mining and waste disposal - with waste disposal arguably
the historically most hazardous. This is in _addition_ to the
statistical risk that fossil plants pose by carcinogen emission,
SO2 emission and potential greenhouse effects.
If you dislike low probability scenarios with high potential costs
then just the potential greenhouse effects should rule out fossil
fuel plants.
* Steinn Sigurdsson Lick Observatory *
* ste...@lick.ucsc.edu "standard disclaimer" *
* But, oh, love is strange *
* and you have to learn to take the crunchy with the smooth, *
* I suppose - B.B. 1983 *
Carl J Lydick
>In article <19v8o3...@gap.caltech.edu> ca...@SOL1.GPS.CALTECH.EDU writes:
>>In article <53...@dime.cs.umass.edu>, yoda...@chelm.cs.umass.edu (victor yodaiken) writes:
>>>By your "reasoning" the risks involved in manufacturing
>>>hydrogen bombs are equivalent to those involved in manufacturing
>>>ice cream as one never knows when a tornado might send lethal fragments
>>>of ice cream flying through the air at high speeds.
>>
>>No, Victor, it's YOUR reasoning (the way to compare risks is to compare
>>worst-case scenarios) that yields the above conclusion.
>
>This is a gross distortion of my actual argument.
Well, in that case, since it's what quite a few readers of this newsgroup
thought you were saying, how about clarifying your argument?
Carl J Lydick
>> and condensation of facts and opinions is required. It is exactly this
>> forcing of a "readily-digestable" answer that got the nuclear industry
>> into the bind it is fighting its way out of. It also forces people to
>> overstate their claims in order to prove their points quickly (if
>> watching this newsgroup makes this believable).
>
>I don't think that there is a good excuse for overstating claims.
Yackadam, the eternal bleater of "let's base our analysis on worst-case
scenarios" doesn't think there's a good excuse for overstating claims. Well,
at least he seems to have a sense of humor.
Carl J Lydick
>In article <53...@dime.cs.umass.edu> yoda...@chelm.cs.umass.edu (victor yodaiken) writes:
>>Right, but we can easily tag on other numbers which expose the difference.
>>Suppose that there are 100 people living in my village and that we
>>chose between the risk of losing 20 people in one disaster and losing
>>20 people over 10 years. Clearly, the village itself may not be able to
>>survive a sudden loss of 1/5 of the population, but it may be able to
>>cope with a longer term loss.
>
>Change the data the model has to work with, and you change the model. What
>you've done above is provide additional information to work with. So, with
>the additional information, I agree that the two cases (losing 20 people in
>one disaster, with no deaths during the remaiing time and losing 20 people
>over a period of 10 years) are not equivalent.
>But this is only because you've provided a more detailed context.
No, it's because he lied to us in the original statement of the problem.
Since, by his own definition, "the village itself may not be able to survive
a sudden loss of 1/5 of the population," we're *NOT*, as Yackadam claims in the
original premise, talking about 20 people over 10 years vs. 20 people in one
disaster. We're talking about losing 20 people over 10 years vs. losing 20
people in one disaster and the remaining 80 over the next 10 years.
victor yodaiken
>In article <53...@dime.cs.umass.edu> yoda...@chelm.cs.umass.edu (victor yodaiken) writes:
>>Not my assumption, tacit or otherwise. I've been arguing that people who
>>actually know something about this problem are not as enamored of
>>over-simplified models as are some of our friends here on the net.
>
>This may be the case, but are they as unwilling to use these models as you
>seem to be?
I have not seen any claims in the scientific literature that risk models
invalidate subjective concerns such as doubts about the efficacy of
the NRC or doubts about whether probability estimates are correct
within an order of magnitude. Note that the AEC's own review of the
Wash-1400 report produced some predictions of a major accident every
50 years of 200 reactor operation (or something like that, perhaps even
more).
>Are they as insistent on concentrating on the worst-case scenario as you are?
>
>I doubt it, but that is mere speculation on my part.
I strongly suggest you look at the literature.
>>From my, limited, readings of the PRA literature, I have the impression
>>that there is a lot of reasonably clever research going on and that
>>people are experimenting with models. I don't claim to have any special
>>insight into what models should be used. I do claim that, contrary to
>>what some people have claimed here, PRAs are limited tools which should be
>>regarded as data points, but not as definitive answers. And I claim that
>>it is a grave error to imagine that a linear risk model can be used to
>>obtain an objective answer to the question of whether or not it is a good
>>thing to develop nuclear power.
>
>You have made a stronger claim than this. You have claimed that *no* risk
>model can be used to obtain an objective answer, since you have claimed that
>no objective risk model exists.
Yes.
>It is not a question of whether we should be using nuclear power. It
>is a question of how nuclear power fares when compared with viable current
>alternatives, e.g. coal power. The need for power generation is there.
>*Something* has to be used to generate that power.
>
>You can't have your cake and eat it too. If you *insist* that linear
>models don't apply to nuclear power, then you must also insist that they do
>not apply to the alternatives, unless you have a good, solid argument to
>back you up for the coal power case (one which *cannot* apply to the nuclear
>power case, i.e. the two must be somehow fundamentally different in a
>significant, with respect to the choice of risk models, way).
It seems obvious to me that the risks posed by coal are of a different
kind than the risks posed by nuclear fission. Coal has many obvious
problems, but given current technology, no coal plant can cause the
same level of devastation that we see at Chernobyl (unless we want to
take the improbabilities of statistical mechanics into account and
thus render the whole discussion futile). So the question
with nuclear (one question anyway) is whether US PWR/BWR plants can fail
on a similar scale. My current belief is that such a failure is possible,
even though I realize that US reactors are far better designed and
maintained than XSSR ones (partly because the Soviet nuclear power designers
never had to deal with the "irrational and hysterical" fears of the populace).
My belief is based on a deep skepticism of the design, construction,
and siting process used for US plants, suspicion of the abilities of
the utilities and NRC to properly manage the plants, and concern over
the long-term management of nuclear mishaps (i.e. what happens 100 years
after a meltdown has caused us to entomb an entire reactor building in
concrete.) For examples of why I'm skeptical of the design process,
consider the failure to orient turbines properly in some plants, the
nutty Brown's Ferry control cabling and the TMI hidden displays (on the
back of the panel where the operators could not see it). For examples
of construction and siting problems, see Okrent's "History" and the
Union of Concerned Scientists "Safety Second". Safety Second provides
some very well documented reasons to be suspicious of the NRC/AEC's
vigilance, and Okrent's book also illustrates how concern for public
relations and keeping the very politically powerful manufacturers happy
was allowed to override engineering concerns, time and again.
Since there appears no pressing need for immediate construction of
new generating capacity, my belief is that it pays to be cautious here.
I'd rather spend money on solar, conservation, insulation, natural
gas, fancy pollution control equipment .... For detailed descriptions
of alternate energy paths see proposals by the UCS, EPRI, and the
Rocky Mountain Institute. In these you will find in depth analysis of
comparative costs.
