Content uploaded by Jiri Herza
Author content
All content in this area was uploaded by Jiri Herza on Jun 18, 2019
Content may be subject to copyright.
A consequence-based tailings dam safety framework
J. Herza
Technical Director, GHD Pty Ltd, Perth, Australia
M. Ashley
Senior Engineer, GHD Pty Ltd, Perth, Australia
J. Thorp
Senior Engineer, GHD Pty Ltd, Perth, Australia
A. Small
Senior Geotechnical Consultant, KCB, Canada
ABSTRACT: Despite the development of tailings dam safety standards, guidelines, bulletins, risk
assessment tools and management tools, tailings dam failures with high consequence have occurred at
a similar rate over the last ten years as they did before. At the time of writing this paper (January 2019)
another tailings dam failed in Brazil resulting in a large number of fatalities. The authors of this paper
concluded that the main underlying reason for recent failures of large tailings dams is a systematic
failure to recognize the potential geotechnical hazards, their consequences, trigger mechanisms and
the appropriate method of stability analysis. The individual prescriptive measures currently applied
during some stages of tailings projects to control the risk, such as scaling the design loads based on
the consequence, are not considered sufficient to overcome the systematic deficiency. Instead, a ho-
listic dam safety management system overarching all phases of tailings dam projects from planning to
closure is required, including the management of the facilities. The authors suggest the widely used
and understood consequence-based principles should be extended to cover the entire life span of tail-
ings dams, including the dam safety management system. This approach is an extension of the tailings
dam safety frameworks presented in MAC (2017) and Morgenstern (2018) and may provide the min-
imum requirements for all relevant aspects of the tailings dam safety framework. The suggested ap-
proach would respect the risks posed by the tailings dam while taking into account the economical
aspects of the project. The approach should considered by the ICOLD Technical Committee for Tail-
ings Dams and Waste Lagoons and the International Council on Mining and Metals.
RÉSUMÉ : Les outils d´évaluation et de gestion, les barrages de résidus miniers ayant de lourde
conséquence se produisent à un rythme similaire au cours des dix dernières années. Au moment de la
rédaction de cet article (Janvier 2019) encore un barrage de résidus miniers s´est écroulée en Brazil
causant un vaste nombre de victimes. L´auteur de cet article a conclu la raison de base la plus impor-
tante des défaillances des larges barrages de résidus miniers est le manque systématique de reconnais-
sance des risques potentiels géotechniques, de leurs conséquences, des mécanismes déclencheurs ainsi
que des méthodes appropriées de l´analyse de stabilité. Les mesures normatives individuelles qui sont
actuellement appliquées lors de certaines étapes des projets de contrôle de risque des résidus, telles
que la réduction des charges de projet en fonction des conséquences, sont considérées comme insuffi-
santes pour surmonter la déficience systématique. Cependant, un système holistique de gestion de la
sureté des barrages de résidus miniers est demandé couvrant toutes les phases de la planification
jusqu´à la fermeture y compris la gestion des installations. L´auteur propose des principes fondés sur
les conséquences compris et largement utilisés d´être étendues afin de couvrir toute la durée de vie des
barrages de résidus, y compris le système de gestion de la sécurité des barrages. Cette approche est
une extension du cadre de sécurité débarrages de résidus miniers présentée au MAC (2017) et Mor-
genstern (2018) et peut fournir les exigences minimales à tous les aspects du cadre de sécurité des
barrages de résidus miniers. L´approche suggérée respecterait les risques posés par les barrages de
résidus miniers tout en tenant compte l´aspect économique du projet et devrait être prise en considé-
ration par le Comité Technique ICOLD pour des Barrages de Résidus Miniers et les Lagons de déchet
ainsi que le Conseil International de L´industrie Minière et des Métaux.
1 INTRODUCTION
The consequence-based dam safety principle is an approach to dam safety that has commonly
been used for the design of tailings dams. In this paper, the authors propose that the consequence-
based dam safety principle be extended to provide a framework to manage all of the life phases
of a tailings dam. It is a contemporary version of the “traditional” approach to dam safety, as
defined by ICOLD Bulletin 154, and it differs in some aspects from the risk-based approach that
has been gaining momentum in the industry. Note that in this paper, the terms consequence cate-
gory, hazard rating and hazard category are considered synonymous.
A holistic consequence-based approach aims to address the commonly inconsistent and generic
guidance currently in place, which makes it difficult for dam owners, regulators and designers to
agree on the dam-safety approach to be adopted for a project. This approach covers all life phases
of a dam, including the governance considerations such as nominating clear responsibilities and
requirements for change management.
This paper aims to contribute to the development of an international tailings dam safety man-
agement approach that can be applied to all phases of tailings dam projects. The approach could
be considered an advance of the traditional approach that could be adopted dam practitioners
worldwide.
