Córrego do Feijão Tailings Failure 1/25/2019

Our banner is an ever looping image of the actual failure by static liquefaction of Vale’s Dam I at the Córrego do Feijão  mine. This page will eventually have our complete preliminary assessment and an annotated bibliography of resources providing authoritative background information relevant to this failure, which ranks as the 6th worst in recorded history, based on data available as of February 6, 2019.

We have had a team of volunteers translating and summarizing the entire portfolio of key official government documents of relevance to this failure. We thank Bruno Milanez for acquiring and transferring this important portfolio to us.


It seemed likely from the first descriptions and images of the Corrego do Feijao Dam1 failure that the cause was static liquefaction, a transformation of the nature of the tailings resulting from over saturation that makes them perform more like water than soil causing a greater force that propels released tailings much further than the same volume of compacted tailings would travel in a foundation failure. Some experts are saying that while a liquefaction clearly occurred it is not clear whether it followed from or caused the failure.

 Fundao, Stava, and Merriespruitt ,Los Frailes, Sullivan Mine &Pinto Valley are a few well known examples of static liquefaction failure in history. It is a subject of growing interest and concern that has evolved over at least two decades . It has forced a re evaluation of what was taught in mining schools and has been best knowledge for a large part of the worldwide community of dedicated mine engineers. It is not now a major focus of education and apprenticeship as the worlds future engineers set to work . A great deal of knowledge has been advanced by our chief Compiler, Engineering, Roberto Rodriguez, by revered engineers like our colleague Andrew Fourie and by the venerable Mike Davies & Todd Martin among many others..

It’s just not finding its way it into practice in the feld. It is just not art of the thought and decision making that has shaped the world’s existing portfolio of some 18,000 TSF’s and those that are planned to receive the worlds mineral production this decade.

… design practice in many mining regions has in fact discounted the possi-bility of the mechanisms and criteria for this failure mode, the possibility of its occurrence has often been overlooked in the search for other causes of failure.  …

 The simplistic view is that by defining the friction angle and pore pressure of the sand we can predict the strength of that sand, the drained strength. The exception these references allow for sands is during an earthquake when the sand may become ‘liquefied’. Clays on the other-hand are deemed to be cohesive and have an undrained strength. Those readers who have benefited from a more enlightened geotechnical education may not find this a credible proposition, but it is clear to the authors that even as we enter the 21st  century, a range of educators, regulatory and quasi-governmental groups,and an alarming number of geotechnical consultants still have not un-learned their first series of lectures in soil mechanics based on textbooks expounding the views noted above.Until these simplistic models have been un-learned by all involved with the design, licensing, and construction of tailings impoundments, a major contributor to failures, i.e. inappropriate and incorrect designs based upon a lack of understanding of the tailings strength, will continue ” (emphasis added)

Davies, McRoberts , Martin ( 2002)

It is in this dark crevice of “unknowing” that the ever present liabilities of Dam 1 Corrego do Feijao matured to the 5th largest failure ever in recorded history based on what we have now as deaths, runout and release.

Dam Vi is the pool of water n the foreground. The fully disgorged Dam 1 is the big black whole beyond that.

It is a good fortune of history that by coincidence Pirete, a geotechnical engineer for Vale, elected to do his thesis on static liquefaction using Dam1 Corrego do Feijao as his case study.( Pirete 2010) His thesis advisor, a published author, Gomes, and Pirete later published a journal article(Pirete & Gomes 2013) expanding the concepts of the thesis and again including Dam 1 Corrego do Feijao as the case study.

These two studies give us a window into critical measurements of Dam 1 over time that evidence it to be in a continuous state of extreme vulnerabity over its entire life to the moment of failure .

Pirete and Gomes conclude a resistence to collapse by static liquefaction that all our experts and others agree are not supported in the actual data they present and yet it is possible Vale, inappropriately relied on this as gospel even though reliance should have been placed on the data indicating a constant state of extreme vulnerability to collapse.

