1 February 2019

Brumadinho disaster: the extent of the environmental impact

Posted by Dave Petley

Brumadinho disaster: the extent of the environmental impact

The human cost of the Brumadinho disaster in Brazil is becoming increasingly clear. Latest reports suggest that there are 110 known fatalities, of whom 71 have been positively identified, and a further 238 people remain missing.  Six people remain in hospital.  The cost of this disaster feels very real when you look at this news report, which provides details of just of few of the victims.  In addition, the environmental impact of the disaster is also apparent.  The image below, collected by Planet Labs, shows the area covered by the tailings:-

Brumadinho disaster

Planet Labs image showing the area covered by the tailings in the Brumadinho disaster. Image copyright of Planet Labs, used with permission.

.

The area affected is somewhat smaller than that of the Samarco disaster in Brazil.  The total ecological destruction in the river is clear.  There is now an urgent need to retain the remaining tailings left in the storage area and to contain the waste, and any toxins draining from it, in the river channel.

Meanwhile, there remains considerable speculation as to the causes of the disaster.  The image below is a Planet Labs image collected on 25th January 2019, showing the dam that collapsed to generate the Brumadinho disaster:-

Brumadinho disaster

Planet Labs image of the site of the Brumadinho disaster, collected on 25th January 2019, two days before the collapse. Image copyright of Planet Labs, used with permission.

.

This image shows no obvious indications of ongoing work on the tailings dam itself, and there are no signs of work on the tailings stored in the pond.  Indeed. there is also no obvious indication of the incipient failure developing.  Thus, the events that led to the collapse remain intriguing, and it is going to be important to ascertain the causes if similar accidents are to be avoided.

The paper on the dam that I highlighted earlier this week (Pirete and Gomes 2013) shows that the dam was built using the upstream method.  This is a cross-section of the structure:-

Brumadinho disaster

Indicative cross section of the tailings dam that collapsed to generate the Brumadinho disaster. Diagram from Pirete and Gomes (2013).

.

In this case most of the retaining structure is built upon the tailings themselves.  This is not unusual, and it is not necessarily unsafe, but it does assume that the properties of the tailings are well-understood.  Note in particular that the tailings are described as being dilative – there is a nice primer on the mechanical behavior of tailings here, but in essence dilative tailings are assumed to show strain hardening, i.e. they will become stronger as they deform.  This may be a key assumption on the context of this failure.  It is perhaps worth looking at the report of the Samarco tailings dam accident, which attributed the collapse to an extrusion process.  In this case, failure also developed in a dam built using the upsteam method.  The original design assumed that a layer of sand sitting over the tailings would be kept in an unsaturated state, but this was changed during construction, and the sands were allowed to become saturated.  Ultimately, this allowed a so-called lateral extrusion event to develop, precipitating the collapse of the dam.

The operator of the site, Vale, are reported to have decided to remove all tailings dams of this type.  It will be interesting to see the timescale for these plans, and how they will monitor the stability of these structures in the interim so as to ensure no repeat of the Brumadinho disaster.

References

Pirete, W. and Gomes, R.C. 2013. Tailings liquefaction analysis using strength ratios and SPT/CPT results. Soils and Rocks, 36 (1), 37-54.

Planet Team (2019). Planet Application Program Interface: In Space for Life on Earth. San Francisco, CA. https://www.planet.com/.

Acknowledgement

Thanks to Robert Simmon of Planet Labs for his help in acquiring and processing the imagery.