快猫短视频

TOM JORDAN is one of the few people who know what an earthquake really feels
like. 鈥淎ll you feel is a whump whump bump. It鈥檚 a sharp motion,鈥� he says. That
may not sound like the normal experience of a quake鈥攖he minute-long
rolling motion you feel while standing on the quivering surface of the Earth.
But that is because Jordan, a seismology researcher at the University of
Southern California in Los Angeles, wasn鈥檛 on the surface of the Earth. He was,
quite literally, caught in the middle of the quake 4 kilometres underground,
down among the moving rock.

Jordan was in a South African gold mine when an earthquake that registered
3.7 on the Richter scale struck, killing three people. It wasn鈥檛 a natural
earthquake: South Africa lies on a stable part of the continent and rarely sees
large quakes. The quake struck because mining blasts and excavations had
dislodged rock beds and gradually built up huge pressures and stresses in the
walls of the mine. Eventually, like glass squeezed in a vice, the rock exploded,
throwing out thousands of tonnes of rock at phenomenal speeds.

Rockbursts like this kill more than 100 South African miners annually, and
over the years they have cost thousands of lives worldwide. Miners urgently need
an early warning system, but despite decades of research no one has yet been
able to see a rockburst coming (see 鈥淗ow to handle a rockburst鈥�). However,
engineers at the Southwest Research Institute (SWRI) in San Antonio, Texas, now
think they have found a way to provide reliable predictions. Simon Hsiung and
his colleagues claim that subtle sound patterns emitted by hard, brittle rock
can reveal when it is about to reach breaking point. 鈥淭his technique can be
applied to hard-rock mines anywhere in the world,鈥� Hsiung says. 鈥淭here is no
geographical or geological limitation.鈥� If their initial findings prove
significant, they could save lives and livelihoods across the globe.

Engineers have known for decades that the best way to find out anything about
the state of the rocks is to listen to them. Every time a microscopic fracture a
few tenths of a millimetre long occurs it makes a tiny 鈥減op鈥�. Most of the sound
produced is of a very high frequency and is only partially audible to people. In
danger areas, mine workers install sensors to listen for these sounds. And when
they hear a high-frequency chatter鈥攃aused by many microfractures happening
at once鈥攖hey know that trouble is approaching. However, translating the
sounds into precise information about the state of cracking in the rock is
tough. Combining the patterns of the cracks with numerical modelling techniques
can show where the stresses are building up, but knowing where the stresses are
does not tell you when it will rupture. This is the same problem which hampers
earthquake prediction.

Hsiung is confident that he has a solution鈥攁 new way of listening to
the rocks talk. In a laboratory at SWRI, Hsiung and colleagues Amitava Ghosh,
Asadul Chowdhury and James Lankford used a machine to squeeze 10-centimetre-high
pillars of tuff鈥攁 consolidated volcanic ash. As they carefully increased
the pressure, the researchers used an array of sensors, linked to a computer
running a specially developed program, to monitor the symphony of cracks
occurring inside. Eventually, the tuff burst, leaving complete records of the
acoustic emissions of the rock pillars as they were being crushed.

Hsiung鈥檚 analysis of the sound is based on the fractal nature of cracking in
the tuff. Many things in nature show fractal or scale-invariant behaviour (
快猫短视频, 15 September 1990, p 38). Looking at a picture of a desert
sand dune, for example, you can鈥檛 tell if the dune is millimetres or kilometres
long, unless you know the scale. Whether you look at ripples in the sand from an
aeroplane or through a magnifying glass, the wind creates the same patterns.

Similarly, the microfractures are distributed in a fractal pattern within the
tuff, and they appear at time intervals that also follow a fractal pattern. As
each tiny crack appears, the sensors 鈥渉ear鈥� the pop and the computer program
notes the time. The program relates the number of cracks occurring within a
particular period of time to the 鈥渇ractal dimension鈥� of the cracking. This
factor, which gives a measure of the complexity of fractal patterns
(快猫短视频, 20 December 1997, p 22)
also provides an easy way to predict rockbursts.

As the stress on the tuff increases, the computer repeatedly calculates the
fractal dimension. At first, it grows steadily. Then, suddenly, the fractal
dimension starts to drop. This, Hsiung says, is the point of no return: once the
fractal dimension has peaked, the tuff soon blows apart under the stress. This
is because the peak in the fractal dimension occurs when the cracks coalesce
into a single plane. From then on, most of the cracking will occur on this
horizontal slice of the pillar, known as a failure plane. The structure is now
unstable and bursting is inevitable.