2 THE EVOLUTION OF TAILINGS DAM SAFETY GUIDANCE AND LEGISLATION
Tailings dams safety guidance and regulation has evolved over time. By the mid 1960s, ge-
otechnical risks and the need for improved safety of tailings dams were recognised, culminating
in regulations developed by British Columbia, Canada in 1971. This regulation was preceded by
the Chilean decree (1970), which banned upstream tailings dam construction as a response to a
major earthquake in March 1965, in which 11 tailings dams were either badly damaged or failed,
the most famous being El Cobre.
A number of significant updates to tailings (and water) dam safety regulation have occurred in
the decades since. Such updates were commonly in response to major dam failures; such as the
Buffalo Creek Dam and Teton Dam in the 1970s and the Stava tailings dam failure in 1985. A
sub-committee was formed by ICOLD in 1976 to address tailings dam safety issues and has de-
veloped a design manual (1982) and technical bulletins.
The more recent failures of the Mt Polley Mine (Canada) tailings dam in 2014 and the Fundão
(Brazil) tailings dam in 2015 has prompted many commentators and agencies to address the issues
and provide guidance.
Presently, in Canada and Australia, dam safety is mostly legislated at the provincial/state level,
and the legislation commonly refers to guidelines produced by Canadian Dams Association
(CDA) and Australian National Committee of Large Dams (ANCOLD) for the technical require-
ments. Some states (such as New South Wales and Western Australia in Australia) and mining
companies (such as Rio Tinto) also produce their own tailings dam safety guidelines. Further
guidance is also available from a number of other institutes.
Morgenstern (2018) provides a history of the development of tailings (and water) dams safety
guidance and regulation, and a summary of the more noteworthy recommendations. ICOLD Bul-
letin 167 (preprint) provides a comprehensive review of current legal and regulatory arrangements
for safety of dams.
A number of conclusions were drawn in ICOLD Bulletin 167, including “There is a wide range
of approaches to dam classification” and “In most countries with large dams, the state administra-
tion and the dam owners follow certain dam safety management procedures. However, the levels
of complexity differ from country to country”.
Tailings dam guidance continues to be updated by CDA, ANCOLD, Mining Association of
Canada (MAC), International Council on Mining and Metals (ICMM), and ICOLD amongst other
national and internal bodies. The authors also understand that the World Bank are currently pre-
paring their own dam safety guidance for mining projects financed by the World Bank.
In November 2016, the ICMM published a position statement that set out ICMM members’
approach and principles for the governance of tailings storage facilities (TSFs) for the mining and
metals industry to minimise the risk of catastrophic failure. The ICMM key principles, listed be-
low, are supported by the authors of this paper as key commitments to be included in any tailings
dam safety framework.:
1. Accountabilities, responsibilities, and associated competencies are defined to support appro-
priate identification and management of tailings storage facilities risk. (ACCOUNTABILITY
and COMPETENCY)
2. The financial and human resources needed to support continued tailings storage facilities
management and governance are maintained throughout a facility’s life cycle. (CON-
SISTENCY)
3. Risks associated with potential changes are assessed, controlled, and communicated to avoid
inadvertently compromising facility integrity. (CHANGE MANAGEMENT)
4. Internal and external review and assurance processes are in place so that controls for facilities
risks can be comprehensively assessed and continually improved. (REVIEWS)
There is an opportunity for a common approach to tailings dam safety through the application
of consequence-based principles as presented in this paper, adopting the commitments in the
ICMM position statement. It is important to acknowledge that many others, including Morgen-
stern (2018) are also advocating for change, with widely varying views on what this should entail.
3 CURRENT APPROACHES TO DAM SAFETY
3.1 Traditional approach
ICOLD Bulletin 154 states “The traditional approach to dam safety analysis (often called de-
terministic or standards-based) begins with the potential hazard or consequence classification and
follows with calculations to ensure that the dam system conforms to a deterministic set of princi-
ples, rules and requirements (traditionally called design standards).”
The main limitations of current “traditional approaches” are the lack of specific guidance on
the agreed approach for all phases of the dam project and for items beyond “design standards”.
For example, guidance on the investigation and design phases is limited and minimum require-
ments are generally not available. The dam safety management systems are typically not linked
to the dam consequences. The importance of hazard rating or consequence category is often ref-
erenced, however the specific applications are not provided.
For example, ICOLD Bulletin 139 “Improving tailings dam safety” states that the hazard rating
influences design factors such as acceptable levels of seepage or levels of factors of safety for
stability, considerations of the reliability of the site and tailings material investigations, and the
reliability of data being used in the design.