We also present an authoritative explanation of static liquefaction by our own expert in static liquefaction, Roberto Rodriguez Pacheco ,whose life works and analysis of TSF’s in Spain, Peru and elsewhere leads us toward static liquefaction as a cause of failure. Dr Rodriguez’ work embodying hundreds of peer reviewed papers and thousands of citations emphasizes the level of saturation (80% being critical)and the particle size and shape of tailings . As grades fall it takes more grinding to extract minerals and that results in tailings with two many fines and with characteristics that undermine attainment of a permanently dry consoilidated state resistent to failure.

Finally we present analysis of a series of google earth images over a period of 6 years by Frederico Lopes Freire which shows increasing saturation from constant natural drainage into and over the TSF that is not diverted or captured. Frederico Lopes Freire notes a deformation in the upper left face of the dam where, in our view, the collapse initiated based on hundreds of loops of our banner. (The first indication is a large cloud of black dust right behind this deformation. Three small areas in the lower right were after this first upper left dust. According to Dr. Rodriguez-Pacheco saturation is a key pre cursor to static liquefaction with 80% being a critical level.

The image below is the exact moment the first crack appeared in the upper left of the top layer of the dam. This is the area where Feirers interpretation of the google earth images shows increasing saturation.Simultaneously a large circular area in the front of the lower dam cracks. This area is also emphasized in Feirer’s analysis of his google earth photos which show collapse of the main drain leading from the center top to the foot of the original starter dam. Pre failure Feirer’s photos show erosion all the way down the face. A 2017 safety audit shows massive erosion outside the toe of the starter dam.

The July 11, 2011 image below is the first in the series presented in Frieres analysis of increasing saturation.

“Drainage of superficial waters originating from rain or other sources is collected at two locations as indicated at levels 912 and 913. The flux and direction are shown by the arrows.  At the top of the image, at level 948, there is a clearly visible water intrusion, possibly originating from the higher elevation points located behind the dam, possibly a brook.The dense forest extending back from the dam is possibly another source of drainage into the dam.” Frederico Lopes Friere.

The image below is the last in the series presented by Frieres and he notes a deformation at the top of the dam right before the change in direction that seems to correspond to the area of the first break at the moment of initiation of the collapse.between his numbered areas 945 and 926.

July 2018 is the date of the TUL-SUD stability analysis , presented below, which makes no note of the deformation but does emphasize the drainage problems within the dam. The TUL-SUD seems to support Frederico Lopes Fereirs analysis.

In presenting the two studies of the potential of static liquefaction both  using Dam 1 as a case study, it is important to recognize that the purpose of these studies was to provide data for evaluation of a trigger level for static liquefaction and not to evaluate dam 1 per se. WMTF uses  all such studies in our failure narratives where the data is authenticated and valid and has a bearing on a historic description of the dam or documentation of facts associated with failure.

 It is important though to bear in mind that both works were  a good faith effort to establish a workable static liquefaction trigger so that future failures could be assesed reliably for liquefaction potential.  This kind of work is essential to eventually being able to better assess dam stability and prevent future losses. The finding that Dam1 per their triggers was not subject to liquefaction is not the same as a finding that in fact that there was no potential for static liquefaction .  It takes a large body of work by many to finally evolve a tool we can rely on and use.  Andy Fourie one of the world’s leading authorities on static liquefaction is still searching  for that holy grail of a critical threshold and he too is using case studies ( see www.tailliq.com).

Pirete & Gomes 2013

As it happened   Pirete  the doctoral candidate  on the thesis and the co-author  of the  published paper was and is an employee of Vale.  The recommendations he made in the thesis to prevent saturation that might set up liquefaction conditions were made to, and apparently adopted by, Vale at Dam1 and as of 2013, the journal version of hi s thesis these water management issues were in place and working. We respect and value these two works but we also share the reservations others have expressed expressed that the data presented supports their conclusion that the dam was strong enough to withstand a massive sudden liquefaction and that the probability of liquefaction was very low. As we will expand on below a close associate of WMTF’s who is among the worlds top experts on static liquefaction is very specific and very harsh on the short comings of this work even as a meaningful contribution to the the search for a”Liquefactin Trigger”.