--
Carl J Lydick
>It seems obvious to me that the risks posed by coal are of a different
>kind than the risks posed by nuclear fission. Coal has many obvious
>problems, but given current technology, no coal plant can cause the
>same level of devastation that we see at Chernobyl (unless we want to
>take the improbabilities of statistical mechanics into account and
>thus render the whole discussion futile). So the question
>with nuclear (one question anyway) is whether US PWR/BWR plants can fail
>on a similar scale.
No, one question is whether they likelihood of US PWR/BWR plants failing on a
similar scale is comparable to the likelihood of US coal plants failing on a
similar scale. You can't argue that something's too improbable for
consideration with regard to coal plants, then use worst-case scenarios on
nuclear plants. Come on Victor, try to show at least a modicum of integrity.
victor yodaiken
>In article <53...@dime.cs.umass.edu>, yoda...@chelm.cs.umass.edu (victor yodaiken) writes:
>>> Unfortunately, we do not have an infinite amount of time on this planet,
>>> and condensation of facts and opinions is required. It is exactly this
>>> forcing of a "readily-digestable" answer that got the nuclear industry
>>> into the bind it is fighting its way out of. It also forces people to
>>> overstate their claims in order to prove their points quickly (if
>>> watching this newsgroup makes this believable).
>>
>>I don't think that there is a good excuse for overstating claims.
>
>Yackadam, the eternal bleater of "let's base our analysis on worst-case
>scenarios" doesn't think there's a good excuse for overstating claims. Well,
>at least he seems to have a sense of humor.
Your content free, sophmoric postings are really out of place in a sci
newsgroup. If you wish to base your arguments on witty transformations
of my name, you would be better off in one of the alt groups.
--
Ray Depew
[ lots of stuff ]
Ken, I don'thtuink that you have TOO much to worry about, except for the
possibility that this Howard Fast character has a golden tongue.
> .... And in all truth, beyond the fact
> that they are enormously profitable to those who build them and supply
> the atomic fuel, there is no real necessity for atomic power.
>
> We have enough natural gas, combined with other sources, to provide clean
> energy for years to come. New agricultural techniques make it possible to
> grow sufficient plant sources of fuel to take care of any petroleum
> shortfall. We have vast sources of coal, and scientists are finding new
> nonpolluting ways to use it, and the advances in wind-power and sun-power
> open a whole new, clean world of power.
At best, this is a horrible misunderstading. At worst, it is an outright
and intentional lie.
I took an "Energy and the Environment" course back in 1980. One assignment
was (1) to research the world's projected energy
needs for the present, the next decade and the next century, (2) to research
the world's projected energy resources for the same time periods,
including petroleum imports, and (3) to compare the two. We reached the
sobering conclusion that WE ARE GOING TO NEED EVERY DROP OF ENERGY WE CAN
SQUEEZE FROM EVERY RESOURCE WE HAVE, UNLESS HUMANS ARE WILLING TO
DRASTICALLY REDUCE THEIR PER-CAPITA CONSUMPTION OF ENERGY. (This means
solar, wind, photovoltaic, biomass, nuclear, coal, ... you name it.)
Even ZPG wouldn't be enough to solve the problem, because per-capita
energy consumption in the world continues to increase, despite all our
efforts to conserve. (Sorry about the ALL CAPS, but Mr. Fast might need
to see them as well.)
Anyone who thinks we can eliminate one current or future source of energy,
be it coal, nuclear or hydro, is living in a fool's paradise. If we were
to shut off nuke power, be it a little at a time or all at once, quality-
of-life would suffer the world over.
> When one thinks of the trillions of dollars that have been poured into
> creating atomic-energy plants Q all of them doomed by the dangers they
> threaten and by their innate shortness of life, as compared to other
> sources of power Q one can conclude only that we have an inborn and
> unshakable streak of insanity.
I'd demand some documentary proof for this emotional, inflammatory and
unprovable statement.
I take it Mr. Fast hasn't driven northward into Canada to see all of the
hardwood forests which have been killed by acid rain. Scientific American
contained a series of articles not too long ago (no references, sorry),
complete with maps, demonstrating how the plumes from the USA's eastern
coal-fired power plants and factories correlated with the forest kills in
eastern Canada.
He's probably never concerned himself, either, with the health hazards
associated with coal mining, including the fatal respiratory illnesses
suffered by miners and entire mining towns -- or with the high
concentrations of toxic heavy metals in the water which leaches out of
piles of coal ash -- or with the fact that coal and crude oil contain
lethally high concentrations of proven (not suspected) carcinogens, such
that merely handling the stuff is a no-no.
>
> Howard Fast is an author, playwright and screenwriter.
>
HAHAHAHAHAHAHAHAHAHAHA -- Ken, this is the best line in the entire article!
Such credentials! I know, if you make fun of his credentials you will
lose credibility, not he, but if you've got the nerve, you might suggest
that he stick to what he knows best, and leave the scientific thinking to
those who know enough not to pull their facts out of thin air.
The newspaper should have used this tag line:
"Howard Fast is a writer of fiction."
Man, good luck to you. Guys like this need to be shut up before their
arm-waving and screaming creates more damage than it prevents.
Regards
Ray Depew
Integrated Circuits Business Division
Hewlett Packard Co, Fort Collins, Colorado
r...@hpfiqa.fc.hp.com
BSChE BYU '81
MSEE Stanford '88
Carl J Lydick
>In article <1992Sep23....@nuchat.sccsi.com> ke...@nuchat.sccsi.com (Kevin Brown) writes:
>>in the history of the world to reactors of the kind used in the U.S. and
>>Canada? I dare say none have.
>>
>>How many average-case failures have occurred? I dare say more than zero.
>>
>>Since more average-case failures have occurred than worst-case failures,
>>it seems ludicrous to suggest that a worst-case analysis is a better
>>means of comparison than an average-case analysis.
>>
>>And so I reiterate my question: what is this obsession you have with
>>worst-case failure modes?
>
>By this logic, since I've never died and I've lived a relatively
>long time already, and all the bad things I've ever experienced
>were far less than 'worse case' (i.e. dying) I shouldn't worry
>about the possibility that I might someday die, since it has never
>happened in my history and must therefore be so remote as to be a
>ludicrous thing to even consider...
Bullshit. You've presumably seen (or heard of) other people dying.
>Ditto the case for condoms. Since I've got a very low chance of getting
>AIDS (or getting someone pregnant) why worry about it? Ditch that
>condom today!
Again, there are numerous examples of people getting AIDS or getting pregnant.
>The 'average-case' failure of lack of a condom is that
>you have a great time. Once in a while you get some treatable disease.
>Only very rairly would you get AIDS or have a pregnancy. So what is
>this obsession with the worst-case failure modes?
Well, what if I were to tell you that there's some horrible disease, much worse
than AIDS that you might get. Of course, nobody's ever been known to have this
disease, it's just that it's possible to imagine it, and the action of the
disease is plausible. Is the possibility of getting that imagined disease
going to be a deterrent to you?
>The fact is that we havn't had enough reactor-years of experience to
>have had the very low probability modes of failure manifest.
Right. Now, look closely at the circular argument you just made: No matter
HOW MANY reactor-years of experience we have, you can STILL make that claim.
