Samarco / Vale — Fundão tailings dam collapse (Mariana, Brazil)

1. Episode summary

Samarco Mineração S.A., a 50/50 iron-ore joint venture of Vale S.A. and BHP Billiton, operated the Germano mining complex near Mariana, Minas Gerais. The Fundão tailings dam, commissioned in 2008, stored both sandy tailings (coarse, free-draining, but liquefaction-susceptible when loose and saturated) and slimes (fine, clay-like, low-permeability). Between 2011 and 2015, Samarco was executing a roughly US$3.5 billion expansion ("Project Fourth Pellet Plant") aimed at lifting nominal pellet capacity from ~22 to 30.5 Mt/y, commissioned in 2014. Over the same period, the Fundão structure was progressively raised using the upstream method and its design was modified after a concrete gallery beneath the left abutment was found, in late 2012, to be structurally deficient — forcing a setback of the embankment alignment over previously deposited slimes. Seepage and cracking episodes were recorded on the left abutment during 2013, and instrumentation showed elevated saturation. Raising continued. On 5 November 2015, the dam suffered a flowslide failure, releasing roughly 43.7 million cubic metres of tailings into the Gualaxo do Norte and Doce river systems, destroying the village of Bento Rodrigues, killing 19 people, and contaminating some 668 km of watercourses to the Atlantic. The strategic question the episode turned on was whether the joint-venture operator would arrest a recognised and accumulating geotechnical risk on an active tailings structure whose design had been altered under operational pressure.

2. Sources

Primary:

1. Morgenstern, N. R.; Vick, S. G.; Viotti, C. B.; Watts, B. D. Report on the Immediate Causes of the Failure of the Fundão Dam (Fundão Tailings Dam Review Panel, commissioned by Cleary Gottlieb Steen & Hamilton LLP on behalf of Samarco, Vale and BHP Billiton), 25 August 2016 — principal technical investigation; chain of events, left-abutment setback, slimes extrusion, liquefaction triggering.
2. Samarco Mineração S.A., 2014 Annual Sustainability Report — production figures, expansion project status, Fourth Pellet Plant commissioning, 2014 operating context.
3. Ministério Público Federal (Brazil), indictment of 21 individuals and four corporate defendants (Samarco, Vale, BHP Billiton Brasil, VogBR), filed October 2016 — charges including qualified homicide, flooding, building collapse, environmental crimes; procedural history through acquittal reported November 2024.
4. High Court of England and Wales, judgment on liability in Município de Mariana & Ors v. BHP Group (UK) Ltd & Anor, 14 November 2025 — strict-liability finding against BHP; documentary findings on foreseeability, absence of certain studies, and dam-raising decisions.

Secondary (with justification):

1. "Mariana dam disaster" — Wikipedia, consolidated encyclopedic entry drawing on the Morgenstern panel report, Brazilian federal prosecutors' filings, and contemporaneous news coverage. Used as a cross-reference index to primary documents.
2. "Technical note: Fundão Mine tailings dam failure," Ground Engineering (Emap), 12 February 2020 — peer-directed engineering synthesis of the panel findings; unpacks lateral-extrusion mechanism and setback geometry for a professional geotechnical audience.
3. Mining Technology (Verdict Media), "Samarco dam disaster: dealing with the fallout of a tragedy" — investigative feature synthesising documentary evidence, leaked internal warnings, and regulatory response.
4. House of Commons Library, research briefing CDP-2023-0133, "Mariana Dam disaster" — UK parliamentary synthesis of facts, litigation status, and settlement architecture.
5. Finance Uncovered, "Dam winners: Mining bosses at companies being sued over Brazil's worst environmental disaster awarded $516m in pay since 2015 catastrophe," financeuncovered.org, 2023 — documents the internal Samarco bonus-tracking document and the remuneration committee's maximum risk-management score awarded less than three months before the collapse, and the year-prior downgrading of dam-collapse risk in the internal risk register. Added during §4 research.
6. IBTimes (International Business Times), "BHP-Vale's Samarco Was Warned About Crack In Dam Responsible For Deadly Burst In Brazil, Engineer Says," ibtimes.com, 2016 — reports Pimenta de Ávila's September 2014 findings, crack identification, and recommended remedial measures including buttressing and daily water-level monitoring. Added during §4 research.
7. Samarco Mineração S.A., Relatório Bienal 2015–2016 (Biennial Report), section "About the failure of the Fundão dam," samarco.com — documents the absence of a contingency plan or evacuation routes in the event of dam failure, and the gap in downstream warning for Bento Rodrigues residents. Added during §4 research.

