India Nuclear News And Discussion

[arnab]

Sir since you are so much in love with yourself - why not stand in front of a mirror and repeat what you write about yourself on BRF to yourself? This will help us all instead of steadily weakening the signal to noise ratio on the forum with pointless fearmongering. You are not canvassing for votes here you know

Ah but my speaking here causes so much irritation to known charlatans and purveyors of blatant falsehood that they jump in and expose themselves very thoroughly, without much effort.

So you see there is good in repeating a basic truth over and over again.

er..like what? basic truths like LWR is inferior PWR?. This has to be in the level of your assertion that smoothbore is better than rifled barrel because all western tanks uses it

[Sanku]

er..like what? basic truths like LWR is inferior PWR?. This has to be in the level of your assertion that smoothbore is better than rifled barrel because all western tanks uses it

You know, I am surprised at considerable inability shown by some to get even a single point right without either tripping over basics or bringing in data points which exist in some parallel multiverse.

On the bore thingy ALL (not only western tanks guns are smooth bore, including eastern, northern and penguin tanks)


Also my statement was "The HESH advantage of rifled is not significant enough to trade off for design complexities for stablizing firing AFSPDS and commonality with rest of tanks"

But yes, that was probably too complicated for consumption of some folks here, who can only faithfully repeat what is written in some press handout by the shrine where they go for their "matha tek"

[abhishek_sharma]

Can you tell me how you'll factor in cost of a an yet occur nuclear accident that may or may not happen some time in the future. The severity of a nuclear accident could be from anywhere near this interesting compilation of nuclear incidents in India to a Cherynobl type of mega disaster.

You first raised the point of quantifying everything. Yet you still seem to not have an idea of how one can do that.

Well, obviously we have to use historical data for finding the probabilities. The probabilities of earthquakes and tsunamis can be estimated by scientists. Similarly, the probability that a component could fail is studied in reliability engineering. Given a reactor, they should tell us the probability that it would fail in X years.

I will provide a toy example. I am sure many members here can improve it.

Let us assume that we are told that given a new nuclear plant, we have the following probabilities:

No problems in 80 years: Probability 0.85 --> Liability = 0
1 Small accident in 80 years: Probability 0.13 --> Liability = 10
1 Big accident in 80 years: Probability 0.02 --> Liability = 1000

Total expected liability = 0*0.85 + 10* 0.13 + 1000* 0.02 = 21.3
Total electricity generated in 80 years: 1000000 units
Liability surcharge per unit electricity: 21.3/1000000

True cost of electricity = Cost of generation + Liability surcharge

This is, of course, based on class 10 level of Mathematics. People here will tell you there are opportunity costs and it should be discounted with time etc etc.

[arnab]

er..like what? basic truths like LWR is inferior PWR?. This has to be in the level of your assertion that smoothbore is better than rifled barrel because all western tanks uses it

On the bore thingy ALL (not only western tanks guns are smooth bore, including eastern, northern and penguin tanks)


Also my statement was "The HESH advantage of rifled is not significant enough to trade off for design complexities for stablizing firing AFSPDS and commonality with rest of tanks"

OT but Chally has rifled barrel and I think the 'commonality' argument was proven to be complete BS. and 'design complexities' is basically handwaving

[somnath]

Abhishek-ji, Soemwhat simlar, thought hugely more intrcate probabilities are used by Insurance companies to calculate risk premia...The problem with black swan events is, well, there are no probability tables and liability estimates thereby to compute...That is precisely why they are kept out of the purview of ordinary contracts and insurance policies as well...Arnab gave a pertinent example - if insurance companies were "forced" to quote a price for a risk they couldnt compute, they would simply have to jack up the risk premia across the board in order to be conservative...

What merits more careful review is evironmental costs - there is enough data, enough monitoring and enough empirican standard of estimating the costs...All industrial activity should have environmental costs stacked up in the final price...On that BTW, on known data nuke power is likly to come up trumps against coal or hydro...But that is really a different discussion...

[somnath]

basic truths like LWR is inferior PWR?.

Well, I ave seen this claim made a few times...Bizarre is all I can conclude in response...