Because of the fractal nature of the formation of microfractures, the SWRI
researchers believe that fractures will occur in larger samples in exactly the
same way. 鈥淭he microfracturing process in this tuff rock should be similar to
that in pillars in hard-rock mines,鈥� Hsiung says. So the analysis of small-scale
samples should help the researchers understand how microscopic cracks grow in
walls and pillars hundreds of metres below ground.

But it is still unclear exactly how much actual warning time this technique
provides. In the laboratory experiments failure occurred in a matter of minutes,
and the peak in the fractal dimension occurred anywhere between midway and
three-quarters of the way through the loading tests. 鈥淯nderground, if you don鈥檛
mine too fast, you probably have several days before it bursts,鈥� Hsiung
says.

That would give mine engineers time to take action to de-stress the rock, or
to simply move the workers elsewhere in the mine. But Hsiung has a lot to do
before he needs to worry about the exact amount of warning time the fractal
dimension provides. It鈥檚 going to be a long, hard task convincing a jaded
industry that rockburst prediction is even possible.

鈥淭here have been a few notable predictions that could just as well have been
coincidences,鈥� says Brian White of Spokane Research Laboratory, part of the
National Institute for Occupational Safety and Health in Washington DC. Many
efforts at prediction have failed to materialise into rockbursts, he adds.
Heather McLoren, a rock mechanics engineer at the Lucky Friday silver mine in
northern Idaho, is even more sceptical. 鈥淵ou might as well treat anyone who says
they can predict rockbursts like a snake-oil peddler,鈥� she says.

Jordan鈥檚 experiences in South Africa have also led him to believe that there
may be more to rockburst prediction than the SWRI researchers have accounted
for. When miners excavate a tunnel, he says, the opening reduces the load that
normally holds together the beds of rock in and around the mine. As a result,
the beds can slip, causing a seismic tremor. The resulting rockbursts can attack
the tunnels and working spaces from below and from the sides, not just with the
simple downward pressure that Hsiung modelled in the laboratory. Because of
this, Jordan says, there could be a different kind of sound in the rock prior to
the burst.

But Hsiung is confident that his technique will work for all rock types and
under all rockburst conditions. He believes that the fractal analysis should
still provide some warning, because the slip still compresses the rock and
causes a fracture plane to form. But he admits that it is early days: the team
presented their first report on the technique this August at the North American
Rock Mechanics Symposium in Seattle, Washington.

They now plan to gain the industry鈥檚 confidence gradually, starting by using
acoustic data taken from mines that have experienced rockbursts. 鈥淲e hope to
show from that data just how good this technique is,鈥� Hsiung says. Then they
will set up their equipment in a mine and wait for a real, live rockburst. It
might take some careful planning to make the prediction accurate, Hsiung admits,
but he is confident that they can make it work. Only then does Hsiung expect the
mining industry to take notice. It will be worth the wait鈥攁nd the scorn,
he believes. There might be a few accusations of peddling snake oil, but Hsiung
is convinced that the fractal chatter of rocks will save hundreds of lives.

ROCKBURSTS can happen wherever mining puts stress on brittle, hard rock such
as quartzite and granite. So far they have been seen in more than 100 places
worldwide, including Canada, the US, Britain, Germany and Australia, as well as
South Africa. The explosions can have the force of a small atomic bomb,
generating seismic events greater than magnitude 5 on the Richter scale. Miners
who have survived them report witnessing solid granite walls waving like flags
in the breeze.

The dangerous path to a rockburst begins when a mine is drilled or blasted
open. The Earth鈥檚 natural response to a blasted mine is to close up the hole,
but the hard rock walls and the pillars that the miners leave in place for
support do not give in easily. As a result, stresses mount in the structure,
producing tiny fractures. The subsequent mining process removes more rock and
adds to the problem. In effect, everything the miners do brings catastrophic
collapse a little closer.

The best solution found so far has been to adopt new mining strategies,
moving away from the harder rock, which bursts more violently, and reinforcing
the excavated areas with sand fill and concrete. The Lucky Friday mine in
northern Idaho implemented these ideas in 1987 after suffering three fatalities
in as many years. The mine owners also covered the interior with chain link
fencing to catch bursting walls. These strategies have been largely successful
in preventing rockburst injuries. But rockbursts still occur about once a month,
and they are big enough to kill anyone caught standing in front of them.

Heather McLoren, a rock mechanics engineer at the Lucky Friday mine, believes
that by watching high-risk areas for the signs of rockburst activity, they are
getting better at identifying the factors involved in bursts. But it鈥檚 still not
prediction, she admits. 鈥淲e can come fairly close in some instances, but we
still can鈥檛 definitively say when and where until after the fact.鈥