The significance of hazard and risk in Dam Safety Management was also recognized and
ICOLD Bulletin 139 states “The amount of effort and resources that an owner should put into a
dam safety program is determined by each dam’s hazard and risk” and “The Hazard Category …
is fundamental to the nature and cost of a dam safety program.” However, the Bulletin does not
provide specific guidance on the approach for different hazard ratings.
A consequence-based dam safety framework has been introduced in the Australian National
Committee on Large Dams (ANCOLD) Guidelines on Dam Safety Management (ANCOLD,
2003). The guide is prescriptive with tabulated, consequence-based requirements for surveillance,
monitoring, comprehensive surveillance review, and dam safety reviews. However, requirements
for the investigation, design and construction phases are much less prescriptive, and generally
only distinguish between very low to low hazard category dams, and significant or higher hazard
category dams.
The Canadian Dam Association (CDA, 2007) also has a consequence-based dam safety frame-
work and states as one of their principles “The standard of care to be exercised in the management
of dam safety shall be commensurate with the consequences of dam failure.” As with the AN-
COLD guidance, the CDA guidance describes the requirements for design criteria and reviews,
but has only limited information on investigation and construction phases.
There are many schemes and frameworks for classifying dams in accordance with the potential
consequences of failure. Some systems use three levels (low, medium, and high) and others use
up to seven levels (very low to extreme). For the purpose of this paper, the terms low, significant,
and high are used. Failure of a low classification dam would not result in loss of life, environ-
mental damages, loss of infrastructure, or have a significant impact on the mine owner. Failure of
a high consequence dam would involve the loss of one or more lives, major environmental dam-
age, major damage to infrastructure, and major impact on the mine owner. Significant classifica-
tion dams are between low and high. A classification scheme that can be used internationally and
adopted by the many jurisdictions is beyond the purview of this paper, but the use of the terms
low, significant, and high are sufficient to support the concept that is proposed in this paper.
3.2 Risk-Based Approach
The risk-based approach is based upon a pre-selected level of tolerable risk and all elements of
a project are designed such that the overall probability of failure does not exceed the pre-selected
level. The process is described in the ANCOLD Guidelines on Risk Assessment (ANCOLD,
2003) and referenced in other ANCOLD guidelines including the Guideline on Tailings Dams
(ANCOLD, 2012). The risk-based approach divides the dam into components such as embank-
ments, spillways and outlet works, evaluates the likely failure modes of each component and the
initiating events that might lead to failure. This process includes consideration of the probability
of each event (such as a design flood) occurring and therefore considers probabilities in the con-
text of the required design life of the component. The process then considers the behaviour under
each loading and estimates the probability of failure, sometimes by reference to historical fre-
quencies of similar failures. Event trees (failure pathways) are used to calculate the individual and
the total risk of failure.
The tolerable risk is usually determined based on the consequences of the dam failure and is
typically based on loss of life and economic damages to third parties. There is no tolerable risk
guidance for environmental losses. The higher the failure consequences the lower the tolerable
probability of failure. The ANCOLD Guidelines on Risk Assessment (ANCOLD, 2003) recog-
nized that the tolerable risk cannot be reduced to zero and the suggested tolerability level is trun-
cated at 10-6 for new dams and major augmentations as shown in Figure 1. It is noted that at the
time of writing this paper, the ANCOLD (2003) Guideline was under review.
Figure 1 ANCOLD Societal Risk Guideline (ANCOLD, 2003)
The risk-based approach is based on a certain level of tolerable probability of failure, in the
case presented later, with respect to loss of life. The limit of tolerability represents the tolerable
likelihood of failure to the community when loss of lives is a potential consequence. Figure 2
shows the annual probability of death between 2008 and 2018 taken from the Australian Bureau
of Statistics Report 3302.0 (ABS, 2018) and the limits of tolerability suggested by Guidelines on
Risk Assessment (ANCOLD, 2003). As can be seen, the tolerable limit for the probability of a
single death is 1 x 10-4 per annum, which aligns closely with the lowest annual probability of
death across Australia. The broadly tolerable risk imposed by an external party on a number of
people (more than 100) is two orders of magnitude lower, being for a risk in the wider community
by an external party. For an existing dam, the existence of which is presumably known by the
public, the ANCOLD acceptance levels are one order of magnitude higher. This also considers a
practical acceptance of the cost effect associated with upgrading numerous dams should the more
stringent criteria apply.
The selection of a tolerable risk of failure upon which the risk based principle is built is a very
difficult and widely discussed topic. While ANCOLD supports the risk-based approach, other
practitioners and organizations dismiss the tolerable risk principle.