[Sanku]

OT but Chally has rifled barrel and I think the 'commonality' argument was proven to be complete BS. and 'design complexities' is basically handwaving

Proven to be BS, by who? Who cant even read and understand what the other person is saying.

Go fish buddy, not everyone is MMS.

[amit]

I think there's a huge disconnect here.

A nuclear reactor is not like a plane or ship or a tank gun barrel.

If NPCIL decides to buy, say, a reactor from GE that doesn't mean that after the dotted line is signed then NPCIL will sit back and everything from ground excavation, civil construction to actual criticality of the reactor will be done by GE with no inputs from NPCIL, DAE and other GoI institutions. So, while the point made by Vina is very valid, there's no reason to believe that such multiple redundancies would not be built into Indian nuclear plants which have foreign made reactors. Heck we already run Russian ones don't we? So do we pray that there's no Chernobyl redux in India?

Take the Fukushima plant built in the late 1960s-early 1970s. According to WiKi the reactors for the different units were supplied by GE, Toshiba and Hitachi. The plant design was by GE, the architectural design by Ebasco and the actual construction was by Kajima. I'm sure the Tokyo Electric company signed separate deals with each entity.

So we can see that its not a simple case of sticking to whatever design that the original reactor equipment supplier provides. There's several layers and NPCIL and DAE can build in all the safeguards they are comfortable with.

[amit]

Well, obviously we have to use historical data for finding the probabilities. The probabilities of earthquakes and tsunamis can be estimated by scientists. Similarly, the probability that a component could fail is studied in reliability engineering. Given a reactor, they should tell us the probability that it would fail in X years.

Problem is, we don't know for certain that all these have not already been factored in, do we? The liability caps, as they exist today weren't pulled out of somebodies musharaff, I'd imagine.

[abhishek_sharma]

Abhishek-ji, Soemwhat simlar, thought hugely more intrcate probabilities are used by Insurance companies to calculate risk premia...The problem with black swan events is, well, there are no probability tables and liability estimates thereby to compute...That is precisely why they are kept out of the purview of ordinary contracts and insurance policies as well..

You are obviously correct that my example is not sophisticated. I am sure many members here can improve it. As I mentioned, reliability engineering can provide valuable insights in estimating the safety of components. And I don't think these earthquakes are really "black swan" events. I am pretty sure scientists can issue assessments like "The probability that a huge earthquake will hit Assam in 50 years is 0.52". So if they try honestly, they can surely get a reasonable estimate.

[abhishek_sharma]

Problem is, we don't know for certain that all these have not already been factored in, do we? The liability caps, as they exist today weren't pulled out of somebodies musharaff, I'd imagine.

Why is there a cap? Similar (or better) calculations should be done for estimating the surcharge. There should be no cap on liability. If the probabilities are correct, then the liability will be recovered from the surcharge. No losses for companies and people will see the true costs.

[amit]

"The probability that a huge earthquake will hit Assam in 50 years is 0.52". So if they try honestly, they can surely get a reasonable estimate.

Sure scientists can make such predictions. But the problem is to pin how huge, huge is. Is huge 7.0 on the Richter, or is it 7.3 or 7.6 or is it 8.0, 8.5, 9.0 or 9.1?

I'm sure you understand the massive difference in magnitude in each figure and the level of destruction these can do. So the standard way to do it is to look at historical data and work with the highest recorded since seismological instruments were developed 100 years ago and work with that. Which why the Japan on is a Black Swan event, it was the greater, by several orders of magnitude to what was the greatest previous record.

Now new calculations will be based on 9.0 or 9.1 in Japan.

[arnab]

Why is there a cap? Similar (or better) calculations should be done for estimating the surcharge. There should be no cap on liability. If the probabilities are correct, then the liability will be recovered from the surcharge. No losses for companies and people will see the true costs.

Probabilities by definition - are 'probabilities'. So the question of being correct does not arise.

[amit]

Why is there a cap? Similar (or better) calculations should be done for estimating the surcharge. There should be no cap on liability. If the probabilities are correct, then the liability will be recovered from the surcharge. No losses for companies and people will see the true costs.

