Brumadinho — a Tailings Dam That Liquefied in Seconds and Killed 270

At 12:28 on 25 January 2019, Dam I of Vale’s Córrego do Feijão iron-ore mine, near Brumadinho in the Brazilian state of Minas Gerais, collapsed without warning and released roughly 9.7 million cubic metres of saturated mine tailings as a fast-moving mudflow. The wave swept over the mine’s own administrative complex and a canteen where workers were eating lunch, then ran more than five kilometres to the Paraopeba River. It killed 270 people, most of them Vale employees and contractors. The cause was static liquefaction: a loose, water-saturated, brittle deposit of iron tailings lost almost all of its strength in an instant and flowed like a heavy liquid. No earthquake, no overtopping and no rainstorm on the day triggered it. The dam failed under its own static weight.

Dam I was an upstream-raised tailings dam, about 86 metres high, built and raised in stages from 1976 onward and last fed tailings in 2014. The upstream method is the cheapest way to raise a tailings impoundment and the most dangerous: each lift is founded partly on the soft, previously deposited slimes behind the wall, so the embankment is built on the very material it must retain. At Brumadinho that retained material was fine iron tailings, deposited loose and kept saturated by a high internal water table, with no functioning drainage to draw the water down. Loose saturated tailings are metastable. They can stand for years at a factor of safety barely above one, then collapse catastrophically when a small disturbance pushes them past their brittle strength.

The collapse was filmed by the mine’s own surveillance cameras. The footage shows the downstream face bulging and then disintegrating across roughly 80 percent of its width in about five seconds, with no slumping or cracking beforehand. The Expert Panel commissioned to investigate found that internal creep deformations under sustained load, combined with a loss of soil suction as a wet 2018 rainy season raised pore pressures in the upper tailings, drove the deposit past its peak strength. Once a sliver near the crest liquefied, the failure ran backward through the impoundment in a retrogressive chain, each liquefying block undermining the next, until the whole face had flowed out.

The investigation, led by the geotechnical engineer Peter Robertson and released in December 2019, concluded that the failure occurred by static liquefaction with no detectable warning, despite extensive instrumentation. That finding is the indictment. The hazard was inherent in the structure — a brittle, saturated, upstream-raised deposit with no drainage and almost no reserve of strength — and the monitoring regime was watching for movement that, in a brittle flow failure, never comes. Brumadinho, four years after the same company’s Fundão failure at Mariana, became the disaster that ended upstream tailings dams in Brazil and forced the global mining industry to confront a failure mode it had treated as theoretical.

Fundão (Mariana) — the Flow-Liquefaction Breach That Poisoned 668 km of River

On the afternoon of 5 November 2015, the Fundão tailings dam — an upstream-raised iron-ore tailings embankment operated by Samarco, the joint venture of Vale and BHP Billiton, in the Germano mining complex near Mariana in Minas Gerais, Brazil — breached and released roughly 43.7 million cubic metres of saturated mine tailings down the Gualaxo do Norte and Rio Doce valleys. The mudflow reached the village of Bento Rodrigues within about thirty minutes, killed 19 people, and spread contamination along some 668 kilometres of watercourse to the Atlantic seventeen days later. It was, by volume and reach, the largest tailings-dam disaster on record at the time. The embankment did not overtop and was not undermined by piping. It liquefied — the saturated tailings sand momentarily lost its strength and flowed like a fluid.

The Fundão dam was raised by the upstream method, the cheapest and most common way to grow a tailings dam, in which each new lift is built partly on top of previously deposited tailings rather than on competent ground. The method is sound only if the tailings beneath remain drained, dense, and capable of carrying load. At Fundão they did not. A chain of construction and design decisions — an under-built starter drain that was never properly repaired, a forced realignment of the left abutment back over a deposit of weak fine “slimes,” and a rising water table — left a wedge of loose, fully saturated sand sitting on soft clayey slimes near the left abutment. That sand was a loaded gun: dense enough to stand under static conditions, loose enough to collapse and liquefy if disturbed.

The disturbance came in stages over 2015. As the embankment was raised above the slimes, the soft layer extruded sideways under the load, stretching and loosening the sand above it and nudging the whole left flank toward instability. The process was already far advanced when, at roughly 14:30 on 5 November, three small seismic tremors registered in the region. Computer modelling later showed those minor shocks added a final increment of horizontal movement to the slimes and the overlying sand — enough to tip a system on the edge into collapse. Around 16:00 the left abutment slid, the sand liquefied, and the dam burst.

The Independent Panel commissioned by the owners — chaired by Norbert Morgenstern and reporting in August 2016 — concluded that the failure was a flow liquefaction caused by a specific, traceable sequence of construction defects, not an unforeseeable act of nature. The case, paired with the Brumadinho disaster that followed in 2019, drove a worldwide reckoning with upstream tailings dams and produced the first global tailings-management standard. Fundão is the canonical demonstration that a tailings dam is a soil-mechanics problem in which saturation and density, not the height of the wall, decide whether the structure lives or dies.