Key limitations to the risk-based approach are that some of the decisive factors cannot be quan-
tified by probability, the required process can be costly and time-consuming for dam practitioners
and fall-back positions are often adopted. Lack of experience, technical and administrative com-
petence of the contractor building a dam or within the organization responsible for the dam, neg-
ligent and insufficient maintenance or inadequate procedures for operation are decisive factors in
dam safety, but they cannot be considered by a probabilistic assessment.
Similarly, a quantitative determination of the area impacted by the dam failure and the conse-
quences to society, the environment and the dam owner is very difficult and there is no interna-
tionally accepted framework for consequences evaluation. Hence, both the probability and the
consequence of the risk-based approach are not consistently well defined.
An owner can use the risk assessment approach to support internal decisions with their own
framework for assigning likelihood and assessing consequences. This is consistent with the paper
by Morgenstern (2018) that advocates for the use of risk assessments. But the challenge is then
conveying the results of this risk assessment approach and consequence framework beyond the
owner’s organization where communities of interest may have different perspectives on how the
consequences are assessed.
Figure 2 Average background risk (Bureau of Statistics, 2018) and tolerability levels
4 THE NEED FOR CHANGE
Despite the development of the tailings dam safety standards, guidelines, bulletins and risk
assessment and management tools, tailings dam failure records, such as the collation of failures
by WISE, shows that failures with high consequence have occurred at a similar rate over the last
ten years as the ten years before. At the time of writing this paper (January 2019), another tailings
dam had failed in Brazil resulting in a large number of fatalities.
It is noted that the statistics maintained by WISE (www.wise-uranium.org/mdaf.html) are in-
complete, as indicated by the website itself and Li et al. (2016), and many more tailings dam
failures have likely occurred in the last several decades.
Morgenstern (2018) noted that understanding of undrained failure mechanisms leading to static
liquefaction with high consequences is a factor in about 50% of the failure cases. Inadequacies in
site characterization, both geological and geotechnical, is a factor in about 40% of the failure
cases.
Herza et al (2017) demonstrated that adoption of a satisfactory factor of safety value does not
necessary prevent a slope failure and concluded that the minimum FoS does not directly account
for fundamental misunderstandings of the material behaviour, omissions of critical materials and
conditions or poor construction practices.
The tailings dam failures of the Mt. Polley TSF in 2014, the Fundão Dam in 2015 and Cadia
Mine TSF in 2018 are recent examples of failures caused by omission, misunderstanding, misin-
terpretation, or misrepresentation of the behaviour of critical materials.
It is alarming that the slope failure of one embankment at Cadia Mine TSF occurred in 2018,
though with no significant consequence, despite the well advanced dam safety culture of the
owner, involvement of a reputable engineering firm and stringent dam safety legislation in NSW,
Australia.
It appears that the main underlying reason for recent failures of large tailings dams is a system-
atic failure to recognize the potential geotechnical hazards, their consequences, trigger mecha-
nisms, and the appropriate method of analysis for stability. Individual prescriptive measures cur-
rently applied during some stages of the project to control the risk, such as scaling the design
loads based on the consequence, are simply not sufficient to overcome the systematic deficiency.
To address this issue, a holistic tailings dam safety management system overarching all phases of
the tailings dam project from planning to closure is required.
5 TAILINGS DAM SAFETY FRAMEWORK
It is recognized that each tailings dam project is different. However, each tailings dam project
progresses through similar stages and each stage has to be adequately managed. The managerial
requirements for each stage are similar even for different projects with vastly different conse-
quences. These requirements include the need for a suitably qualified personnel and adequate
practices and documentation for investigation, design, construction, operation and closure.
In his excellent lecture, Morgenstern (2018) recommends a Performance-Based, Risk-Informed
Safe Design, Construction, Operation, and Closure (PBRISD) system for tailings projects. That
system is built upon accountability achieved through multiple layers of review, recurrent risk
assessment, and performance-based validation from construction through closure.
The PBRISD presents personnel who should be involved in the tailings dam safety program
and the activities to be completed at various stages of the tailings project. These items are irre-
spective of the tailings dam consequence and it further develops the management approach rec-
ommended by MAC (2017). A simplified version of a framework of the key personnel and activ-
ities involved in the tailings dam projects is shown in Table 1. The definitions of the roles used
in this simplified example are as follows:
Executive Tailings Manager (ETM) is a person appointed by the owner of the TSF, to own,
be accountable and responsible for setting up and implement a tailings safety management frame-
work. The ETM assigns accountabilities and responsibilities for the personnel involved in the
tailings dam safety management.
Responsible Manager (RM) is a person appointed by the owner to be responsible for the
implementation of the tailings dam safety management and to be accountable for the safety of a
tailings facility. The RM, or their delegate, must be based on site during the construction and
operational phases of the facility.