Boss can you tell me which country or industry works with that kind of liability?

I think there's disconnect in what you desire and what is practical or industry standard. As it is the 80 year liability clause is a world's first.

[somnath]

^^^Plus, "likely damage", simplistically, is Pk X expected loss...Now even if there was a way of predicting the Pk of each "category" of earthquakes, there can be no reliable estimate of damage...The topgraphy of the area, the economic composition of the same, the inflation prevalent et al will dramatically alter those calculations...No economist can estimate these numbers either...For example, if something happens to Tarapur, the worst affected area will perhaps be New Bombay...Can anyone say that the expected damage to New Bombay 20 years back is the same as what it is today? Or can anyone predict the expected damage in New Bombay 30 years hence?

Above all, "who is responsible" isnt a viable question in "acts of God"...As I said before who is responsible for global warming that is causing erratic weather patterns?

[abhishek_sharma]

Probabilities by definition - are 'probabilities'. So the question of being correct does not arise.

I meant our estimates should reflect actual frequencies of events in the real world.

[abhishek_sharma]

Boss can you tell me which country or industry works with that kind of liability?

I think there's disconnect in what you desire and what is practical or industry standard. As it is the 80 year liability clause is a world's first.

Yes, there is a "disconnect" between what I desire and what is "practical". There is no doubt about it. What is your definition of "practical"? True blame assignment can cause genuine takleef.

[arnab]

Probabilities by definition - are 'probabilities'. So the question of being correct does not arise.

I meant our estimates should reflect actual frequencies of events in the real world.

but that is the point - earthquakes are probabilistic in nature, not deterministic. So we cannot with any degree of certainty predict an earthquake (or any natural disaster) and as Somnath has added we cannot predict the second variable either - (the extent of damage). So such an exercise will not be fruitful in the context of potential damage that may happen.

[Suppiah]

Discovered this in the course of googalling for tsunami and reactors etc..very useful article...sorry if posted earlier, as is likely to be

http://www.wired.com/magazine/2009/12/f ... ukes/all/1

[abhishek_sharma]

but that is the point - earthquakes are probabilistic in nature, not deterministic. So we cannot with any degree of certainty predict an earthquake (or any natural disaster)

Yes, and our conclusions will be probabilistic. I think we can say that "The Probability that an earthquake of magnitude 7+ will hit Assam in next 50 years is greater than 0.6".

[abhishek_sharma]

Regarding damages: How are these assessments made:

http://www.nytimes.com/2011/03/16/busin ... nsure.html

Japanese insurance companies, global insurers and reinsurers, hedge funds and other investors in catastrophe bonds are all expected to bear a portion of the losses that seem likely to exceed $100 billion. Total damage from the 1995 earthquake in Kobe, Japan, was estimated at $100 billion, according to the Insurance Information Institute, but only about $3 billion of that was covered by insurance.

[arnab]

but that is the point - earthquakes are probabilistic in nature, not deterministic. So we cannot with any degree of certainty predict an earthquake (or any natural disaster)

Yes, and our conclusions will be probabilistic. I think we can say that "The Probability that an earthquake of magnitude 7+ will hit Assam in next 50 years is greater than 0.6".

Since these are random events, any probability assignments based on past occurences will necessarily mean GIGO (Garbage In, Garbage Out). You can identify earthquake prone regions based on geology but that is about it. Second, even assuming your 0.6 for Assam is correct - how will you assign potential damage costs? presumably you will have to 'cap' it somewhere? How do you know whether that cap is appropriate?

[abhishek_sharma]

Since these are random events, any probability assignments based on past occurences will necessarily mean GIGO (Garbage In, Garbage Out).

It is not clear that they are random events. We probably do not know their mechanisms so they seem random. Therefore, probability assignments from past occurrences may or may not be GIGO. However, I agree that we need more information about it.

[abhishek_sharma]

Is this fair? No. Does this make economic sense? Yes.

If the probability of a nuclear accident is low, then a fair solution could be economically viable as well. Actually it depends on the costs from climate change related problems.

