Early in the afternoon of Friday January 25th, 2019, an 86 meters high tailings dam at the iron ore mine of Córrego do Feijão, in the region of Minas Gerais, Brazil, ruptured1,2, discharging 11.7 million cubic meters of toxic mud3, killing at least 121 people with more than 200 persons still missing at the time of writing. The dam had been inactive since 2015, and decommissioning plans had been made in 20163,4.
In the aftermath of the event, Vale announced it would ramp up efforts to dismantle 10 dams similar to the one that burst, which should cost the company R$5 billion ($1.37 billion) over three years5,6, and incur a 10% reduction in annual production, relative to expectations. Vale was immediately fined close to $95 million by Brazilian environmental agencies, $430 million were frozen for direct indemnification of workers, and courts requested that a further $3 billion of Vale’s assets be frozen7,8,9,10,11,12,13,14. This was accompanied by lawsuits, arrest warrants against company personnel, and an 18% dip in the company’s share value the following Monday15.
Córrego do Feijão is located roughly 100 km away from Mariana, where a tailings storage facility (TSF) at the Samarco mine, owned by BHP and Vale (each with a 50% share), failed on November 5th, 2015, destroying the village of Bento Rodrigues, with 19 deaths, 60 million of cubic meters of iron ore waste released – some of it eventually reaching the Atlantic Ocean, more than 600 km away16. Thousands of fishermen were left out of work and several lawsuits ensued, including one for R$20 billion ($5.4 billion) that was settled in 2018, and a separate one for R$155 billion ($42 billion) currently suspended to allow negotiations for a settlement – which might be impacted by the recent event according to the prosecutor in charge. Samarco already spent $4.4 billion ($1.2 billion)17,18,19 on compensation and clean-up. In August 2014, the Mount Polley tailings dam operated by Imperial Metals also failed in Canada, leading to public outcry, and liability20,21.
Mining-related environmental liabilities have received relatively little focus in Western Europe and among investor communities. The repetition of such tragic events should prompt a deeper interest into the topic. The Córrego do Feijão seems a perfect example of a underestimated risk22.
TSFs are supposed to store waste from ore processing for an indefinite amount of time. Some of them are amongst the largest infrastructures built by men.
Figure 1: Chiquicamata TSF. Source: Google CNES / Airbus DigitalGlobe, extracted from Google Earth
No one knows exactly how many exist around the world. Estimates of 3,500 are often put forward23,24,25,26, yet, in 2014, there were of the order of 18,000 registered active mines accounting only for copper, gold, iron ore and rare earth27.
Three broad categories of TSFs exist, sketched on Fig. 2: upstream, centerline, downstream, in decreasing order of safety and cost. The 10 dams that Vale pledged to decommission within three years are its remaining upstream dams.
Independent studies suggest that TSFs fail more often than regular dams23,24,25,30,31,32,33, and that recent years have seen more impactful events. As ore grades are depleting, tailings dams are getting larger. In British Columbia, at least four tailings dams two to six times higher than Mount Polley are planned34.
In the past few years, initiatives from the industry to monitor TSFs and environmental issues better have emerged. For instance, the Mining Association of Canada released a guide which was updated to incorporate the conclusion of the task force that analyzed the Mount Polley disaster35,36 – before its failure, Mount Polley was seen as adhering to excellent standards. After Mariana, the International Council on Mining and Metals (ICMM) released new guidelines for tailings management37, water management38, as well as a binding agreement involving its member companies for better TSF management39 (Vale re-joined ICMM in 2017).
Coming up with cogent risk profiles is an arduous task. First, TSFs are highly engineered features, especially at modernly operated mines, and risk factors vary across operations. Contrary to regular water dams, they can be “raised” several times, on a rhythm that depends on initial planning, commodity prices and reassessment of reserves. TSFs are expensive to maintain properly, constituting almost exclusively a liability. Second, disclosure in the mining industry has historically been low, and initiatives led by NGOs usually focused on greenhouse gases (GHG) emissions seem cumbersome and not necessarily useful.
As large failures keep happening, and as more earth is moved to satisfy economic activity and the energy transition, the lack of asset-level public data is bewildering. The availability of a public database covering the entire industry with simple information such as dam age, height, location, the number of times it was raised, the main construction material, and the main method of construction, should be a minimum.
Beyond TSFs, a cornerstone of environmental monitoring in the mining industry is environmental impact assessments which occur before operation starts, much too early to forecast actual impacts of a multi-decadal operation. For instance, remediation costs are systematically underestimated at that stage40.
Mining is slowly coming under the spotlight in the context of increasing metal needs and finer understanding of environmental issues beyond GHG emissions. However, it is still a sector whose overall liability is easy to push to the bottom of the list of concerns in countries like France where no mine is currently operating. The current policy focus on climate change is important, but only captures part of the environmental liabilities caused by human activities. Public opinions – especially in countries where mining has a negligible footprint but where consumption of manufactured good is high, along with long-term investors, have a responsibility to push for more disclosure on environmental risk in the mining sector.
Luc Bonnafous, Energy-Climate Analyst Sources available via hyperlinks