Even if they were excellent as to the serach for a liquefaction trigger, these documents do not have the same purpose or the same weight as the 2018 TUL-SUD stability analysis previously presented below.  The Tul -Sud  analysis  and the fact  that all they could say was that no factor of safety  could be determined has far more weight  than the two studies and the information they contain with a bearing on this failure.

Minas Gerais Law required stability analysis periodically and presumably had standards for that that which we have yet to review.   That may change the relative importance of all the documentation we have.

We use all such studies in our failure narratives where the data is authenticated and valid and has a bearing on history of the failed dam and is pre-failure state.  It is important though to bear in mind that both works were a good faith effort to establish a workable static liquefaction trigger so that future failures could be presented.  The finding that Dam 1 per their triggers was not subject to liquefaction is not the same as a finding that in fact that there was no potential for static liquefaction and any use or reliance on it in that way would be questionable.  It takes a large body of work by many to finally evolve a tool we can rely on and use.  Andy Fourie one of the worlds leading authorities on static liquefaction is still searching for that holy grail of a critical threshold and he too is using case studies ( see www.TAILLIQ.com)


It was not previously known that the risk of failure by static liquefaction had actually been evaluated pre -failure.  A few days ago one of our esteemed AdvisoryCommittee members Roy Wares, a Due Diligence expert, forwarded  a photostatic copy of a 2010 thesis in Portuguese by a Vale Employee, Pirete, on which Vale had apparently relied in taking the steps Pirete recommended to keep the tailings from achieving saturation levels that threatened static liquefaction.

Our Chief Compiler, Environmental Consequence, Dr. Steve Emerman, a hydrologist, translated and summarized the thesis for WMTF. His summary and the original photostat of the thesis can be downloaded below.

The conclusion of that thesis gives more assurance of the dams resilience to failure by static liquefaction than was warranted by the finding of TUV-SUDthat there was not sufficient data to calculate a factor f safety. A Factor of safety of at least 1.5 is mandatory by widely accepted best guidance for any dam of any construction with a top hazard ranking . Further this dam exceeded the adviseable safe height for a downstream construction as the venerable Klohn Crippen Bergin reminded us all. in their 2/9/2019 statement. Huge amounts of water were flowing into and over the TSF over 6 years as documented below. in photos and well informed observations of apparent saturation conditions.

Pirete & Goes Particle Size Distribution Dam1 Corrego do Feijao

We have not yet confirmed a valid Factor of Safety immediately pre failure or over time leading up to failure. We have not yet obtained or reviewed records on two anecdotally confirmed major leaks, one on July 2018 and one just before failure.

This post was prepared by Lindsay Newland Bowker in consultation with our in house experts and others but is not yet fully peer reviewed. We welcome any corrections or additions and any documents or records on factor of safety, saturation and the reported leaks. Please feel free to contact Dr. Emerman direrctly with any questions or comments on his summary of the thesis. ( contact information provided in header of his summary) February 14, 2019

Download Steven Emerman English Summary

Download Thesis


Three years after the thesis was  completed , a portion of it was  used as a case study in a journal article by Pirete& Gomes .which restated the Pirete thesis conclusion that the dam, per their critical threshold calculation, was not vulnerable to static liquefaction  “Tailings dams constructed using the upstream method generally have relatively low- density materials with a high degree of saturation. Such conditions can generate the phenomena of liquefaction, which is potentially critical in slurry tailings disposal systems. In this paper, this approach was applied for stability assessments to verify liquefaction potential in an upstream tailings dam built by the hydraulic fill technique and located in the QuadriláteroFerrífero(Iron Quadrangle) region, southeastern of Brazil. The results ratified the safety condition of the impoundment although they have demonstrated that the tailings tend to exhibit contractile behavior during shear, indicating liquefaction susceptibility.”