[amit]

What is your definition of "practical"? True blame assignment can cause genuine takleef.

My definition of what is "practical" is the same as that of the rest of the world, that is what is doable and what is achievable.

Since you're all for quantifying damages and then fixing liability, let me tell you another variable that you need to factor in. That is opportunity cost. What would be the economic as well as well-being opportunity cost for Indian citizen due the lack of electricity due to the non building of power plants because the liability clause was so high that companies who are the in business for commerce with profit stay away from India.

Unfortunately, and sad as it must be, India does not float in a vacuum Nirvana. World best practices in terms of liabilities is just about the best we can hope for. A few pages ago I linked some articles which talked about the likely liabilities for the Fukoshima incident. I suggest you have a relook at them.

There's no reason to think that the Japanese government cares less for the safety of its people than what an ideal (maybe nationalist) Indian GoI would do.

[arnab]

It is not clear that they are random events. We probably do not know their mechanisms so they seem random. Therefore, probability assignments from past occurrences may or may not be GIGO. However, I agree that we need more information about it.

Interview with a scientist from the US Geological Survey

http://www.usgs.gov/corecast/details.asp?ID=76

Jessica

First I would like to know, can the USGS or any other agency predict earthquakes?

Mike

Jessica, there's currently no organization or government or scientist capable of successfully predicting the time and occurrence of earthquake. However scientists are very good at saying things more general about earthquake hazards and earthquake risks. For example we can look at faults and patterns of earthquakes over many years and we can do a pretty good job of saying where on the landscape we're likely to have earthquakes on which faults, how big those earthquakes are likely to be and about how heavy the shaking is likely to be from those earthquakes.

Using that information, um, we can improve building codes, we can do ah, land use planning, we can avoid buildings next to faults that are hazardous and so forth. So we can forecast in the long term where the earthquake hazard is likely to be."

Jessica

Has the USGS done experiments to predict earthquakes?
Yes. USGS has done and has sponsored much research over several decades on earthquake prediction, both specific prediction experiments and also more general research to understand the predictability of earthquakes. For example is there something that happened in the earth ahead of an earthquake that would allow it to be detected at the surface?

The USGS and the state of California engaged in a really careful experiment on earthquake prediction. In Central California, there's a stretch of the San Andreas fault that runs through a small town called Parkfield in Central California and back in the 80s it was noticed that there had been a pattern of earthquakes of about a magnitude 6, several of these earthquakes over time spaced out about every 20 to 25 years, the last one occurring in 1966. On the basis of that pattern, the USGS and the state predicted that there would be another one coming probably in the mid 80s. And that led to a very intensive experiment to try to capture all the information possible about that earthquake with a variety of instruments and also to predict it if possible. It turned out that the earthquake did not come in the 80s and it did not come in the 90s. It actually waited until 2004.

[abhishek_sharma]

Since you're all for quantifying damages and then fixing liability, let me tell you another variable that you need to factor in. That is opportunity cost. What would be the opportunity cost for Indian citizen due the lack of electricity due to the non building of power plants because the liability clause was so high that companies who are the in business for commerce with profit stay away from India.

And what is the opportunity cost for an Indian citizen who doesn't get sufficient compensation? I suggest that you think more about this issue. We do not live in a world, where the profits of corporates is more important than fair compensation for average citizens. I think that a "nationalist" government might want to think about it.

[ramana]

So what are you guys arguing here about?

[arnab]

Is this fair? No. Does this make economic sense? Yes.

If the probability of a nuclear accident is low, then a fair solution could be economically viable as well. Actually it depends on the costs from climate change related problems.

I'm sorry you have lost me. What does this mean? you are assigning 'real' unlimited insurance payouts with a 'probabilistic' nuclear accident (however low it is, it is still positive). The benefits of nuke power have to be seen in the context of the energy matrix for long-term energy security. Coal mine accidents kill more people every year than nuke accidents have over the last 60 odd years of operation. Even the Japanese disaster won't change that statistic.

[abhishek_sharma]

Interview with a scientist from the US Geological Survey

...

