杏吧原创

BILL GLASSLEY wasn鈥檛 looking for trouble, but he could be facing mountains of
it. He and his team at Lawrence Livermore National Laboratory in California only
wanted to create some tools for geologists like himself to study the Earth鈥檚
anatomy. His aim was to build a giant computer model that would travel in time
and reveal the changing structures of rocks and mountains. But Glassley鈥檚 foray
into virtual geology might be about to make his life a little uncomfortable.
With his first look into the geological future he has foreseen some rather
serious surprises in store for an underground nuclear waste dump planned by the
government.

Glassley鈥檚 prototype is a model of Yucca Mountain, a flat-topped volcanic
ridge that rises some 360 metres out of the Nevada desert. The mountain is the
proposed site for dumping nearly 80 000 tonnes of radioactive waste from
American nuclear reactors and military facilities. Glassley鈥檚 simulations show
that drilling tunnels and storing the waste will produce some harmless鈥攁nd
possibly even advantageous鈥攃hanges to the mountain鈥檚 structure. The bad
news is that, according to Glassley鈥檚 model, the waste will create corrosive
waters that could threaten waste canisters. There will also be some dramatic
changes in the chemistry of the rock below the tunnels. And Glassley has not yet
begun to factor in the effects of geological forces that grind, shake and split
the region鈥檚 rocks.

He never meant the software to be a political bombshell. The idea was for it
to help geologists understand everything from groundwater supplies to earthquake
prediction, by creating virtual landforms such as deep-sea vents, volcanoes and
geological fault lines. But whatever Glassley鈥檚 intentions, his Yucca Mountain
results seem certain to be seized on by government officials and environmental
campaigners. The Department of Energy鈥攚hich also happens to be Glassley鈥檚
employer鈥攈as already poured billions of dollars into checking the geology
of Yucca Mountain. And as it has no alternative plan if the mountain is deemed
unfit for use as a waste repository, there is a lot of pressure on the
government to approve the disposal site.

Glassley鈥檚 program is not the sort of application you鈥檇 want to try running
on your desktop PC. The virtual Yucca Mountain exists inside more than 1400
parallel microprocessors, controlled by the Blue Pacific supercomputer. At its
full potential the mountain will be split into 40 million cells of various
sizes, each one described by 120 geological parameters such as stress, heat,
chemistry, grain size, porosity and permeability. In key areas, the cells are
just centimetres across. The computing power, fed with the best geological data
available, can take the rocks forwards and backwards through millions of years,
simulating every conceivable process going on within this portion of the Earth鈥檚
crust. The whole thing has grown from equations developed in the late 1980s to
describe the movement of water through rock pores and fissures鈥攁n
essential part of modelling geological features. At that time no one had the
software or the computing power to crunch the complex coupled equations, but now
Glassley and mathematician John Nitao have done the necessary programming and
proved that it works.

Once Glassley and Nitao were over that obstacle, they hit another one: too
little geological data. To get credible results from a simulation, everything
needs to be broken down to the scale of centimetres. Lawrence Livermore鈥檚
supercomputing facility and the virtual mountain software make that possible;
the problem, says Glassley, is that it needs to be given data of comparable
accuracy. And, despite extensive surveying with boreholes, gravimetric surveys
and detailed mapping of the main tunnel already drilled into the mountain, there
still wasn鈥檛 quite enough data to build its digital doppelg盲nger.

To compensate, the researchers generated their own data. They decided what a
younger version of Yucca Mountain would probably have been like, grew it into a
high-resolution structure and used the computer-generated data to make up for
the missing geology. It might seem like a fudge, but it is surprisingly
accurate: the virtual data even produced features that geologists had observed
on the real mountain but had never been able to explain.

Once they were satisfied with their simulation, Glassley and his team drilled
miles of tunnels in their virtual mountain, loaded them with canisters of hot
nuclear waste, and sent the mountain on some preliminary voyages through time.
It did not take long before the surprises started to appear. The tunnels
themselves are a major disturbance to the mountain, Glassley says. 鈥淭hen you add
these hot things and it just goes nuts,鈥� he says.

