DGR Why is model building important? 

How would you build a risk model for doing nothing?  It can be done.  What do you think its risk is over let's say 100,000 years?


What am I for and what I know

 I'm for the selection of the safest spot for DGRs based upon facts including all the risks for long term storage of nuclear waste.  It's too important an issue to be derailed.

Further, I don't care where it goes as long as it is the best site geologically and strategically with risk minimized.  There are no scientific breakthroughs required.


Written for Canadian Community News by Mike Sterling

To Comment on this article Click Here

In watching the Joint Review Panel (JRP) hearings and the course of events before them, it is clear there is barely hidden a disdain for model building from some interveners and questioners because they think it allows them to say nobody will give them 100% assurance of no risk. So they proudly say:

"What good are models?"

Some otherwise knowledgeable people are wholly out to lunch about the subject.  It is so foreign to them that they have to roll their eyes, rather than discuss the subject.  Why is this so?

The answer is simple.  They just don't encounter models in anything other than the weather.  They don't realize that many of the things that are so important to their everyday lives have been subject to some form of risk analysis.

By the way, weather modeling is the penultimate in difficulty and should not be compared to the work on the DGR.

Let's ask ourselves a question.

Would anybody question modeling, security and risk analysis, if we were going to bury Captain Kidd's treasure for 250,000 years?  Would anyone question the risk of doing nothing?  Of course few believe in doing nothing, just do nothing over their lifetime.

Why are models useful?

How can the usefulness of models be explained?   There is a misconception that there is one giant model and data is being poured into it to create a sort of witches brew, complete with a misty mirror showing the future. The witch then adds some magic vial of chemicals and out of the mist in the clouded mirror pops the results.  That's not how it works.

People wonder about the accuracy of the input data and the reliability of the model makers.  So let's try to explain them another way.

First of all, there is rarely one model.  Models are set up to exercise and stress particular things.  The results are then combined to probe deeper with more models and simulations for an overall view.   The output of one model may be the input to another.

While data input is important, we miss something if we don't realize that models are used to tell the engineer/scientist/official what factors are important.  That is, they tell the engineers what types of inputs are most sensitive and potentially can cause trouble.  If you don't grasp that concept, you cannot understand models' usefulness.

You may say this is nonsense, but that's the very point.  You cannot trust what you call your magic common sense.  So your sense of what is common is rarely valid. Common sense should not exist in good model making.  It is too often dead wrong.

If you have not solved any problems using models, then don't be so sure of your opinions.  Some of the people testifying at the JRP hearings who scoff at models do so out of a flawed sense of their own ability to deal with complexity.

Common Sense

If I've heard it once, I've heard it a hundred times in the hearings.  A naysayer will say: 

"Well, it's just common sense ......". 

That's just the point.  There is no place for our unverified 'common sense' in model building or the DGR project in general.  You have to assume that what you are dealing with is starkly uncommon.  That's the very essence of risk analysis.  Would the naysayers like to present a risk analysis for doing nothing?  We repeat that. Have they properly modeled the do nothing strategy?  After all all waste now stored was always considered to be in temporary storage.

Complicated Processes

In complicated processes, it is useful to find out what exactly is the most sensitive set of input variables and scenarios.  These are often not known before the modeling begins.

In every significant model that I've ever worked on in the past, I was always surprised by what factors were the most significant.  I learned to never trust my common sense, whatever that is. I did not and could not trust others' common sense either.  Once you find the sensitive parts, you can then tune the process to eliminate their influence on the outcomes.

Click the orange arrow to read the second column

Models work and are useful for four main reasons:

  1. They allow you to test many, many  scenarios millions upon millions of times.

  2. They allow you to find the most sensitive parts of the process. (very important)

  3. They allow you to bring together participants in the process to give you more scenarios to test.

  4. They allow you to get a fix on the economics of minimizing risk or designing around it.

What is a Scenario?

A scenario is a description of what could happen.  It's a what if?  Get in a room, cram as many smart people as you can inside and ask them to give you any and all scenarios that they can dream up.  'Drain' them of all their thoughts on ways things can go wrong.

In model building you want to present to the model as many scenarios as you can.  The modeling process will vary the basic scenarios millions of times, thereby stressing the system.  This will reveal places in the process that need more attention, more scrutiny and maybe even an extension to the model or a new type of model.  

Simple examples of DGR scenarios are:

  • A group of terrorists attack the DGR before it is closed and decommissioned by stealth from the lake side at night.

  • A millennium storm arises on the Great Lakes.

  • A small plane sneaks into the airspace above the sight and drops a bomb.

  • Once sealed the underground chambers collapse after 50,000 years.

  • One hundred thousand years from now a glacier covers the site of the DGR with one mile of ice.

  • The DGR site is hit by a meteorite.

  • There is an accident on the transportation route.

  • Some chemical process takes place at great depth.

  • In lowering the waste, a mechanical failure takes place.

Notice that the above scenarios are very much more of a risk with above ground storage.

You can see why you may need separate models to study these individual scenarios and then combine them in a final risk model.

Also, note that the above scenarios are very high level and they contain a myriad of smaller scenarios.

Model building in the case of a DGR project is very useful and well suited to the task.  It is much, much, much smaller than the weather, but complicated enough to require good analysis and lots of scenarios.

It is a lot better than uncommonly flawed, quick to judge 'common sense' we hear so much about.  Forget common sense.  It's very uncommon.

In April of 2012, I wrote an article about Risk analysis.  You will get more examples of what I mean by the uncommon use of modeling methods and my experience with them.

Science Risk management and intuition Read More


Survey Participate in our latest Kincardine Times survey Read More Survey Participate in our latest Saugeen Times survey Read More Survey Participate in our latest Walkerton News survey Read More

Scrolling stops when you move your mouse inside the scroll area.  You can click on the ads for more

for world news, books, sports, movies ...

Wednesday, October 23, 2013