Tertiary (flagged):

1. IUCN, "The Fundão Dam Failure" (2022 retrospective story) — flagged tertiary, used for frame only.

3. OTA narrative

Observe. The observation apparatus existed and produced usable signals. Piezometers and inspection records on the Fundão structure flagged elevated saturation and seepage on the left-abutment setback from at least 2013; surface cracking and seepage episodes were logged and reported internally; a 2013 internal review documented structural issues and alternative-design considerations; the concrete gallery defect that drove the 2012 alignment setback had itself been identified by instrumentation and inspection. The technical literature on upstream-raised tailings structures built over soft slimes, liquefaction susceptibility of loose saturated sands, and the specific risk of lateral extrusion was industry-available to an Archetype-level mining major and to a panel-level consulting cadre. The observation task was therefore routine for the peer group of large iron-ore tailings operators: the signals required to form a correct diagnosis were generated and were read by the operator's own instrumentation team. Observe was not a root cause; it functioned as a transmission step that carried the failure signal to decision-makers who then did not act on its implication.

Think. The reasoning step from the observed signals to the governing risk picture was the critical failure. The correct interpretive framework — that a setback placing embankment loading over soft slimes, combined with rising saturation and continued upstream raising, could trigger static liquefaction and a flowslide — was accessible in the geotechnical literature that the Morgenstern panel itself drew on, and within the competence envelope of the firms advising Samarco. Instead, the design change and subsequent raising were treated as manageable within existing operational parameters, and the 2013 seepage-and-crack events were read as drainage-management problems rather than as indicators of a developing liquefaction mechanism. The reasoning failure was therefore an Easy-Wrong Think: the correct framework existed, was documented in the company's own monitoring data, and was not applied. Think is a root-cause phase in this episode.

Act. Execution compounded the reasoning failure rather than catching it. The operator continued to raise the embankment, did not arrest operations to reconcile the setback geometry with the governing risk picture, did not commission the studies that the 2025 English High Court later identified as absent, and did not put adequate downstream warning arrangements in place for the communities below the dam. These were not exotic capabilities — dam-halt protocols, independent-panel review before further raising, and downstream alarm systems are routine moves for a large mining joint venture operating a tailings structure of this class. Act is a root-cause phase in this episode and is classified Wrong at the easy end of the task-difficulty axis: the relevant checks and interventions were standard for the peer group and were omitted. Observe and Act together transmitted a correctable situation into a catastrophic flowslide; Think was the pivot.

4. Modality evidence

Direction. The episode's directional frame was set by Samarco's board-approved "Project Fourth Pellet Plant" expansion, which committed the company to lifting pellet capacity from approximately 22 Mt/y to 30.5 Mt/y through a roughly US$3.5 billion capital programme commissioned in 2014 (Samarco 2014 Annual Sustainability Report). The Fundão dam was a load-bearing physical constraint on that expansion: it needed to continue receiving and retaining tailings at rising volumes to keep the mine operating at target throughput. The strategic choice that bound all downstream decisions was therefore not simply to expand, but to expand through the Fundão structure — an upstream-raised tailings dam whose design had already been altered in 2012 after the discovery of a structurally deficient concrete gallery beneath the left abutment (Morgenstern Panel Report, 2016). No evidence in the sources identifies a discrete, attributable decision by Samarco's board or executive team to arrest the expansion trajectory and re-evaluate the dam's fitness for purpose as the geotechnical picture worsened from 2013 onward. Direction's contribution here is real but secondary: the expansion direction itself was not intrinsically wrong, but it was held fixed as a constraint rather than interrogated against the evolving risk picture, compressing the space available for Think and Act to function.