The radioactive decay of spent fuel rods鈥攖he primary waste targeted for
Yucca Mountain鈥攇ives off a lot of heat in the tunnels, often raising the
temperature to over 100 掳C. That鈥檚 hot enough to drive water out of the
14-million-year-old compressed volcanic ash known as 鈥渢uff鈥�, which is the
stratum geologists favour for tunnelling into the mountain. Some 20 per cent of
the mass of the tuff is water, bound to the rocks鈥� crystalline minerals.

According to the simulation, the combination of heat and water causes
physical processes and chemical reactions that change the structure of the rock
around the tunnels. The water that boils off moves up and away from the heat
through fissures and pores in the rocks. On reaching cooler rocks, however, this
vapour condenses and starts to run back down, dissolving some minerals out of
the rock as it goes. Eventually this mineral-laden solution reaches the hot
rocks again, where it vaporises, deposits its load of dissolved minerals and
starts the cycle again (see Diagram).

Just a hundred years into the simulation, the heat and the movement of carbon
dioxide has caused changes in the pH of rocks below the tunnels. Long,
keel-shaped regions of more alkaline rock begin to appear, stretching towards
the groundwater below.

After 300 years, the leaching cycle has created dome-like regions in the rock
above and to the sides of the tunnels. The pores and fissures that initially
carried the water have become clogged like household pipes encrusted with
hard-water deposits, and this eventually prevents the water from moving through
the rocks. It is a neat self-sealing trick, and since water percolating down
from the surface is the main vehicle for spreading radioactive waste at Yucca
Mountain, these relatively impermeable domes are probably a good thing, Glassley
believes.

Chemical attacks

This benefit comes at a cost, however. Once the initial heat of the waste
materials abates, and water is able to seep back into the tunnels, local
variations in the chemistry of the rock mean this water is more acidic in some
areas and more alkaline in others. This creates zones of highly variable acidity
and alkalinity just centimetres apart within the tunnels and surrounding rock.
That鈥檚 a problem, says Glassley, because it means that waste canisters have to
be resistant to a variety of chemical attacks.

Matters aren鈥檛 improved by another surprise that the simulation revealed.
Some waste packages produced more heat than others, and water evaporating from
around the hot packages was continually condensing on their cooler neighbours.
The result was dripping-wet canisters that, thanks to the mineral chemistry in
the tunnels, also had high concentrations of chloride salts on their surfaces.
And salt plus metal equals corrosion, as anyone who has seen the effect of
seawater on steel will know.

The simulation also threw up a perplexing result: no matter how carefully the
researchers tried to make the virtual tunnels the same as each other, each one
developed differently. After puzzling over this for a while, the modellers
realised that the difference was a matter of location. The mountain rock
provides a huge heat sink, but each tunnel has a different amount of rock around
it, depending on its position in the mountain. This affects the way the heat and
water behave in each of the tunnels. It鈥檚 a factor that probably would never
have occurred to the repository鈥檚 developers if the virtual mountain had not
been built.

鈥淭his is an unbelievably powerful learning tool,鈥� Glassley says. Almost every
time he runs the simulation it triggers new ideas and provides new insights into
the challenges facing the Yucca Mountain project鈥攁nd this, he insists, is
good news for the DoE. By discovering these problems at the simulation stage, it
may be possible to devise methods of arranging different types of waste with
different heats to minimise the pH changes and moisture problems. At
the very least, the designers will know where to place monitoring devices during
the planned 100-year 鈥渃onfirmation period鈥�: a century of checks to ensure that
nothing unexpected is going to happen in the repository.

But the timescale that really matters is much longer: the 10 000 years for
which the government stipulates that the mountain must safely contain the waste.
During that time Yucca Mountain could experience massive earthquakes or even the
beginnings of an ice age. Neither situation has yet been simulated in the
virtual mountain.