Responsible Engineer (RE, also referred to as Engineer of Record) is a person or a party
appointed by the RM to provide verification that the design intent of the tailings dam is met
throughout all stages from planning to closure and maintaining the tailings dam documentation.
Preferably, the RE will be the same individual through all stages of the dam life to retain account-
ability, responsibility and preserve the site specific knowledge.
Design Engineer (DE) is a person or a party appointed by the RM to prepare the Design
Documentation and provide construction oversight and construction QA. The DE may or may not
be the same entity as the RE.
Independent Reviewer (IR) is a person or a party appointed by the ETM to provide inde-
pendent reviews of the tailings dam and check that the design intent is in accordance with the
current engineering practice and is being met throughout all life stages of the tailings dam.
The definitions of the documentation shown in Table 1 are as follows:
Storage Management Plan (SMP) is a live document outlining the short term, medium
term and life of the project plan for the tailings dam including the facility closure.
Design Documentation (DD) is a set of studies, reports, calculations, drawings, specifica-
tions, closure plans and other documents, which clearly define the design intent of the tailings
dam and present assumptions and conditions that are critical for the dam integrity and safety. DD
must be available for the original tailings dam and all modifications.
Construction Documentation (CD) is a set of reports, calculations, quality assurance and
quality control forms and certificates and other documents, which describe the construction ac-
tivities and document that the construction was carried out and completed in accordance with the
design intent as presented in the DD. CD must be available for the original tailings dam and all
modifications to the tailings dam.
Operations and Maintenance Manual (OMM) is a document that outlines the operations,
maintenance, surveillance, monitoring and emergency procedures for the tailings dam and defines
roles and responsibilities for the dam operations. The OMM forms part of the DD but must be
reviewed periodically and updated as required. It must be updated after each stage of construction
based on the CD.
HAZID is a hazard identification process and HAZOP is a hazard and operability process.
Table 1 Tailings Dam Safety Framework – Key activities matrix
Executive Tailings
Manager (ETM)
Responsible Manager (RM)
Responsible Engineer (RE)
Independent Re-
viewer (IR)
Planning
Sets up, owns and
implements of the
tailings safety man-
agement frame-
work. Defines ac-
countabilities and
assigns responsibil-
ities for personnel
involved in frame-
work, nominates
IR.
Initiates and oversees planning activi-
ties, appoints RE and approves TMP.
Prepares (or oversees preparation of) and
endorses the TMP.
Reviews TMP.
Design
Oversees the design process, appoints
the Design Engineer (DE), facilitates
HAZOP*, approves DD.
Prepares (or oversees preparation of) and
endorses DD, facilitates HAZID*.
Reviews DD, partici-
pates at HAZID* and
HAZOP*.
Construction
(inc. closure)
Appoints a contractor, oversees con-
struction activities, approves CD and
OMM.
Provides oversight to the construction activ-
ities with respect to conformance with the
design, maintains (or oversees the mainte-
nance of construction QA, if RE is not the
DE), endorses CD, updates (or overseas the
update of) OMM.
Reviews CD and
OMM.
Operation
Appoints Dam operators, ensures that
the requirements of the Operation and
Maintenance Manual are met.
Verifies that the Dam is operated in accord-
ance with the design intent and the Opera-
tions Manual, provides training for the op-
erators.
Provides inspections
and audits as re-
quired in the OMM
and as requested by
the ETM.
6 APPLICATION OF CONSEQUENCE-BASED PRINCIPLES
6.1 Key Uncertainties in Dam Projects
ICOLD Bulletin 154 states: “In dam safety assessment, it is not possible to have complete and
perfect knowledge of the condition of a dam or its performance and, as such, knowledge uncer-
tainties permeate the dam safety assessment and management processes” (ICOLD Bulletin 154).
As outlined in CDA (2013), the appropriateness of factors of safety depends on the conserva-
tism of the assumptions made regarding stratigraphy, strength of materials, pore-water pressure,
and loading.
To understand how structures are expected to perform and what level of deviation from the
normal condition is tolerable, dam safety analyses should consider the full range of applicable
conditions. Design, construction, and operation should all be considered in the analysis to ensure
that the intent of the design has been achieved (CDA, 2013).
The consequence-based principles should therefore aim to account for and reduce the uncer-
tainties in all of the tailings dam life-phases to acceptable levels with respect to the tailings dam
consequence.
6.2 Planning and Investigation
During the planning and investigation stage, a key uncertainty is the foundation materials, their
distribution and potential changes. The final structure and the associated loading conditions may
also not be fully defined in the planning and investigation stage, especially in tailings dam pro-
jects.
Insufficient understanding of the foundation characteristics and the ultimate loading conditions
may lead to catastrophic failures. Therefore, it is crucial that the ultimate consequence of the
tailings dam is considered and that the planning and investigation is conducted accordingly.