Scoring note (zero-modality rationale): the directional layer described in this subsection is acknowledged in the §4 evidence as present and specific but is not load-bearing for the strategic failure causation of the episode — the operative failure causation mechanism was located in Structure, Processes, Culture rather than in the directional choice itself. Direction is therefore recorded at zero per cent on the rationale of modality acknowledged in narrative but not load-bearing for the strategic value created in the episode. Categorisation under METHODOLOGY-ota-scoring-v4.md §5: modality acknowledged in narrative but not load-bearing.

Structure. Samarco was governed as a 50/50 joint venture, with its board constituted entirely of nominees from Vale S.A. and BHP Billiton; board decisions required the joint agreement of both shareholders, giving both parents structural ability — and structural responsibility — to block or condition operating decisions (High Court judgment, Município de Mariana v. BHP Group, 14 November 2025). Day-to-day operations were delegated to a Samarco executive committee operating beneath the board, while BHP exercised oversight through various committees and sub-committees at multiple levels, from strategic decisions to operational risk management, and conducted financial and technical audits whose findings were reported back to BHP (ibid.). The design and raise authority for the Fundão structure was vested in external geotechnical consultants: Joaquim Pimenta de Ávila held the original design mandate from 2008 to 2012, after which his contract was not renewed in the same capacity and VogBR Recursos Hídricos e Geotecnia was retained to design the planned further raise beyond the original Pimenta de Ávila envelope (Wikipedia, Mariana dam disaster; Mining Technology, Samarco dam disaster). This succession of consulting relationships — and the transition of Pimenta de Ávila from designer to part-time inspector — created a structural gap: no single body held continuous, integrated authority over design intent, monitoring data, and raise decisions simultaneously. The High Court found that BHP's governance over Samarco was substantive enough that it knew, or ought reasonably to have known by August 2014, that there were serious problems with the dam's drainage and stability, yet the governance architecture did not convert that knowledge into a halting instruction (High Court judgment, Município de Mariana v. BHP Group, 2025).

Processes. The monitoring and reporting machinery generated warning signals but did not close the loop from observation to remediation. Piezometers and inspection records documented elevated saturation and seepage on the left-abutment setback from at least 2013; surface cracking was logged and reported internally; by August 2014 the replacement blanket drain intended to control saturation had reached maximum capacity (Morgenstern Panel Report, 2016; Wikipedia, Mariana dam disaster). In September 2014, Pimenta de Ávila documented a crack in the structure and communicated recommended remedial measures including buttressing and daily water-level monitoring (IBTimes, "BHP-Vale's Samarco Was Warned About Crack in Dam," 2016; Wikipedia). The process failure was not in generating these observations — they were generated and transmitted — but in the review and escalation procedure that should have evaluated their cumulative significance. The Morgenstern Panel identified the absence of certain requisite studies — a proper liquefaction analysis and a comprehensive stability analysis reflecting the changed geometry — as a process-level absence rather than an individual error (Morgenstern Panel Report, 2016; High Court judgment, 2025). No evidence in the sources indicates that Samarco operated a dam-safety review process that would have halted further raising pending the completion of those studies. VogBR's role in assessing stability was later characterised, in the Brazilian federal prosecution, as having presented a false stability report; VogBR subsequently stated that it had not monitored the drainage works (Wikipedia, Mariana dam disaster). The emergency preparedness process was similarly deficient: Samarco and the downstream communities lacked a contingency plan or evacuation routes in the event of dam failure, and residents of Bento Rodrigues received no functional warning before the tailings wave reached the village (Samarco Biennial Report 2015–2016; IUCN, "The Fundão Dam Failure," 2022).