E. Michael Smith
>So: how many worst-case failures, of the kind you refer to, have happened
>in the history of the world to reactors of the kind used in the U.S. and
>Canada? I dare say none have.
>
>How many average-case failures have occurred? I dare say more than zero.
>
>Since more average-case failures have occurred than worst-case failures,
>it seems ludicrous to suggest that a worst-case analysis is a better
>means of comparison than an average-case analysis.
>
>And so I reiterate my question: what is this obsession you have with
>worst-case failure modes?
By this logic, since I've never died and I've lived a relatively
long time already, and all the bad things I've ever experienced
were far less than 'worse case' (i.e. dying) I shouldn't worry
about the possibility that I might someday die, since it has never
happened in my history and must therefore be so remote as to be a
ludicrous thing to even consider...
Ditto the case for condoms. Since I've got a very low chance of getting
AIDS (or getting someone pregnant) why worry about it? Ditch that
condom today! The 'average-case' failure of lack of a condom is that
you have a great time. Once in a while you get some treatable disease.
Only very rairly would you get AIDS or have a pregnancy. So what is
this obsession with the worst-case failure modes? Heck, I've known
couples who tried for YEARS to get pregnant. Who needs condoms!
The fact is that we havn't had enough reactor-years of experience to
have had the very low probability modes of failure manifest. This is
a testimonial to the designs we've used, but does not PROVE the the
failures WILL NOT happen! A one in ten thousand reactor-year failure
ought not to have happened in the first few years of reactor use!
>>Low probability/high consequence
>>events are not necessarily equivalent in riskiness to high probability
>>low consequence events.
>
>By what measure is *this* true? I was under the impression that the
>very *definition* of statistical risk was probability times cost.
See my other posting on the evaluation of risk to sort this one out.
The two of you are using 'risk' differently. One includes the value
placed on the outcome, the other just notes the cost in abstract terms
but does not value the quality of the risk.
>In terms of demonstrating objective correctness, correspondence to human
>intuition of risk is irrelevent. All that matters is that the predictions
>match real-world observation.
This dodges the issue of the value put on a particular magnitude of risk.
'Quantity has a quality all its own' holds here...
>In terms of convincing people who are unable or unwilling to put aside
>their intuition in favor of an objective model, I agree that one must
>show the correspondence between the equation and human intuition.
This again dodges the issue of valuation of a risk. Value is, and
must always remain, a human intuitive process. I've watched BOTH
my parents die from cancer. I would value a quick death from some
other process much more highly than a slow wasting death from cancer.
This VALUATION of the risk of cancer death will change how I value
the risk of extra cancers in the population ... even if the abstract
model predicts equal deaths from cancer and from accident for two
competing scenarios, I'd VALUE them differently. This is NOT irrational,
the is respecting the human condition. This doesn't deny the accuracy
of the analytical model. It does challenge the conclusions drawn from it.
--
E. Michael Smith e...@apple.COM
'Whatever you can do, or dream you can, begin it. Boldness has
genius, power and magic in it.' - Goethe
I am not responsible nor is anyone else. Everything is disclaimed.
victor yodaiken
>>In article <1992Sep24....@nuchat.sccsi.com> ke...@nuchat.sccsi.com (Kevin Brown) writes:
>[...]
>>>While this may be true, I was under the impression that the only types
>>>of severe failure that would be of any concern at all involve breaches
>>>of containment. I would think that this would simplify the problem
>>>space considerably.
>>
>>Not really, small failures can add up to big failures. For example,
>>suppose that a backup power system fails during a SCRAM, the loss
>>of a battery or the improper care of a diesel or the misinstallation
>>of lightning surge protectors .... could all impact on probabilities
>>of big failures.
>
>Okay. Well, let's suppose that the things you're talking about happen.
>What sort of "big failure" are we talking here?
>
>My point is this: I don't see how the kind of scenario you're so concerned
>with can occur without some kind of release of highly radioactive material
>from the reactor. And as far as I know (this could be sheer ignorance on
>my part, of course), the only way such a release can occur is when the
>containment is breached. After all, the *job* of the containment is to
>prevent such a release, yes?
Well, that's the big question isn't it? What if a TMI style
screwup happens at an older plant and when they finally start
cooling it the reactor vessel splits open due to the loss of
ductility associated with nuclear radiation, suppose that the
plant has a weaker containment dome than did TMI (which was
especially strengthened due to the nearby airport), suppose that a
venting valve from containment is stuck open, or that there is a
lightening strike and the operators lose instrument power, or that
someone panics and opens a venting pipe or that the turbine cracks and
sends a large metal fragment through the containment wall ...., suppose
its a BWR plant and the rods won't withdraw, and the cladding catches fire
or that a fuel barge tips over on the cooling water intake or ....
Is the probability of all this equivalent to the probability of a meteor
strike? If so, what's the evidence?
>>>Yes, we *do* pay more attention to accidents with terrible consequences
>>>than those with minor consequences. I do not question that this is so.
>>>I question that this is wise.
>
>I wanted to reiterate this.
>
>We humans have a fascination with the spectacular or the unusual. That
>we have a tendency to emphasize such things in no way indicates that we
>should base our decisions on those tendencies. For some reason, however,
>you seem unable to keep yourself from doing just that, at least with
>respect to nuclear power. Why is it that you seem inconsistent in this
>regard? Why is it that your decisions with respect to transportation
>(to name one example) are not similarly based primarily on the spectacular?
We have to make decisions about what we think are credible scenarios and
what we think are worthwhile risks.
>>>As for your car, what *I'm* saying is that you're probably better off
>>>using the reliability of your headlight system than the reliability of
>>>your brakes as the prime determining factor in the decision of what
>>>form of transportation to use (note, however, that *neither* is probably
>>>a good factor to use in the decision. It's a relative thing).
>>
>>Did you learn to drive in Boston?
>
>No. I learned to drive in Houston. Is there a significant difference?
>
Unfortunately no. Driving in either town is suicidal.
>By the way, I have experienced electrical system failure that took the
>headlights with it before. The danger such a failure presents should not
>be underestimated. It is quite possible that the danger is very close to
>the danger presented by a brake system failure.
Look, you have your risk aversion strategies and I have mine. We are
both given a choice of a broken down car to get from A to B. I
pick the car with working brakes, missing one headlight, and you pick the
car with both headlights and no brakes. I'll even give you a head start,
since I don't want to be anywhere near you on the road.
>
>>>Interesting stuff. I would ask, then: what is the proper metric to use for
>>>deciding which risk model to use (note: there may not be a good or simple
>>>answer to this one, in which case references would be appreciated)?
>>
>>Look at the EPRI book previously referenced. Also see Brittle Power by
>>Lovins for another view of risk.
>
>Okay. These are views, though, correct? Are they considered definitive?
>
>If not, then please direct me to something that *is* considered definitive
>(by the scientific community). If no such thing exists, then I must
>conclude that there *is* no single proper metric to use.
The EPRI work reflects the mainstream of the scientific community, there is
no definitive work in this relatively new and contentious field.
>>>>deaths over 3 days embedded in 10 death free years.
>>>>I disagree. Who's being objective?