“The author collected thirty-three case histories of liquefaction flow failures that were back-analyzed to evaluate the yield and liquefied shear strength. Relationships between yield strength ratio and corrected SPT and CPT resistance were developed for use in liquefaction triggering analyses and also, those between liquefied strength ratio and corrected SPT and CPT resistance were developed for use in post-triggering stability analyses”

The part of the paper addressing Dam1 as the case study Starts at page 51 . The conclusions ar quoted below

“This liquefaction analysis procedure was applied to a Dam I from “Córrego do Feijão” mine, an 81 m high ore tailings disposal system (CF tailings), located in the Quadrilátero Ferrífero (Iron Quadrangle) region / Brazil, resulting in the following conclusions:

• CF tailings tend to exhibit contractile behavior during shear and then these materials are susceptible to liquefaction;

 • The referenced subdivided section of the downstream slope of Dam I resulted in nine layers being susceptible to liquefaction with resistances given by the mean values obtained from both SPT or CPT profile zones

 • From the specific critical failure surfaces along the downstream slope of Dam I (obtained from SPT or CPT corrected values and subdivided into 16 segments), was obtained an average static shear stress ratio value ( d/[1]’v0) equal to 0,207 through the critical domain of tailings susceptible to liquefaction;

 • In the hypothesis of only static loading, the (FS)triggering values varied between 1.14and1.36, indicating that CFSR tailings are unlikely to liquefy;

• Considering a rapid rise of the phreatic line through the tailings deposit reaching the toes of the intermediate rising dykes, with complete saturation of tailings layers susceptible to iquefied, a post-triggering analysis indicated that the flow failure susceptibility of the Dam I is low even under a such critical loading event;

 • The conclusions of these analyses, in addition to laboratory testing program results and based on rigid management procedures adopted in field, demonstrate that Dam I constitutes a safety structure against mechanisms from liquefaction-induced failures;

 • Although the Olson (2001) and Olson & Stark (2003b) liquefaction analysis has been proposed mainly for cohesionless soils, the methodology is consistent and suitable for preliminary analyses of liquefaction potential in tailings deposits (generally relatively low-density materials with a high degree of saturation), particularly upstream tailings dams.”

Within the text Pirete &Gomes indicate that they made a series of maintenance and control measures  on drainage to reduce or avoid conditions associated withe set up to possible static liquefaction  No comment on whether  those were monitored and remained effective.in keeping contents from reaching conditions which could approach critical thresholds for static liquefaction. 

Vale has not commented specifically on how either the thesis or its further examination in Pirete & Gomes was relied on but public statements suggest that it may have been taken as gospel by Vale’s Board and Senior decision makers. That reliance may explain the strength of their repeated assertions that the dam was in excellent condition pre failure and also why they were unvoved by the TUL SUD September 2018 report which we present below. As we set forth in the introduction to this series of posts on static liquefaction reliance on these two studies would not satisfy “due diligence” whether or not government accepted it as evidence of stability and of non suceptibility to failure by static liquefaction.

The opinion of non susceptibility was with reference to the very important “critical threshold” work Pirete an dthen Pirete and Gomes were pursuing. Other experts including Dr. Andrew Fourie are still pursuing this critical threshold . So in 2010, 2013 the Pirete& Gomes was neither proven nor wodely accepted. I was and is an important step towards useful loss prevention measurers with doable field applications that could be effective in continual assessment of static liquefaction risk.

Pirete and Gomes are to be thnaked and recogized for this important work but it would not have been appropriate for Vale to relyon on their fnding of non siceptibility to static liquefaction failue in the context of this reserach as an assurance of non suceptibility to staticliquefaction. The recommendations made are inalinement with best practice and are ckearly key to preenting a static liquefaction failure.

It appears that Vale did accept and act on those recommendations to improve drainage but the 2018 Tul -Sud analyzed below includes a long list of drainage problems that increase the risk of static liquefaction failure.If Vale relied on Pirete Gomes assurance that the dam could withstand even a sudden saturation without succumbing to static liquefaction that would not be “responsible practice”. It is not clear at this time, though, what reliance Vale actually did place on the Piretet& Gomes finding of low risk of static liquefaction.

