This is a very distinguished panel but as with all such commissioned works it is contractually limited to technical causes of failure and does not address the quality and competence of the Vale geotechnical team or the consultants and contractors involved over the life of the TSF or otherwise address prevailing practice or best practice.
It makes the point that the particular characteristics of the tailings in combination with an unsuitable design allowed the TSF stack to hover in a continuous state of not fully understood or clearly recognized vulnerability over its entire operating life. Applying standard tests and following standard guidelines for stability analysis and other assessment produces false and misleading results which on the whole wrongly state the stability condition and state of the tailings mass. In one test the panel themselves did in the laboratory the tailings showed a compliant 1.5 factor of safety in undrained conditions immediately pre-failure.
This mineralogy is significantly different from most natural soils that are predominately composed of quartz (i.e., silica-based minerals). The mineralogy also is significantly different from the soils used to develop the traditional empirical correlations used to interpret in situ tests, such as the CPTu, which would indicate that these empirical correlations may not provide reliable results in these tailings. p16
Historical stability calculations using LEM performed prior to the failure that used an undrained strength ratio of su/σ’vo = 0.26 calculated an FS close to 1.0.( TUV SUD 2018) However, if the FS was very close to 1.0, the dam would be expected to show various signs of distress, such as cracking and deformations, because the materials must strain (deform) to reach their peak strength. The fact that the dam showed no signs of distress prior to failure suggests that the results of traditional LEM were misleading. P59Re-analysis of the stability of Dam I using the peak undrained strength values determined from laboratory testing performed by the Panel, which reflect the additional strength due to bonding, resulted in a FS close to 1.5. However, this ignores possible regions of high shear stress where the local FS can be less than 1.0. The additional strength due to bonding essentially allows a steeper slope to be constructed without signs of distress, even though it could be potentially unstable
Bonding here refers to an apparent tendency for iron particles to almost immeditaely bond with other particles creating a brittleness that adds a strength easily lost in un drained conditions ( the chronic state of the Brumadinho dam 1) resulting in sudden abrupt failure under only a very small increase in strain. Brittleness is long understood to be a characteristic associated with abrupt and complete flow slides and is currently frequently referenced in connection with oil and gas wastes .
The overall conclusion is that it was something called a “creep rupture” ( a cumulative increasing strain within the mass deriving principally from the unusual weight of the tailings). In the case of Brumadinho the additional strain was brought on, according to their modeling, by increasing saturation of the previously unsaturated layer above the chronically , critically high water table within the structure.
The report settles the widely held speculation that something more could have been done to prevent failure because the main conditions of concern were known pre-failure. The panel specifically noted that little change had occurred in the water table over Vale’s focused and intensive 3 year effort to de water due to continuous influxes and the poor hydraulics of the stack itself .
The panel confirms that Vale did conduct a stability analysis for every raise, a basic of good practice. But the limitations in “sophistication” , suitability to actual conditions and reliability of results on which Vale in house and consulting engineers based their conclusions and decisions are noted.
Vale’s geotechnical team insisted on one model for all stability analysis which is not correct procedure. Tuv Sud complained about that in their 2017 stability report. Vogbr complained about Vale requiring a specific model and supplying the data for that model at Samarco. It is important to select the correct method for liquefaction analysis for the conditions, stress and loading concerns at that moment in time and having the technical capacity at hand to apply and interpret the results of those analytic methods. One model for all situations, as Vale required does not satisfy that requirement. Further, the specific model Vale insisted be used for all government required stability analysis is only appropriate for assessment of prospective changes in the structure, for example dam raises. It is not appropriate for assessing actual conditions of a structure at any point in time.
The panel, because of the contractually imposesd limitations on “technical causes” does not mention Vale’s imposition of a particular model on all its consultants for all official stability analysis . The panel due to this limitation on its inquiry was not in a position to offer comment on Vale’s policy of one model for all situations or to comment on whether a licensed engineer functioning as an independent expert ( as required under Brazilian Law) should agree to use of a model it does not feel is suitable to actual conditions.
TUV SUD, after failure refused to certify any more dams under Brazilian Law where its checklist risk of failure lead to a proforma result of “little risk of failure”. No engineer should be required to put its name and signature on a conclusion that is not its own.
The report in general does not conflict with any of the analysis we had posted at World Mine Tailings Failures. It does though add many new insights and additional data which we are in the process of including in our failure narrative.
The panel specifically agrees that the design of the TSF itself was a major never corrected contributor to the failure. The panel only cites its steepness and the type of construction (upstream) and its poor hydrology without applying what existed as widely recognized standards e.g. Vick’s that upstream should not be used if the tailings to be deposited are not 40% to 60% sand. or Davies et. al. on the cardinal rules for safe upstream construction ( size, slope, rate of raise etc.) The report adds a fact that had not been brought out in previous analysis associated with the dam’s initial design and history: that the starter dam blocked a creek that had been draining up stream groundwater resulting in a recurring and continuous back up of groundwater within the structure ( in addition to the inflow we note from the higher draining elevations to which the structure attaches.
The panel adds that a centerline setback undertaken pre -Vale early in the life of structure contributed to failure in that future raises built up from a foundation of finer less strong tailings.
It would be useful and fruitful to commission this exact panel to address all our other questions about whether Vale and its consultants executed a standard of care that met or exceeded prevailing practice. Vale seems to have exceeded prevailing practice. They at least had an in house geotechnical staff and that staff put considerable store in mandated reporting and standards under Brazilian Law . Overall and on tailings Vale is right in the middle of the global pack on the various elements of the Responsible Mining Index 2018.
It would greatly contribute to the ongoing Global Tailings Review to have this same panel offer a civil society/best practices perspective on this same data.