>>>
>>>Whether or not these are "equivalent" depends on how you're measuring them.
>>>In terms of the numbers alone, they are quite obviously equivalent. Dead
>>>is dead, no matter how it's concentrated.
>>
>>Right, but we can easily tag on other numbers which expose the difference.
>>Suppose that there are 100 people living in my village and that we
>>chose between the risk of losing 20 people in one disaster and losing
>>20 people over 10 years. Clearly, the village itself may not be able to
>>survive a sudden loss of 1/5 of the population, but it may be able to
>>cope with a longer term loss.
>
>Change the data the model has to work with, and you change the model. What
>you've done above is provide additional information to work with. So, with
>the additional information, I agree that the two cases (losing 20 people in
>one disaster, with no deaths during the remaiing time and losing 20 people
>over a period of 10 years) are not equivalent.
So, expected deaths from fossil induced lung diseases cannot be directly
compared to expected deaths from a nuclear melt-down without some
additional explanation.
>If you're trying to say that our decisions must account for the specific
>circumstances surrounding them, you'll get no argument from me. However,
>the discussion of nuclear power versus coal power has thus far been one
>without such detail. It has thus far been about the "general case". As
>such, there is no such detail of the type you provide above, and as such
>the models will necessarily seem somewhat simplistic.
I have not been arguing that nuclear is worse than coal, I've been arguing
that simplistic justifications of nuclear will not fly.
>>The real world is very complex, and we cannot
>>hope to capture it all in any formula.
>
>Ah, but this was not my intent. Rather, my intent is to capture the basic
>essence of the real world, and to do so with as little reliance on my own
>value set as possible.
Of course your value set determines what, in this issue, you find to be
essential.
>>Your choice to value what your
>>model can calculate over the subjective, but real, phenonemna which your
>>model ignores, is not an objective choice.
>
>Come again? You think it's not an objective decision to place less
>significance on that which you cannot determine objectively?
Yes, there is no objective reason to prefer what can be easily
quantified over what cannot be easily quantified. The importance of
real phenomena does not correlate well with the ease with which we can
construct mathematical models of them.
>This may be. Ultimately, the desire to attempt to model the real world as
>accurately as possible is a value judgement. However, I would argue that
>at the very least it is a *useful* value judgement (which, one may argue,
>is in itself a subjective term).
We ask about the relative advantages of N vs. F. You capture some of the
relative advantages via a formula which I suggest ignores some key
criteria. For you to insist on ignoring those criteria is not a sign of
objectivity.
This is not a correct assumption. As a simple example, suppose, that
I argue that the reliability of computer programs can be acertained
by dividing the number of lines of code by the number of variables. You
object, and argue that such subjective considerations as one's faith in
the diligence of the programmers and the novelty of the methods used are
critical indicators. Am I being more objective, more accurate, more
scientific than you are?
>Yes, it *is* the best we can do for many issues. The problem is that, at
>least with respect to the issue of nuclear power, it *isn't* the best we
>can do. In the case of nuclear power as an alternative to coal power, we
>have behavioral models and cost/benefit models which have real-world data
>to support them, and that means that, if we choose to base our decision on
>them, we can do better than chance in getting the results we want. Basing
>our decision purely on subjective judgements, however, is effectively the
>same thing as relying on chance to make the decision.
Of course, even accepting your argument, we still have to decide which of the
conflicting cost/benefit analysis we wish to accept.
--
John De Armond
Damn, damn, yackadam. You made my day. I've never seed a better case
of the pot calling the kettle black than YOU, of all people, accusing
Carl of making content-free, sophmoric postings.
Only thing I want to know, yackadam, and that is how you do it. I mean, I'm
semi-retired and I don't have time to create half the noise you do. Even
allowing for your firing out your ass instead of researching your posts,
I don't see how you enter that much text. My fingers would be aching at
half the daily volume.
Oh... I see... You are really several people working for EPRI or GPU or
someone who is trying to make the anti-nook-kooks look like, well, kooks.
Damn if you ain't doin' a FINE job, whomever you are.
John
--
John De Armond, WD4OQC |Interested in high performance mobility?
Performance Engineering Magazine (TM) |
Marietta, Ga |Interested in high tech and computers?
j...@dixie.com |Write me about PE Magazine
Need Usenet public Access in Atlanta? Write Me for info on Dixie.com.
John De Armond
>>My point is this: I don't see how the kind of scenario you're so concerned
>>with can occur without some kind of release of highly radioactive material
>>from the reactor. And as far as I know (this could be sheer ignorance on
>>my part, of course), the only way such a release can occur is when the
>>containment is breached. After all, the *job* of the containment is to
>>prevent such a release, yes?
Let's take a look at what our resident armchair reactor engineer, yackadamn
proposes:
>Well, that's the big question isn't it? What if a TMI style
>screwup happens at an older plant and when they finally start
>cooling it the reactor vessel splits open due to the loss of
>ductility associated with nuclear radiation,
If you're going to walk the walk, at least talk the talk. The theoretical
problem is called Neutron-Enhanced Nil Ductility Elevation or simply
Neutron embrittlement. Not a problem, of course, since a) all reactors are
now charged with fuel enrichments that minimizes neutron irradiation of
the pot and b) since the problem has been identified, all cooling water
including the passive accumulator cooling system is heated at all times
to a temperature above the theoretical worst case nil ductility point. End
of problem. As long as we're postulating, we can also postulate that the
same core melt that would set up the conditions for nil ductility failure
would also heat the pot hot enough to anneal out the neutron induced defects.
As long as we're postulating.
But let's assume for the sake of argument that the pot breaks. What then?
Well either the bottom of the pot falls off or it stays partially in place.
If it falls off, it falls into the reactor sump and is cooled by the couple
of million gallons of water present. If it stays in place, it is cooled
by the safety injection system, HPSI, LPSI, passive accumulator, concensate
cooling system, and/or river water system. Yeah, we really do think
of these kinds of things when we design the multiply redundant safety
systems.
>suppose that the
>plant has a weaker containment dome than did TMI (which was
>especially strengthened due to the nearby airport),
No it wasn't. All containment buildings must withstand the impact of
the largest known aircraft that would be flying in the area. Most all
FSARs I've read use the 747 as the prototypical crasher.
>suppose that a
>venting valve from containment is stuck open,
Then the operator would simply close one of the other redundant valves
in the train. All containment penetrations have two valves inside and
two valves outside, all in series and all designed to fail closed.
>or that there is a
>lightening strike and the operators lose instrument power,
Then the PWR plant would cool on natural circulation until power was restored.
Natural circ cooling is a test performed ALL PWRs. BWRs are cooled by
first the HPSI and then the LPSI pumps. Both are steam driven turbine
pumps that require no outside power to function. I startup tested these
pumps at Browns Ferry NP, Unit III. One of the tests performed is to
rack out the control panel power breakers and verify the pumps operate
normally and that all valves fail in the proper direction. Note too
that all components including the heat sink are located inside the secondary
containment building so containment isolation does not affect their
operation. (I've saved the definitions of the acronyms for later in
order to watch Yackadamn stumble over these new-to-him terms.)