Fair point. It implies that we need to improve our understanding of these issues. My point is that we should start with an estimate. It wouldn't be perfect, but it would be certainly better than a probability of zero which is commonly assumed.

[abhishek_sharma]

I'm sorry you have lost me. What does this mean? you are assigning 'real' unlimited insurance payouts with a 'probabilistic' nuclear accident (however low it is, it is still positive). The benefits of nuke power have to be seen in the context of the energy matrix for long-term energy security. Coal mine accidents kill more people every year than nuke accidents have over the last 60 odd years of operation. Even the Japanese disaster won't change that statistic.

I have been saying since yesterday that it is possible that nuclear energy could be better than coal. We should just compare the true costs. That is all.

[abhishek_sharma]

So what are you guys arguing here about?

*sigh*

This argument about nuclear/coal has continued for too long. I am going back to work.

[amit]

And what is the opportunity cost for an Indian citizen who doesn't get sufficient compensation? I suggest that you think more about this issue.

Sorry boss that's just rhetorical. When you yourself are not clear how compensation would/should be calculated and when it's been pretty well established the number of variables that exist, how can you assume that any compensation that may accrue in the event of an accident would be insufficient? Heck we don't even know if such a hypothetical accident would be operator fault or equipment supplier fault - there would be different compensation liability calculations for each case.

Moreover, we haven't signed a single deal and yet how can we be sure that compensation would be "inadequate"?

At the same time opportunity cost for lack electricity is happening today even as we debate this on the Internet. Pity a lot of folks who are experience power cuts at this moment can't take part in this debate.

But you're right this discussion has continued for long enough. Need for a timeout.

[Sanku]

So what are you guys arguing here about?

Basically that since we cant predict accidents, its safe to discard them in any calculations we make.

Compared to -- if we cant predict, lets not use a dangerous tech whose outcomes we dont know and can be catastrophic.

[amit]

So what are you guys arguing here about?

Basically that since we cant predict accidents, its safe to discard them in any calculations we make.

Compared to -- if we cant predict, lets not use a dangerous tech whose outcomes we dont know and can be catastrophic.

After so many pages if this is the conclusion, I can only say WOW!

[abhishek_sharma]

Sorry boss that's just rhetorical.

It is not rhetorical at all. It is the main point (as far as I am concerned).

When you yourself are not clear how compensation would/should be calculated

Let us not go there. I am pretty sure how it should be calculated. Actually it was discussed in this thread many months ago.

What should I pay you if I burn down your car/house? See it is not very difficult.

In the case of loss of lives, how about ensuring that their family members get the salary of deceased person for next N years? This is too much? I know. I am not "practical" . Okay, let us think about something else. These days govt pays 1-5 lakhs in the case of railway accidents. How about adjusting it for inflation/cost of living?

As somnath was saying many analysts work with far more sophisticated models. It should be pretty easy to use their methods.

... and when it's been pretty well established the number of variables that exist, how can you assume that any compensation that may accrue in the event of an accident would be insufficient? Heck we don't even know if such a hypothetical accident would be operator fault or equipment supplier fault.

I have no idea what you are talking about.

[amit]

I have no idea what you are talking about.

I think you missed the addition I made to that sentence in my post:

- there would be different compensation liability calculations for each case.

Anyway I think we've both raised valid points. It's best if we mull over them for sometime so that we come back after sometime and see if we can reach a consensus. Like I wrote before, I don't think our positions are all that different on the issue.

[abhishek_sharma]

Okay. Let us move on now. More than enough discussion.

[Sanku]

More inconvenient truths by Bramha Chellany who is likely to get called names here pronto for saying the truth once again

While Fukushima spent-fuel fires rage, it's important to remember that spent fuel at the similar Tarapur plant has been accumulating for four decades, with the U.S. unwilling to take it back or let India reprocess

[abhishek_sharma]

Jaitapur clearance may be reviewed

http://www.hindustantimes.com/Jaitapur- ... 73791.aspx

On Tuesday, Ramesh acknowledged that the radiation impact has not been studied and said his ministry was looking at adding more environmental safeguards but refused to revoke the environment clearance