20
the UNEP/ICOLD collected data. It should be noted that all the figures showing tailings
dam data are plotted with the number of incidents as the dependent variable. Fig.2
shows an increase of incidents starting in the mid-1960s. This increase is most likely
due to the larger number of tailings dams constructed after 1960 combined with the more
thorough documentation of tailings dam operation. The USCOLD and UNEP surveys
have provided a total of 211 incidents for comparison. Clearly all incidents have not been
reported, and this collected number form a subset of the actual number of tailings dam
incident that have occurred from the early 1900s to 1996. The following discussion of
results is presented in terms of the number of incidents, rather than downstream effects
or cost of repair or remediation. A comparison of the number of incidents with tailings
dam height is given by Fig.3, where it will be seen that about 57% of the incidents
occurred in dams less than 20 m high. A comparison of actual dam failures with dam
height is given by Fig.4 which shows that approximately 63% of the failures occurred in
dams lower than 20 m.
When an impoundment has become completely filled, or when tailings production
ceases, the tailings dam and its retained impoundment is described as inactive. This
does not make them immune to accidents or failures and, as shown by Fig.7, relatively
few incidents were associated with inactive dams. A cause of failure of inactive dams
is increase of pool water level, bringing the pool closer to the dam crest thereby
bringing the phreatic surface to the downstream slope, leading to eventual slips and
overtopping. A method to avoid the overtopping of closed impoundments consists of
building a subsidiary dyke some distance back from the crest of the tailings dam to
permit the level of the pond water to rise during an emergency without risking the main
dam. The section of such a dam in Germany is shown by Fig.5.
As can well be imagined, the type of construction of the tailings dam must play a
part in its behaviour. One of the earliest and the most common types of construction
is by the upstream method: tailings dams constructed with tailings by the upstream
method have been documented in South Africa in the early 1900s (types of
construction have been described in Bulletin 106): traditional embankment dams when
used to retain impoundments, are referred to as water retaining type dams. Because
of the risk of failure of dams built by the upstream method, particularly when subject to
earthquake shaking, the downstream method was developed; and there is a centreline
method that is a compromise between the former two methods. As Fig.6 shows
clearly, there are many more incidents with dams built by the upstream method than
with other types, but also there are many more of these type of dam than of other
types amongst the examples that we are considering.
An indication of the causes of incidents, for active and inactive tailings dams are
given by Fig.7, where it will be seen that the leading cause of incidents for active dams
are slope instability, overtopping and earthquake. For inactive dams, the leading
causes of incidents are overtopping and earthquake. Fig.8 similarly shows incident
cause for active dams, separating failures form accidents. Finally, Fig.9 shows the
total incidents with their cause separated by tailings dam type. This figure also
indicates that the leading causes for incidents are slope instability, earthquake and
overtopping: particularly so for dams constructed by the upstream method. The
incidents must be reviewed in terms of the number of particular dam types in
operation. The upstream method is the oldest and most commonly used method of
tailings dam construction. This method, as pointed out by Mittal and Morgenstern
(1977), was used at sites prior to the use of foundation investigation and slope stability