Science - USA (2020-08-21)

sciencemag.org SCIENCE

than that of reservoir and power dams, ac-
cording to one estimate.
Each disaster has its own constellation of
causes, but some arise from seemingly trivial
errors. At Mount Polley, investigators led by
Norbert Morgenstern, a geotechnical engi-
neer at the University of Alberta concluded
that part of the dam was built on a weak
patch of silt and clay. Exploratory boreholes
drilled prior to construction were too shallow
to find the problem. Builders further weak-
ened the dam by making its walls steeper
than planned, after the company ran short
of rock. One night, the weight of the sludge
became more than the dam could bear.
It could have been much worse. No one
died. Workers ultimately repaired the dam

and shoveled up much of the mud that had

buried the creek. (The company says the

spill didn’t cause long-term harm to the

Quesnel Lake ecosystem, but some eco-

logists say it’s still too early to tell.)

Morgenstern, who also led the investiga-

tions into the 2015 Brazilian incident and

the 2018 Australia failure, has found that

faulty engineering, including inadequate

scrutiny of the underlying geology, was at

the heart of all but two of 15 major incidents

between 1980 and 2015.

One major problem, he says, is the

“normalization of deviance.” The phrase,

coined after the 1986 explosion of the

space shuttle Challenger, describes how

engineers can be lulled into accepting a se-

ries of seemingly small risks that snowball

into a catastrophe.

There is an unwritten covenant that

regulators and mine owners can count on

engineers to design a safe tailings system,

Morgenstern told a gathering of Brazilian

geotechnical engineers in 2018. “That cov-

enant,” he said, “has been broken.”

THE SEARCH IS ON for fixes. Some mining

watchdogs are calling for replacing one

common type of dam, called an upstream

dam, and banning future use of the design.

Upstream dams are built in stairlike stages,

heading upstream over the accumulating

tailings (see graphic, above). Part of the

weight of each added step is borne by the GRAPHIC: C. BICKEL/

SCIENCE

Liquefed tailings

Core sample

Weak layer of

soil below

sample level

Original construction

Dam modifed

over time in

problematic

ways

Tailings

Starter dam

Weight of dam bearing

down on tailings

Next upstream

dam layer

Soil

Suspended
Water dam material

Solid tailings

Stress

onstruction

difed

ime in

matic

Built-up

tailings

Pipes releasing

mine waste

1

2

3

4

Starter dam

Inconsistent

construction

Prebuild site survey

Mine

tailings

Pooled water

Tailings

beach

Pipe releasing

mine waste

Dam face

Dam

face

Starter dam

Time

Why mining dams fail

Basins filled with leftover sludge from mining can grow to half the

size of Manhattan. Historically, dams containing tailings have failed at

more than 100 times the rate of water-holding dams. In just the

past decade, failures have killed hundreds and contaminated ecosystems

with toxic mud. Many failures have common culprits.

1 Liquefaction
Infiltration of water into the dam
is a chief source of failures. In
extreme cases, water combined
with stress such as an earthquake
can cause an earthen dam to
suddenly turn to liquid.

3 Shaky ground

Geologic weaknesses in the ground

below a dam can leave it vulnerable.

In one of the biggest recent failures,

dam builders didn’t drill deep enough

to discover a weak layer left by

receding glaciers.

2 A risky design

Upstream construction is a common

but failure-prone approach. The dam is

raised gradually, as tailings accumulate.

With each new level, the dam tilts

upstream, relying on tailings below

to help carry the load.

4 Piecemeal changes

Unlike water dams, tailings dams

evolve. They are built bit by bit over

decades as mine waste piles up.

This creates more potential for errors.

908 21 AUGUST 2020 • VOL 369 ISSUE 6506

Published by AAAS