>or that
>someone panics and opens a venting pipe
Then the SRO, the shift supervisor or the STA, all of which are licensed and
all of which are required to be in the control room at all times,
would override the operator and close the valve. Since only the shift
supervisor, with the agreement of either an SRO or STA can override
a written procedure, "someone panics" would not happen. If a sudden
fit of insanity should overtake one of the operators, the armed guard
posted at the entrance to the control room would remove him.
>or that the turbine cracks and
>sends a large metal fragment through the containment wall
A massive turbine failure with half the largest wheel escaping at best
velocity and penetration angle is a design-basis accident that is planned for.
Half the wheel is used because if a fragment constitutes more than half,
it won't be able to escape the shaft. No credit is taken in the analysis
for the fragment penetrating the turbine casing (about 6" thick alloy
steel), the turbine building or aero drag during its trip. Gee, Yack,
maybe you outta actually READ an FSAR sometime.
In other words, the turbine cannot penetrate the containment.
...., suppose
>its(sic) a BWR plant and the rods won't withdraw,
BWR rods insert from the bottom for shutdown, therefore your pretend is
irrelevant. Assuming you knew enough to correctly phrase the hypothetical,
since each rod and associated drive in a BWR is a separate system, with
banks of drives located in at least 4 physically separated areas, and
since the rod insertion is a passive process driven by reactor pressure,
"rods" can't fail to insert. A rod MIGHT. "Rod Ejection" accidents,
as these events are known as, are planned for.
But let's pretend that something could happen to cause a massive failure
to insert. The RO would simply lift the cover on the Standby Liquid
Control system activation button and press it. This would fire explosive-
activated valves and start up a DC motor driven pump, driven from
batteries in the same room, that injects a concentrated boric acid solution
into the reactor to shut it down. If that failed, he'd simply lift
the cover on one of the other three of the quadruply redundant trains
and fire it. If all of those failed, he'd just sit back and let HPSI
do its thing. As soon as the pressure drops in the reactor with the recirc
pumps off, steam voiding automatically shuts down the nuclear reaction.
The operator might occasionally trim the pressure control to keep the
cladding below the ignition point but other than that, not much to do.
>and the cladding catches fire
The cladding burned at TMI. HoHum. Oh, maybe a few thousand people we
somehow missed died because of this.
>or that a fuel barge tips over on the cooling water intake or ....
All plants are required to impound enough cooling water inside the compound
to achieve cold shutdown. Assuming by "fuel" you mean diesel, the fish
might not like it.
>Is the probability of all this equivalent to the probability of a meteor
>strike? If so, what's the evidence?
Well.. Let's see. Meteors strike the earth every day. None of your
fantasy has ever happened so one can step out on a limb and say
that the probability of your fantasy is a bit less than a meteor strike.
>>regard? Why is it that your decisions with respect to transportation
>>(to name one example) are not similarly based primarily on the spectacular?
>We have to make decisions about what we think are credible scenarios and
>what we think are worthwhile risks.
In the venerable words of Tonto, "What's this WE shit, white man?"
[The rest of Yackadamn's interminable anal discharge mercifully deleted]
Yack, you've again ventured out of your normal demagoguery and have tried
to act like you know something about nuclear power. As usual you failed
miserably. Maybe you ought to stick to safe areas like telling us how
you quake in the night from your fear of nuclear power.
victor yodaiken
>yoda...@chelm.cs.umass.edu (victor yodaiken) writes:
>>Well, that's the big question isn't it? What if a TMI style
>>screwup happens at an older plant and when they finally start
>>cooling it the reactor vessel splits open due to the loss of
>>ductility associated with nuclear radiation,
>
>If you're going to walk the walk, at least talk the talk. The theoretical
>problem is called Neutron-Enhanced Nil Ductility Elevation or simply
>Neutron embrittlement. Not a problem, of course, since a) all reactors are
>now charged with fuel enrichments that minimizes neutron irradiation of
>the pot and b) since the problem has been identified, all cooling water
>including the passive accumulator cooling system is heated at all times
>to a temperature above the theoretical worst case nil ductility point. End
>of problem. As long as we're postulating, we can also postulate that the
Every time postings by some of the smart folks working on nuclear power
start to convince me that the technology has potential, you or one of your
equally rational colleagues pops up. Your point is nonsense, despite the
long words you seem so proud of. Anyone wanting technical information
on this issue could consult the record of the NRC hearings which closed
Yankee Rowe precisely because of concerns about loss of ductility.
The rest of your post is similarly well constructed.
--
victor yodaiken
>yoda...@chelm.cs.umass.edu (victor yodaiken) writes:
>semi-retired and I don't have time to create half the noise you do. Even
I don't have to move my lips when I read. That may explain the disparity.
--
Gary Coffman
>
>It seems obvious to me that the risks posed by coal are of a different
>kind than the risks posed by nuclear fission. Coal has many obvious
>problems, but given current technology, no coal plant can cause the
>same level of devastation that we see at Chernobyl (unless we want to
>take the improbabilities of statistical mechanics into account and
>thus render the whole discussion futile).
On the other hand, most of us consider the *routine* risks offered by
coal and oil to be of a different type causing significantly more *proven*
damage than that of nuclear. Even including Chernobyl, which all of us
agree was an unsafe design, the global effects of the *routine* pollution
caused by coal and oil are still greater than the effects of nuclear power
operations *and* accidents over the last 30 years. If you add in mining
hazards such as roof falls and black lung, transportation accidents, refinery
explosions, coal yard fires, steam plant explosions, and the like, the
balance tips even further in favor of nuclear power. It's true that an
absolute worst case nuclear accident involving sudden breach of primary
and secondary containment followed by a fire for dispersal would have
a regional effect worse than any *single* coal plant accident, but coal
plants *don't need accidents* to do severe harm to the environment, they
do it *routinely*.
>Since there appears no pressing need for immediate construction of
>new generating capacity, my belief is that it pays to be cautious here.
>I'd rather spend money on solar, conservation, insulation, natural
>gas, fancy pollution control equipment .... For detailed descriptions
>of alternate energy paths see proposals by the UCS, EPRI, and the
>Rocky Mountain Institute. In these you will find in depth analysis of
>comparative costs.
While it's not urgent to increase *net* generating capacity in the
US or Europe in the next few years, it's essential to replace aging
plant on an ongoing basis over the next few decades. *And* the third
world is rightfully demanding access to electrical power. I don't want
to see any more coal plants built. And I don't want to see the busbar
price of electricity increase dramatically. Conservation and insulation
are fine, they *slow* the rate of increase for new *net* capacity in
already industrialized nations, but they don't address the issues of
replacing aging plant or supplying capacity to emerging nations. Solar
and indirect solar are significantly higher cost sources of electricity
than coal or nuclear. Their use would raise busbar prices sharply.
Natural gas is only a short term solution, and it still emits CO2
and NOx pollutants as a *routine* part of it's operation. It *is*
the current best alternative to nuclear however. If we pin our hopes
on a clean and cost effective replacement for natural gas generation
being developed in the next 50 years, we may be alright, but that's
a mighty big if on which to gamble the future of industrial civilization.