On a simple “facts as known” basis it seems indisputable the no factor of safety approaching or exceeding 1.5 ever existed for dam1. Per broadly established best guidance and widely accepted best practiceathat was a minmum standrd for a dam classifed as high risk as this dam was by Brazils own standards. It would seem to absurd to accept a finding oflow suceptibility to failure by static liquefcation given measured FS of >0 ( Tul Sud 2018) to 1.14 ( Pirete Gomes). 

That range indicates not stability for anything other than a small berm but a state of essentially “still standing”applied to any tailings dam of any construction type. As further characterized by our esteemed Board member Geophysicist Dr. David M. Chambers,would be extremely vulnerable to to any advserse conditions including a severe weather event. 

Vale has announce retention of an expert panel to investigate cause of failure and hopefully Vale’s scope for that work will provide for that consideration of what information Vale assembled and relied on.

This post is the sole work and responsibility of Lindsay Newland Bowker,Executive Dirrector World Mine Tailings Failures based on input from WMTF experts including David M. Chambers and Steven Emerman. It is posted inviting and pending their peer review and review by other experts including WMTF compiler Roberto Rodriguez and coeague Andy Furie two tof the top peopl ein the world on static liquefaction. It is WMTF policy that all posts be peer reviewed and be noted as pending peer review where that has not occurred. Lindsay Newland Bowker, Stonington Maine February 13, 2019.

DOWNLOAD Full article

2018 TUV-SUD Assessment Not Able To Confirm Stability

Our Chief Compiler, Environmental Consequences, and Vice Chair of our Board of Directors, Dr. Steven Emerman, has translated and completed a summary of the much cited TÜV SÜD 2018 dam safety audit. We thank Steve for undertaking this time-consuming work and for his superb summary and analysis of that key document.

Two key parts of Dr. Emerman’s summary are presented below. The full report can be downloaded immediately below those excerpts. Please feel free to contact Dr. Emerman directly by email at SHEmerman@gmail.com


Our information services are all free of charge and none of us take any remuneration. We do need and would appreciate donations to support the technology required to develop and present this information. Checks can be made payable to World Mine Tailings Failures and sent to our Treasurer, John Zacahary Steed, Esq., at  P.O. Box 386, Blue Hill, Maine 04614. We are a nonprofit corporation incorporated under Maine law.


TÜV SÜD recommended correction of all of the above shortcomings in dam maintenance and data collection. In addition, they recommended more efficient conveyance of water (using troughs and pumps) from Dam I into the overflow system. Finally, they recommended the installation of micro-seismometers and a study of the foundation downstream from the dam in order to better understand the potential triggers for liquefaction.


Various news articles have been saying that TÜV SÜD certified the dam as “safe.” This is the strongest statement that TÜV SÜD said that could have suggested that the dam is safe: “Notaram que por vezes o método de Spencer forneceu FS maiores e menores que o exato. Os resultados mostrados por aqueles autores indicam que um fator de segurança superior a 1,05 cobre um possível erro envolvido no método de cálculo utilizado. Conclui-se que a Barragem I se encontra estável quanto à liquefação do rejeito, no cenário de instabilização sob a condição não-drenada, com FS > 1,05 ao serem considerados valores médios para a razão de resistência não-drenada do rejeito saturado.”

My English translation is: “They noted that sometimes Spencer’s method [for calculating factor of safety] provided FS [factor of safety] larger and smaller than the exact one. The results shown by those authors indicate that a factor of safety higher than 1.05 covers a possible error involved in the calculation method used. It is concluded that Dam I is stable in relation to the liquefaction of the tailings, in the scenario of instability under the undrained condition, with FS > 1.05 when considering average values for the undrained resistance ratio of the saturated tailings.”

This is my interpretation of the above statement: TÜV SÜD said that the dam was stable, meaning the factor of safety was greater than one, within the error of their method of calculation. But the real error is the lack of complete and exact knowledge of the physical properties of the mine tailings. For that reason, the Dam Safety Guidelines of the Canadian Dam Association recommend a factor of safety of at least 1.5. Civil engineering structures are not supposed to be built or maintained at a point just slightly safer than the point of failure. In summary, TÜV SÜD did not claim that the dam met generally-recognized dam safety guidelines. They just said that they were certain that the factor of safety was greater than one.