Eurocode 7: Geotechnical Design (under review) applies a limit state philosophy to the inves-
tigation and design of geotechnical structures including embankments. This code may be used to
establish the minimum recommended standards for the planning and investigation of tailings
dams and the qualification and experience of the specialist involved, with respect to their conse-
quence category.
In 2016, the Association of Professional Engineers of British Columbia issued a guidance doc-
ument entitled: “Site Characterization for Dam Foundations in BC” that indicates that the conse-
quence classification should be considered when developing the site characterization program.
However, specific guidance is not provided, and this should be included.
6.3 Design and Stability Assessment
Despite the development of new analytical approaches and advances in soil behaviour under-
standing, the minimum Factor of Safety (FoS) principle still remains the most common approach
to the tailings dam stability design. FoS is commonly used to express the safety margin of slopes
on embankment dams. The minimum acceptable FoS for dam design were anecdotally determined
in the USA in mid-20th century by back-calculating the FoS of existing dams. It was found that
a FoS of 1.5 provided sufficient contingency and was generally considered acceptable. The min-
imum required FoS provides the basis for the deterministic, fall-back method of analysing dams.
A formal set of minimum FoS for various loading conditions was presented in the USACE
Design Manual (USACE, 1970) and the same FoS values, with minor modifications, appear to
have been adopted by many organisations including USBR (1987), Norwegian Geotechnical In-
stitute (1992), BRE (1999), ANCOLD (2012), Canadian Dam Association (2007), and others.
The impact of uncertainties and the reliability of input values is well recognised and USACE
(1970) recommended that these uncertainties be considered in selection of the appropriate FoS.
The revised USACE design manual (USACE, 2003) stated that “Two of the most important con-
siderations that determine appropriate magnitudes for factor of safety are uncertainties in the con-
ditions being analysed, including shear strengths and consequences of failure or unacceptable
performance.” However, the minimum required FoS published by USACE and other authorities
in modern dam engineering do not actually take into account the potential consequence of the
dam failure or the uncertainties, but are solely based on the loading conditions.
Key soil and rock properties such as unit weight, shear resistance and others are determined
from field and laboratory testing and adoption of empirical relationships, which introduces some
levels of uncertainties and potential errors. As shown in Herza et al (2017), a small change of
input parameters can, in some circumstances, have a severe impact on the overall FoS.
With the lack of a clear definition of the FoS purpose, Herza and Phillips (2017) suggested that
an appropriate FoS should “ensure reliability of the dam design and to account for uncertainties
and variability of the foundation and the dam components parameters, uncertainties of the design
loads and limitations of the analysis used”. However, the minimum FoS does not directly account
for fundamental misunderstandings of the material behaviour, omissions of critical materials and
conditions or poor construction practices. The recent tailings dam failures at Mt. Polley, Canada
(2014) and the Fundão Dam, Samarco, Brazil (2015) are examples where the FoS did not suc-
cessfully account for unknowns, nor provide sufficient reliability in the design. In these cases, the
actual failure mechanisms were not considered in the design (Morgenstern et al, 2015), or certain
materials were misidentified (Morgenstern et al, 2016).
The consequence-based principle in the design of tailings dams should require greater rigor for
stability assessment of higher consequence tailings dams and not just from the loading condition
perspective. The consequence category should also be considered in the selection of the analytical
tools, simplifications and conceptualization of the material behaviour and selection of the mini-
mum required FoS. Higher consequence category tailings dams may warrant greater use of stress-
strain modelling to supplement the limit equilibrium approach. Both drained and undrained, dila-
tive and contractive behaviour models should be considered, depending on the material charac-
teristics, to adequately assess normal, unusual and extreme loading conditions on site. The calcu-
lation of the factor of safety should be decoupled from the potential for a triggering mechanism,
especially for tailings dams, the stability of which can be controlled by contractive materials. For
high consequence tailings dams, it may be necessary to assume that there could be a trigger in-
ducing undrained shearing in contractive soils regardless of whether such a trigger mechanism is
identified.
6.4 Construction and Commissioning
Quality assurance, quality control and change management during construction of new tailings
dams and remedial works to existing tailings dams are crucial. If a tailings dam is constructed
under low quality standards and/or deviations from the design are made without understanding
and accepting the consequences of the deviations, the tailings dam safety could be compromised.
The authors of this paper are not aware of any specific international or national guideless for
construction quality management for tailings dams. However, such guidance would be valuable
and could provide minimum required standards for construction quality procedures, change man-
agement procedures and construction, supervisory and managerial personnel requirements. The
level of quality assurance, quality control, supervision and management should be selected based
on the consequence category.