We have better uses for natural gas than generating electricity. There
aren't many better uses for nuclear power than generating electricity.
Gary
Snide Pucker
#>>Did you learn to drive in Boston?
#>
#>No. I learned to drive in Houston. Is there a significant difference?
HAHAHAHAHAHAHAHAHA---*knowing wink*
Actually, there isn't that much difference with having a car in
Boston as it is in Houston as long as you
--don't live near Boston (insurance)
--don't drive there
--don't park there
Other than that, they're pretty much the same =)
--
---
Jerry W. Miller # "Opinions? Of course they
mil...@gn.ecn.purdue.edu # are mine, do YOU want them?"
Purdue Univ. School of Nuclear Engg//Argonne Nat'l Lab--Reactor Engg
Snide Pucker
Just taming some of the assertions about nuclear accident mitigation...
In article <ba_...@dixie.com> j...@dixie.com (John De Armond) writes:
#>yoda...@chelm.cs.umass.edu (victor yodaiken) writes:
#>
#>Well either the bottom of the pot falls off or it stays partially in place.
#>If it falls off, it falls into the reactor sump and is cooled by the couple
#>of million gallons of water present. If it stays in place, it is cooled
#>by the safety injection system, HPSI, LPSI, passive accumulator, concensate
#>cooling system, and/or river water system. Yeah, we really do think
#>of these kinds of things when we design the multiply redundant safety
#>systems.
Errr....actually, few scenarios envision anything more than a partial
cooling of any corium with the sump water. Water in the sump can actually
be a bad thing if enough corium is ejected, in which case you must worry
about the pressure loading on the containment due to rapid vapor expansion
(the "vapor explosion"). I believe the failure of the entire vessel
(cracking in half, etc.) is very^n small at operating (or higher) temps;
this should only be a problem during cold starts, refueling, etc. when
the vessel is cold (and more brittle).
A problem with the entire ECCS methodology is that for serious accidents
(large-break LOCAs and the like), each cooling system must function
properly to cool the core completely. Any failure may cause core damage.
Each system has its own failure mode, so a disastrous malfunctioning of
the ECCS system requires something only a human could do, shut it off
(cf. TMI).
This isn't the same, or nearly as foolproof, as passive cooling techniques
used in BWR MarkII (or its derivatives) or any advanced LWR (pick one).
#>Then the operator would simply close one of the other redundant valves
#>in the train. All containment penetrations have two valves inside and
#>two valves outside, all in series and all designed to fail closed.
All references to TMI's PORV ignored...
#>>or that there is a
#>>lightening strike and the operators lose instrument power,
#>
#>Then the PWR plant would cool on natural circulation until power was restored.
#>Natural circ cooling is a test performed ALL PWRs.
...but has not proven to be 100% effective even for decay heat...
#>Note too
#>that all components including the heat sink are located inside the secondary
#>containment building so containment isolation does not affect their
#>operation.
Which is exactly why a containment breach is required for a nuclear
accident to emit large quantities of radioactive material.
#>>or that
#>>someone panics and opens a venting pipe
#>
#>Then the SRO, the shift supervisor or the STA, all of which are licensed and
#>all of which are required to be in the control room at all times,
#>would override the operator and close the valve. Since only the shift
#>supervisor, with the agreement of either an SRO or STA can override
#>a written procedure, "someone panics" would not happen.
..., but no one's permission is required to hit the wrong button.
#>But let's pretend that something could happen to cause a massive failure
#>to insert. [explanation of possible scenario actions]
#>The operator might occasionally trim the pressure control to keep the
#>cladding below the ignition point but other than that, not much to do.
All of which assumes that his instrumentation readings are fully
functional, necessary components aren't being maintenanced, the
situation is correctly assessed, information is properly exchanged
between coworkers, etc. The WORST possible accident, imho, ISN'T
a _material_ or _component_ failure, but a _human_ or an _instrument_
failure. PWR ECCS is useless unless it is operational (neglecting the
Accumulator Injection System). Backup control rod injection methods
aren't very helpful if the operators think all of the rods are in.
The list goes on (and did at TMI).
The moral of the story is that things _can_ go wrong with nuclear
reactors. Fortunately, there has been a significant amount of time,
effort, and money spent on mitigating every conceivable accident
scenario currently considered possible. This is not to say that
nuclear reactors can never fail completely, but that the likelihood
is remotely small that it is a worthwhile risk compared to the
(relative) benefits of nuclear power.
For nuclear reactors to NEVER fail would require definitive material
behavior and response over the lifetime of the component/instrument
(that is currently unavailable) and a design that acknowledged this.
Mike Van Pelt
>*Something* has to be used to generate that power.
That's where you are failing to comprehend the true genius of the "soft
energy paths" people. It is possible to generate infinite energy by
conservation, with no energy generation at all. That is what has made
Bangladesh is the richest country on Earth.
(And if you believe that one, I've got a carburetor that gets 200
miles per gallon of water that I'll sell you...)
--
Mike Van Pelt Here lies a Technophobe,
LSI Logic/Headland Products No whimper, no blast.
sun!indetech!hsv3!mvp His life's goal accomplished,
m...@hsv3.lsil.com Zero risk at last.
Mike Van Pelt
>>Did you learn to drive in Boston?
>
>No. I learned to drive in Houston. Is there a significant difference?
I lived in Houston for two years, and have had to drive in Boston
on a couple of business trips. I've driven in San Francisco and
Los Angeles. Boston is the worst I've seen anywhere.
I've enjoyed your responses to Victor. Keep up the good work!
victor yodaiken
>In article <1992Sep26.1...@nuchat.sccsi.com> ke...@nuchat.sccsi.com (Kevin Brown) writes:
>>The need for power generation is there.
>>*Something* has to be used to generate that power.
>
>That's where you are failing to comprehend the true genius of the "soft
>energy paths" people. It is possible to generate infinite energy by
>conservation, with no energy generation at all. That is what has made
>Bangladesh is the richest country on Earth.
>
>(And if you believe that one, I've got a carburetor that gets 200
>miles per gallon of water that I'll sell you...)
What I'd like to believe is that you are eventually going to generate
either the apology or the reference that you owe me. Perhaps you've
forgotten your promise already? I may have the old posts on disk if you'd
like a reminder.
--
Kenneth Ng
:invalidate subjective concerns such as doubts about the efficacy of
:the NRC or doubts about whether probability estimates are correct
:within an order of magnitude. Note that the AEC's own review of the
:Wash-1400 report produced some predictions of a major accident every
:50 years of 200 reactor operation (or something like that, perhaps even
:more).
Um, if you are refering to the Lewis Report, it was produced by the NRC,
not the AEC. It was chaired by H. W. Lewis, NUREG/CR-400, 1978. Their
principle findings were that the error bounds were understated, and they were
critical of the executive summary. They also stated that WASH-1400 was the
most exhaustive and thorough study done to date. Lewis himself has testified
in Congress that his personal belief is that WASH-1400 overstated the actual
risks, that the real probablilties were much lower. (Unfortunately I am
unable to find the actual testimony, does anyone know when he testified?)
--
Kenneth Ng
Please reply to k...@eies2.njit.edu for now.