Turn the clock back 10 000 years and Yucca Mountain was a greener, wetter
place. If another ice age should begin within a similar length of time, rainfall
might again increase, sending extra rainwater percolating down through the
mountain, and raising the water table beneath it. This would make it more likely
that radioactive waste might escape into the water supply. Drier weather, on the
other hand, would probably help keep the waste isolated from water. Either
scenario can be easily applied to the virtual mountain to see how it responds,
says Glassley.

Modelling what happens deep underground is rather more difficult. Yucca
Mountain is set in a tectonic province called the Basin and Range. It has arisen
because the North American continental crust is being stretched eastward and
westward, causing blocks of crust to tilt and slide diagonally to fill the
widening gap. This stretching has created a series of long north-south ranges,
with equally long basins in between鈥攅ach bounded by long faults.

According to preliminary measurements by geologist Brian Wernicke of the
California Institute of Technology, the Yucca Mountain region moves about 2
millimetres a year. That may not seem like much, but it translates into 20
metres over the 10 000-year life of the disposal site. Even a remote possibility
of a 20-metre crack in a nuclear waste dump is hardly reassuring.

Unfortunately, Wernicke鈥檚 data won鈥檛 necessarily help make the virtual Yucca
Mountain more accurate. No one knows exactly where and along which faults that
annual 2 millimetres of movement is taking place, and there鈥檚 no way yet to
determine how it will affect Yucca Mountain. The virtual mountain does take
fault lines into account, Glassley says, but only as conduits or barriers to
fluid movement, depending on the minerals they contain.

Ultra-secure designs

Uncertainties like these will not go unnoticed by the Yucca Mountain
project鈥檚 opponents who have already picked scores of holes in previous
geological assessments of the mountain. Despite the DoE鈥檚 strenuous efforts to
prove that the mountain will never leak waste, its case is far from proven and
many of the problems thrown up by Glassley鈥檚 preliminary simulation had never
been anticipated before. The DoE is now trying to counter these worries with
ultra-secure designs for the waste canisters: stainless-steel cases wrapped in
corrosion-resistant nickel-chromoly casings and titanium-palladium drip shields.
Paradoxically, if these leak-proof containers prove to be a success they will
give the protesters another argument against storing the waste underground: if
the canisters are so safe, why do they have to go inside a mountain?

Like every other state, Nevada doesn鈥檛 want to be a dumping ground for
nuclear waste. Unlike most states, however, the majority of Nevada, including
Yucca Mountain, is under federal control鈥攕o the state doesn鈥檛 have much
say in its use. Local politicians claim that Nevada was chosen simply because it
has fewer people and votes than other states.

The virtual mountain won鈥檛 settle the issue: Glassley and his team recognise
that their work is just a scientific drop in a very large political bucket. But
he remains optimistic that virtual landscapes will help to resolve similar
issues in the future. 鈥淚 would like to be in the position where this model can
be used to arbitrate,鈥� he says. 鈥淚 think in a lot of cases it鈥檚 not going to
provide the ultimate answer, but it鈥檚 going to provide a next step.鈥�

He envisions three-dimensional projection systems that would someday enable
decision-makers and scientists to 鈥渨alk鈥� into a time-travelling virtual mountain
at any moment they choose. If they disagree on a scenario they could tinker with
the model鈥檚 parameters as they see fit, and immediately test the results.

Glassley is confident that the virtual Yucca Mountain is worth the
effort鈥攅ven given the controversy it is likely to stir up鈥攂ecause of
the multitude of other potential applications. The same equations and coding
could be applied to test various rock-fluid dynamics theories along the San
Andreas Fault, explore for oil, and investigate damaging intrusions of seawater
in coastal areas. It could even help determine where Mars is hiding its water,
he says.

鈥淭his sort of computing power is going to revolutionise science,鈥� Glassley
predicts. Until now, geologists have only been able to look backwards, but
that鈥檚 all going to change, he says. Thanks to the time-travelling mountain,
geologists will from now on be keeping a wary eye on the future, too.