A rigorous quality control program should be maintained throughout the site preparation and
construction period for significant and high consequence tailings dams. The as-built conditions
should be properly documented. Instrumentation data pertaining to the foundation, embankment,
and appurtenant structures should be reviewed by appropriately qualified engineers.
6.5 Operations and Closure
Catastrophic tailings dam failures are often preceded by warning signs, such as the appearance
of cracks in earth or concrete tailings dams and the discharge of turbid sediment-laden water
downstream of the dam, increasing pore water pressures. The warning time can extend to a period
of years, but in many cases warning time may be as little as hours or minutes. Properly planned
and executed inspections, combined with instrumentation monitoring, review and interpretation
and follow-up, can be highly effective in identifying developing tailings dam performance con-
cerns. The ANCOLD Guidelines on Dam Safety Management (ANCOLD, 2003) provides an
example of requirements for surveillance, monitoring, comprehensive surveillance review, and
dam safety reviews based on the consequence category of the dam.
However, even with a robust surveillance program, an early warning of certain failure mecha-
nisms including a static liquefaction may not be achievable. Therefore, if static liquefaction is a
possible failure mechanism and the consequences are high to extreme, there should not be reliance
on a surveillance program to guard against that failure mode. Instead the potential for static liq-
uefaction for high and extreme consequences tailings dam must be eliminated from the system.
7 CONSEQUENCE-BASED PRINCIPLES IN TAILINGS DAM SAFETY MANAGEMENT
While the activities and the tailings dam safety documentation outlined earlier in this paper are
applicable to all consequence category dams, the level of competencies and qualifications of the
key personnel and the details presented in the tailings dam safety documentation should be
driven by the consequence category.
The consequence-based tailings dam safety framework proposed in this paper could be ex-
pressed as an extension of the managerial framework suggested by MAC (2017) and Morgenstern
(2018). This safety management framework could be seen as three-dimensional (Figure 3) as fol-
lows:
•key activities matrix such as shown in Table 1 (front projection)
•the minimum requirements for tailings dam safety documentation driven by the conse-
quence of failure (side projection)
•the minimum required competencies and qualifications for the key personnel in the
dam safety management framework based on the consequence category (top projec-
tion)
The change management associated with the change of the consequence category of a given
tailings dam would then require filling the gaps between the minimum required levels of person-
nel and documentation for the respective consequence categories before and after the change.
Figure 3 Consequence–based Tailings Dam Safety Management Framework
For clarification, it is noted that the minimum requirements for the details presented in the
tailings dam safety documentation dictates the processes and methods that required to satisfy these
requirements. An example of the proposed consequence-based principles for the qualifications of
the Responsible Engineer, as defined earlier, is provided in Table 2. Table 3 shows an example
of the application of the consequence-based principle for setting up the minimum requirements
for the design documentation.
Table 2 Consequence-based minimum qualifications for Responsible Engineer (example only)
Consequence
Minimum Education
Minimum Experience
Minimum affiliations
Low
Bachelor degree in Civil
or Geotechnical Engi-
neering (or equivalent)
5 years - Dams (appro-
priate to the height and
type)
Member of relevant pro-
fessional organisation
(Engineers Australia, ICE,
etc.)
Significant
Bachelor degree in Civil
or Geotechnical Engi-
neering (or equivalent)
10 years - Dams (appro-
priate to the height and
type)
Member of relevant pro-
fessional organisation
(Engineers Australia, ICE,
etc.)
High*
Masters degree in Civil
or Geotechnical Engi-
neering (or equivalent)**
15 years international –
Dams (appropriate to
the height and type)
across all key areas*
Member of relevant pro-
fessional organisation
(Engineers Australia, ICE,
etc.) and industry organi-
sation such as ANCOLD,
ICOLD, BDS et. **
Notes: *Expert team assembled with minimum 3 persons to provide required experience across
all key areas, ** For at least one member of the Expert Team
Table 3 Consequence-based minimum requirements for Design Documentation (example only)
Consequence
Geological inputs
Geotechnical inputs
Stability Assessment
Low
Marked up regional geo-
logical map and cross sec-
tion of the dam and its
foundations.
Typical values confirmed
by field and laboratory test-
ing.
Simplified limit equilib-
rium assessment for
drained and undrained con-
ditions.
Significant
2D geological/geotechnical
model of the site.
Stress-strain characteristics
of materials taking part in
dam stability. Periodically
verified by field and labora-
tory tests.
Stress-strain analysis taking
into account generation and
dissipation of pore water
pressures.
High
3D geological/geotechnical
model of the site.
Stress-strain characteristics
of materials taking part in
dam stability. Periodically
verified by field and labora-
tory tests.