Apple and AT&T lawsuits: Just say NO!
E. Michael Smith
>Funnily enough that argument strengthens the case for nuclear power!
>If you argue losing 20 people in a single incident is worse than
>losing 20 people statistically over a number of years, then almost all
>nuclear plant accident scenarios are preferred as even the worst cases
>involve mostly statistical risk to a population - I can think of
>no credible scenario where a large population would receive a lethal
>dose from a nuclear plant.
I'd like to support the point that this does seem to be the case.
Especially in the context of things like dam failures, the 'worst
case' of 'concentrated loss' is likely to be low from nuclear when
compared to other systems.
One caviate, though, is that a Chernobyl type failure (which is
admitedly not possible in US designs, though there may be
equivalent modes of failure we don't know about...) can cause great
concentration of grief in a local population. That is, the genetics
of an entire extended family structure can be screwed up by that kind
of release with the attendent lack of evacuation. There is more to
concentrated loss than just death ...
Just as unfortunately, a major toxic spill can be even worse than a
major nuke accident. I've seen a hazard map of the Silicon Valley
that showed the likely death/illness zones around known tankage of
'nasty stuff'. It scared the bigeezus out of me. There was literally
no way I could drive the 15 miles home after an earthquake without
passing through a hazard zone... I now own a gas mask that is in
my 'quake kit ...
Note that I do NOT think this argues for more nuclear, I think it
argues for less toxics in better tankage ... Consider Bohpal (sp?) ...
>In contrast, fossil fuel plant accidents typically
>involve 10-100 deaths on an occasional basis, including transport
>of fuel, mining and waste disposal - with waste disposal arguably
>the historically most hazardous. This is in _addition_ to the
>statistical risk that fossil plants pose by carcinogen emission,
>SO2 emission and potential greenhouse effects.
I thought that there were natural gas explosions projected for CNG
storage sites that involved predictions of thousands of casualties?
Or were these for the (urban) sites that got scrubbed?
Don't coal dust explosions take out a hundred or so miners at a time
every decade or so? (And often with the deaths concentrated in a
single family, since siblings and parents often work the same mine).
>If you dislike low probability scenarios with high potential costs
>then just the potential greenhouse effects should rule out fossil
>fuel plants.
Interesting point ... that also supports a solar path ... (and
supports a 'use a little of each and not alot of any' path as well.)
E. Michael Smith
>Well, having a set of standard designs rather than every unit a custom design
>has been shown to allow better quality control in many other industries. Seems
>like it ought to help in the nuclear power industry, too.
I'd like to ask a bit of indulgence here: What would be your 'dream design'
for this purpose? Would it be a LMFB or a HTGR or an 'inherently save LWR'
or what?
If the industry were given a mandate to come up with THE reactor design
(or maybe two, if there were a synergistic set where one bred and the
other generated, for example) for the next 50 years, what would it be?
I can't help but believe that it wouldn't be what we have today, but what
would it be? Lets forget the past for a few minutes and look into the
future. Show me that the future is one I can believe in!
E. Michael Smith
>All of this is basically irrelevent, however, with respect to the statement
>about the public. The problem is that the public treats nuclear power
>*differently* than it does other forms of power.
This is the result of the public feeling lied to. When someone has
sold you a crock on one occasion, you make 'em work a bit harder before
you part with your money on the next occasion.
You may not like the fact that 'too cheap to meter' and 'no problems
with waste disposal' were touted before you were in the industry. You
may think they are interesting ancient history. You may even think
that the incidents at {Rocky Flats, Savanna River, (That place in
Washington... Hanover?), TMI, Chernobyl, Windscale, ...} arn't relevant
since they happened in the (distant?) past or not in 'the nuclear power
industry' or not in our country or not with our reactor designs or ...
But the public sees a single combined nuclear industry painted with one
brush that has been less than diligent in the past about cleaning up
their trash and telling the truth.
What do you do when you kid lies to you and makes messes? You put him
in 'time out' and then watch 'em real close when he comes out. You let
him know that it will be A LONG TIME before credibility is restored.
You 'treat him differently'.
>The public's view of
>nuclear power seems to be much more negative than their view of other forms
>of power, even though nuclear power isn't *objectively* much worse (if any)
>than any other form of power generation (except *possibly* in the worst
>case, and even *that* is debatable, inasmuch as the worst case scenario
>for other forms of power generation is rarely discussed, and thus not
>clearly defined).
It is about on a scale with their (ever increasing) concern over the
handling of chemically toxic materials. Heck, look at their concern
over chemical explosives. Just go try to buy a box at the local
hardware store like in the good 'ol days ...
>Thus, the public's largely negative view of nuclear
>power is illogical, because it is subjective and highly biased.
BZZZT. Wrong. It may well be highly biased (most likely due to
historical reasons), and it may well be subjective, but that
DOES NOT mean it is illogical.
You have a highly toxic, highly explosive material that it is not
possible for you to easily track and control (as a random civilian)
being used by highly monied interests to make more money in facilities
that have been shown to be shodily run in the past. These interests
have often demonstrated more than a bit of disreguard for The Truth
and for the well being of the random Joe Citizen. You expect the
public to NOT have a largely negative view?
Note that NONE of the 'complaints' I've listed above are with the
technology of nuclear power plants. Most are with the organizations
that run THE nuclear industry and with the nature of the material
itself. (And no, I'm not advocating that reactor fuel is explosive,
only that in THE nuclear industry there are some forms of fuel that
can be used to make bombs and some that can be used to make fuel
rods and that the two functions are not required to be technically
separate ... especially in a breeder/reprocessing cycle...)
>I would argue that most people use the average case as the primary metric
>of comparison of alternatives for most of the decisions they make.
Until it is shown that the worst case is the right one to use ...
I won't knowingly buy a house in a flood plain.
I won't knowingly live near a chemical storage or disposal site.
I won't knowingly live right below a hydro dam.
The 'worst case' is just not acceptable, even though the average
case is ...
I buy insurance for things that have acceptable average DOLLAR losses,
but where the 'burstiness' of the cost is un-acceptable.
>That
>they seem not to do this with respect to nuclear power indicates to me an
>illogical bias.
That they do this with respect to nuclear power indicates to me that
they consider it on a par with chemical hazards ... and act accordingly
in a perfectly logical way. As to why so many folks live below dams and
in flood plains, I can only guess that some folks would be willing to live
next to nuclear plants too ...
>Since you have admitted that the risk assessment of a power-generating
>method (be it nuclear, coal, or whatever) is a subjective thing (inasmuch
>as it's not an *objective* thing), and since your objections to nuclear
>power seem to be based *solely* on your view of the risks of using nuclear
>power (for some reason, the worst-case scenario with respect to nuclear
>power causes you great concern, yet you clearly do not treat the worst-case
>scenario with respect to your car with the same concern, else you would
>reach the same conclusion about driving that you reach about nuclear
>power),
This is an interesting point. One is far more likely to die from a
car accident than from any 'credible' nuclear accident. Even the
scale of nuclear accidents isn't much different from major airliner
crashes. (What's the record, 1000 dead?)