Capability. The geotechnical competences required to diagnose and respond to the developing Fundão risk picture were available to the organisation in principle. Pimenta de Ávila's standing as one of Brazil's foremost tailings dam engineers and his ability to identify the crack and recommend remedial action in September 2014 demonstrate that the relevant diagnostic capability existed in the consulting ecosystem on which Samarco drew (Wikipedia, Mariana dam disaster; IBTimes, 2016). The Morgenstern Panel itself was composed of internationally recognised geotechnical engineers who were able to reconstruct the liquefaction mechanism from the same data stream that had been available to Samarco's operators and consultants, indicating the interpretive capability required was within the peer group's reach (Morgenstern Panel Report, 2016; Ground Engineering, Technical Note, 2020). The upstream-construction method's particular liquefaction susceptibility when built over soft slimes was documented in the technical literature and within Samarco's own monitoring data. The capability gap that the episode exposes is therefore not absolute: it is the gap between having access to capable external advisors and having internal institutional capability to integrate and escalate consultant findings across the design–operations–governance chain. The transition from Pimenta de Ávila as designer to Pimenta de Ávila as part-time consultant, followed by VogBR holding raise-design authority without adequate monitoring of drainage performance, left the integration function under-resourced (Mining Technology, Samarco dam disaster; Wikipedia). Low-confidence note: the sources available do not provide granular detail on Samarco's internal geotechnical staffing levels or the specific technical credentials of the dam-safety personnel employed in-house between 2012 and 2015; the capability assessment here rests primarily on indirect evidence from consultant roles and investigation report findings.

Scoring note (zero-modality rationale): the Capability contribution described in this subsection is classified at the boundary with Processes in the scoring record — the §4 evidence locates the operative driver of the episode's failure causation in Processes rather than in a standalone Capability contribution. Capability is acknowledged in narrative as evidenced but does not carry independent weight in the scoring; weight is borne by Structure, Processes, Culture. Categorisation under METHODOLOGY-ota-scoring-v4.md §5: classification boundary with an adjacent modality.

Culture. The most precisely evidenced cultural signal in the record concerns the incentive and risk-scoring system. An internal Samarco document tracking progress against short-term bonus targets — reviewed by BHP and Vale as the remuneration committee less than three months before the dam's collapse — rated Samarco executives at the maximum score for risk management, including for a risk category that specifically contemplated tailings dam failure; that full-marks assessment gave no apparent weight to the warnings received about the dam's cracks and the decisions that had continued to raise it (Finance Uncovered, "Dam winners," 2023). In the year before the disaster, Samarco's internal risk register had in fact downgraded the perceived risk of a dam collapse (ibid.). Taken together, these documents indicate an organisational environment in which the formal risk-management discourse had been decoupled from the observable deterioration of the structure: the risk was named but scored as controlled, while the signals pointing to inadequate control were not converted into escalation. The Brazilian federal prosecution filed charges that included the presentation of a false stability report, pointing to conduct by consultants that the cultural environment in and around Samarco tolerated or did not effectively challenge (Ministério Público Federal indictment, October 2016; Wikipedia). The English High Court's finding that BHP knew or ought to have known of the dam's critical instability by August 2014, yet approved increased production and continued raising without rectifying the drainage deficiency or commissioning the absent studies, is consistent with a governance and operating culture in which production-trajectory preservation dominated safety escalation (High Court judgment, Município de Mariana v. BHP Group, 2025). These observations are corroborated by the emergency preparedness gap: the absence of a functioning evacuation plan for Bento Rodrigues represents a failure of the safety-culture norm that downstream communities' lives are a non-negotiable planning constraint, not a residual risk to be managed informally (Samarco Biennial Report 2015–2016; IUCN retrospective, 2022).

---