Fully calibrated stress-
strain analysis taking into
account generation and dis-
sipation of pore water pres-
sures, , including shear in-
duced pore pressures
The examples shown in Table 2 and Table 3 are included only to illustrate the application of
the consequence-based principles and are not to be taken as the definitive recommended guidance.
The selection of the minimum requirements for various stages of the tailings dam projects is for
the international tailings community to define and agree upon.
8 CONCLUSIONS
The authors suggest the widely used and understood consequence-based principles and gov-
ernance should be extended to cover the entire life span of tailings dams. Such a system may
overcome the systematic failure to recognize the potential geotechnical hazards, their conse-
quences and the trigger mechanisms, which the authors of this paper believe are the main reasons
for the recent catastrophic failures of tailings dams.
The suggested approach is an extension of the tailings dam safety frameworks presented in
MAC (2017) and Morgenstern (2018) which would provide a clarity for the owners, designers,
regulators and the wider community. The proposed framework would form a road map for tailings
projects from both the technical and managerial perspectives. This road map will refer to the
various existing bulletins and guidelines to provide specific requirements for specific aspects and
stages of the tailings project.
While tailings dam safety management processes, simplistically tabulated in this paper, are
similar for all tailings dams irrespective of their consequences, the level of competencies and
qualifications of the key personnel and the details presented in the tailings dam safety documen-
tation are proposed to be driven by the consequence category.
The suggested approach would respect the risks posed by the tailings dam while taking
into account the economical aspects of the project. The authors of this paper recommend the
suggested consequence-based tailings dam safety framework be considered by the ICOLD
Technical Com-mittee for Tailings Dams and Waste Lagoons and the International Council on
Mining and Metals.
9 REFERENCES
ANCOLD. Guidelines on Dam Safety Management, 2003.
ANCOLD. Guidelines on Risk Assessment, 2003.
ANCOLD. Guidelines on Tailings Dams, 2012.
Association of Professional Engineers and Geoscientists of British Colombia. Site
Characteriza-
tion for dam foundations in BC, V1.2, 2016.
Australian Bureau of Statistics. Report 3302.0 – Deaths, Australia, 2017, 2018.
BRE. An engineering guide to the safety of embankment dams in the United Kingdom. Building
Research Establishment, Garston, 1999.
CDA. Dam Safety Guidelines, 2007.
CDA. Dam Safety Guidelines, revised 2013.
European Committee for Standardization. Eurocode 7: Geotechnical design, 2004. Under revi-
sion.
Herza, J., Phillips, J. Design of dams for mining industry, ICOLD 2017 Conference, Prague,
Czech Republic, 2017.
Herza, J., M. Ashley and J. Thorp. 2017. Factor of Safety? – Do we use it correctly?, ANCOLD
2017 Conference, October 26-27, 2017. Hobart, TAS, Australia.
ICMM. Position Statement on preventing catastrophic failure of tailings storage facilities, 2016.
ICOLD. Bulletin 139 Improving tailings dam safety, Paris, 2011.
ICOLD. Bulletin 154 Dam Safety Management: Operational Phase of the Dam Life Cycle, Un-
published.
ICOLD. Bulletin 167 (preprint) Regulation of Dam Safety: An overview of current
practice
worldwide, Unpublished.
ICOLD. Bulletin 40 Manual on Tailings Dams and Dumps, Paris, 1987.
ICOLD. Bulletin 59 Dam Safety Guidelines, Paris, 1987.
LI, S.; Chen, Q. & Wang, X. Superiority of filtered tailings storage facility to conventional
tailings
impoundment in southern rainy regions of China, 2016.
Mining Association in Canada (MAC) (2017). A Guide to the Management of Tailings
Facilities.
Third Edition, Ottawa, Canada.
Morgenstern N. Geotechnical Risk, Regulation, and Public Policy, Soils and Rocks, São Paulo,
41(2): 107-129 May-August, 2018.
Morgenstern N. R., Vick S. G., Viotti C. and Watts B. Report on the Immediate Causes of the
Failure of the Fundão Dam. Fundão Tailings Dam Review Panel, 2016.
Morgenstern N. R., Vick S. G. and Van Zyl, D. Report on Mount Polley Tailings Storage Facility
Breach, Report of Independent Expert Engineering Investigation and Review Panel to Prov-
ince of British Columbia, 2015.
NORWEGIAN GEOTECHNICAL INSTITUTE. Rockfill dams, design and construction, Oslo,
1992.
USACE. Engineering and Design – Slope Stability, Doc No. EM 1110-2-1902, 1970.
USACE. Engineering and Design – Slope Stability, Doc No. EM 1110-2-1902, 2003.
USBR. Design of Small Dams, 1987.
WISE. Chronology of major tailings dams failures (www.wise-uranium.org/mdaf.html), 2019.