Consider another: The risk of dying from cancer is (roughly)
distributed as 35% from lung cancer mostly attributable to smoking,
35% from colon cancer mostly attributable to HIGH FAT DIETS, and
the remainder from other causes. Thats right, the risk of the
typical American High Fat diet is about the same as smoking.
Vegetarians are MUCH less prone to death from colon cancer.
(like, single digits ... I've forgotten exactly what, but somewere
around 2% to 5% vs 35% ...)
So why does the average person get worried about cancer from smoking,
but discuss it over a Burger, Fries, Shake, ...
For me, finding this information has led to a great increase in my
plant intake and a great decrease in my fat intake ... but not as
much as I'd expected and even less for other folks I've told. There
is SOMETHING to be said for familiarity leading to denial and
unfamiliarity leading to suspicion ...
>it follows that your rejection of nuclear power is entirely
>subjective in nature, i.e. a mere opinion with no objective grounding.
>
>That being the case, why should we take your opinion seriously?
Because 'subjective' is NOT a dirty word, and opinions indicate
underlying causes that will determine how the nuclear industry
fares in the future. If you DON'T take the average Joe Citizens
opinion seriously, you are likely to be faced with a continued
stagnant industry since you won't know what concerns you need
to address in future designs and/or presentations...
E. Michael Smith
>In article <53...@dime.cs.umass.edu> yoda...@chelm.cs.umass.edu (victor yodaiken) writes:
>>No. Please excuse me if I opt out of further elaboration of this issue with
>>you, as you seem to lack a committment to rational discussion.
>
>Inasmuch as you've admitted yourself that the risk analysis is basically a
>subjective thing, I find it rather humourous that you're bailing out
>because the discussion isn't "rational"...
Excuse me, but irrational .NE. subjective. One can be both subjective
and rational. Infact, much of lifes problems are solved by subjective
rational thought... and most of it's creativity.
Or were you proposing that Leonardo Da Vinci's designs were totally
objective? That Bach was totally objective? Or that both were
irrational?
Please lose this notion that to be subjective is to be irrational ...
E. Michael Smith
>In article <53...@dime.cs.umass.edu> yoda...@chelm.cs.umass.edu (victor yodaiken) writes:
>So? Unless you have some personal grudge against nuclear power, you should
>find the worst-case failure of *most* power systems to be worrisome.
What is the worst case failure mode of solar or wind? Maybe it is
self delusion, but I can't think of anything that even comes close
to a core explosion and containment breach upwind of a population center...
(as unlikely as that is ...)
>That
>being the case, what makes nuclear so special that you obviously devote so
>much extra attention to its worst-case failure mode?
Mostly that fact that the adherents of nuclear like to poo-poo the risk
and act as though it wasn't real and they didn't really care about it.
If the pro-nuke side were saying things like: "That is a very real
concern and we have done everything possible to address it, but it
still worries us a bit, so we're still working on making it better" rather
than "It doesn't make sense to spend more time on THAT irrelevant bit
of paranoia, who cares. We've done what we were REQUIRED to do... shut
up and sit down 'cause WE know better what's good for you." I'd be much
less interested and would instead be asking more questions about storage
of toxic chemicals ...
>It wouldn't be because you happen not to *like* nuclear power, would it?
No, it wouldn't. I suppose you'd have to call me a 'closet nuke supporter'
(gads, am I, then, 'comming out of the closet'? ;-) in that I think the
technology has great promise, and I DO wish it were being done right.
I'd really like to be able to support it. But I can't. (Mostly on
the grounds that the management isn't very good and the attitudes need
a tune-up ... with a little worry about the once-through fuel cycle
squandering a valuable resource while causing excessive amounts of mining
and lack of care in quaranteening nuclear materials in the past).
I really admire some of the advanced reactor designs, and would like
to see more of them being used in sizes scaled up to production rather
than just blindly building more of the '60s teakettles we've got now.
>>You claim that n deaths distributed over 10 years are equivalent to n
>>deaths over 3 days embedded in 10 death free years.
>>I disagree. Who's being objective?
>
>Whether or not these are "equivalent" depends on how you're measuring them.
>In terms of the numbers alone, they are quite obviously equivalent. Dead
>is dead, no matter how it's concentrated.
Ah, but lets change the span a little. You have a village of 1000.
What is the difference if 1000 die in a single year followed by 99
death free years, or if 10 die per year over 100 years? In the first
case you have a case of genocide followed by a village empty for 99 years.
In the second you have a normal functioning society with long lives and
normal death in old age.
I don't think this example is really a reflection of what nuclear could
do to a nearby village (I can't envision a way to kill 1000 in a
single year, others may be able to come up with one). But it is offered
only to illustrate the case that two identical numeric quantities have
dramatically different human terms, once the details are unmasked.
You may argue, with some cause, that this type of failure cannot be
caused by nuclear, but to argue that there is no difference in the
two outcomes since 'dead is dead' and 1000=1000 is only to illustrate
a gaping void of perception ...
>Since the danger ("risk") presented by each type of plant cannot be assessed
>in an unbiased and objective fashion, but the benefits and costs *can* be
>assessed in an objective and unbiased fashion (reasoning: we have large
>amounts of data on the amount of power produced by each type of plant, the
>operating expenses for each, the building costs for each, and the pollutants
>produced by each), and since the real world doesn't give a damn about how we
>feel about things, I argue that we would be better off, in real-world terms,
>making the decision *primarily* based on the objective cost/benefit
>evaluation and only *secondarily* based on the subjective risk evaluation.
Ah, so if it is cost effective to commit genocide and kill the village,
we should, but if it isn't clearly cost effective, then and only then
should we think about the subjective value of killing the village ...
While this is an 'absurdum ad reductum' argument, it does illustrate
the point that numbers cannot be trusted in isolation from human values.
'Common sense' and 'sanity checks' must be 'in the loop' ...
Show that the village will not be killed. Show the credible rates of
death are simlarly distributed for nuclear and coal (maybe nuclear is
more 'bursty' than coal, but not by enough to change the community
lifecycle?) But don't try to argue that ANY degree of concentration
of life loss is not relevant. That point cannot be conceeded, since
there is a 'thought experiment' that shows it false...
If you can show that the distribution of deaths would not be
statistically significantly different TO THE COMMUNITY and TO THE
FAMILIES for nuclear and coal, (and ditto for the miserable diseases
that don't kill you), then you have won the case. Until then you
will have a skeptical audience thinking about the devistation of
families in the Nazi era, in American Indian villages, in Armenia,
in Kurdistan, in Bohpal, in ...
>If the judgement of the danger presented by the various power-plant options
>is a subjective one, why is it that the anti-nuclear people insist on
>concentrating on that judgement?
Because they don't trust the particular subjective slant that the
pro-nukers manifest when they discuss their evaluation of that danger...
especially the dogged insistance that it is mathmatically objective
and that they are stupid for not believing the Nuclear Priests who
have shown them the numbers even though the priests know that they
cannot comprehend ...
Nothing gets an IRS auditor interested more than someone saying "I've
got it all figured just exactly right, trust me, no need for you to
audit MY return, I'll just be going now." ...