minerals

Article
A Failure Modes and Effects Analysis Framework for Assessing
Geotechnical Risks of Tailings Dam Closure
Haley L. Schafer 1, * , Nicholas A. Beier 1 and Renato Macciotta 1,2

1 Department of Civil and Environmental Engineering, University of Alberta, 9211 116 Street NW,
Edmonton, AB T6G1H9, Canada; [email protected] (N.A.B.); [email protected] (R.M.)
2 School of Engineering Safety and Risk Management, University of Alberta, 9211 116 Street NW,
Edmonton, AB T6G1H9, Canada
* Correspondence: [email protected]

Abstract: Tailings dams remain on site following mine closures and must be designed and reclaimed
to meet long-term goals, which may include walk-away closure or long-term care and maintenance.
The underperformance of these structures can result in significant risks to public and environmental
safety, as well as impacts on the future land use and economic activities near the structure. In Alberta,
Canada, the expectation is for a tailings dam to be reclaimed and closed so that it can undergo dereg-
istration. To aid in assessing the risks of underperformance during and after closure, a Generalized
Failure Modes and Effects Analysis (G-FMEA) framework was developed to assess the long-term
geotechnical risks for tailings dams in Alberta, with the goal of assessing the potential success of a
tailings dam closure strategy. The G-FMEA is part of an initiative to enhance closure evaluations





 in Alberta in a collaborative effort between industry, the regulator, and academia. The G-FMEA
Citation: Schafer, H.L.; Beier, N.A.;
incorporates the element of time to account for the evolution of the system, which should be applied
Macciotta, R. A Failure Modes and at the planning stage and updated continually throughout the life of the facility. This paper presents
Effects Analysis Framework for the developed G-FMEA framework for tailings dams in Alberta, including the developed risk matrix
Assessing Geotechnical Risks of framework.
Tailings Dam Closure. Minerals 2021,
11, 1234. https://doi.org/10.3390/ Keywords: external tailings facilities; tailings dams; closure; Alberta; failure modes and effects
min11111234 analysis; geotechnical risk

Academic Editors:
Masoud Zare-Naghadehi
and Javad Sattarvand 1. Introduction
Historically, tailings dams were designed with the consideration of geotechnical
Received: 1 October 2021
Accepted: 3 November 2021
failure modes that could occur during the active life of the mine. Little consideration was
Published: 6 November 2021
given to closure of the facility and the long period afterwards. This approach has resulted
in many tailings dams that are in a poor position for closure or have been orphaned
Publisher’s Note: MDPI stays neutral
all together. This is problematic as the failure of these structures can still occur after
with regard to jurisdictional claims in
the closure of a mine site, such as at the Matachewan Consolidated Mine in Ontario [1];
published maps and institutional affil- or when the mine is in a period of care and maintenance, such as at Obed Mountain
iations. Coal Mine in Alberta [2]. Regardless of whether the mine is active or not, tailings dams
can pose a significant risk to the public and the environment if not properly managed
or decommissioned. In the medium-to-long term, after decommissioning, uncertainties
associated with the performance of these structures increase, as there is limited information
Copyright: © 2021 by the authors.
regarding how a tailings dam ages in perpetuity. This knowledge gap poses a significant
Licensee MDPI, Basel, Switzerland.
environmental, public, and financial risk, especially when combined with the serious
This article is an open access article
consequences associated with the failure of tailings dams. As a result, it is vital that risk
distributed under the terms and management practices be developed to assess the risks associated with the long-term
conditions of the Creative Commons behavior. Mitigation strategies should be adopted early in the life of the structure, or as
Attribution (CC BY) license (https:// soon as they are practical for active or closed facilities. There is a transition towards
creativecommons.org/licenses/by/ integrating sustainable mining practices at the outset of mining. Of note, the Landform
4.0/). Design Institute (LDI) advocates for mining with an end in mind, noting that there will

Minerals 2021, 11, 1234. https://doi.org/10.3390/min11111234 https://www.mdpi.com/journal/minerals

Minerals 2021, 11, 1234 2 of 35

always be residual risks present following the closure of a mine [3]. Managing the long-
term residual risks requires a transparent mechanism that includes collaborative design,
construction, reclamation, and aftercare [3].
In Alberta, Canada, the mining industry consists primarily of coal and oil sands
extraction. The external tailings facilities (ETFs) at these mines are at various points in their
lifecycles. With some facilities reclaimed or in the process of being reclaimed, it is important
that systems are in place to aid operators in understanding the regulator’s requirements
for the process of deregistration. Deregistration is the process where the governmental
body in charge of regulating these structures (i.e., Alberta Energy Regulator, AER) assesses
a dam to determine if it can be removed from the dam and pond registry and no longer
be regulated as a dam; Alberta is making ongoing progress towards this goal. This is
evident in the release of the Alberta Dam and Canal Safety Directive (the Directive) and
Manual 019: Decommissioning, Closure, and Abandonment of Dams at Energy Projects
(Manual 019) [4,5]. Manual 019 indicates that “in closure, a dam operator provides active
care and may transition to passive care, depending on site-specific circumstances. As a
result, the Directive adopts a number of formal risk-management principles to ensure
the long-term care and maintenance of tailings dams after the construction and operation
phase has ended” [5].
With the impending closure of a number of external tailings facilities in Alberta,
a Generalized Failure Modes Effects and Analysis (G-FMEA) framework was developed
to assess the landform design for closure. This assessment should ideally be conducted
before mining begins (during permitting) and updated as the project progresses. Ultimately,
the goal of the G-FMEA framework is to aid in managing the residual risks associated with
ETFs in both a practical and economical way. The residual risks must be acceptable to
stakeholders.
In order to complete a comprehensive risk assessment, it may be necessary to conduct
various levels of modelling to evaluate the behaviour and/or assess consequences. For ex-
ample, in operational risk assessments, runout modelling and inundation mapping are
important elements of the risk assessment, as noted by Ghahramani et al. [6]. The utility
of such assessments for conducting a long-term closure risk assessment for the purposes
of deregistration may need to be evaluated in the context of the regulator. For example,
the AER (Manual 019) requires flowable tailings to be removed or mitigated for a tailings
facility to be deregistered, such that they do not pose an unacceptable risk to dam safety [5].
Similar criteria regarding flowable tailings is outlined by Al-Mamun and Small [7]. While
further advancements and an increased confidence in the results of the runout modelling
and inundation mapping are useful for understanding the risks associated with flowable
tailings, they may not be useful tools within the current deregistration framework. How-
ever, they remain useful tools for assessing the long-term risks of tailings facilities where it
is not possible to remove the flowable tailings or for orphaned facilities.
The G-FMEA will fit most structures and failure modes but may highlight failure
modes that require additional analyses, such as quantitative risk assessments. As such,
the G-FMEA is intended to be used as a screening tool for the closure phase of the life cycle
of an external tailings facility, where the risks assessed as acceptable require no further
analyses and higher risks (or multiplicity of relatively low risk) can trigger more detailed
and/or quantitative approaches. The G-FMEA is, therefore, meant to investigate the
potential of the facility to be deregistered as a dam by investigating the risk of geotechnical
failure. This paper presents the developed G-FMEA framework, including a developed
risk matrix framework. The G-FMEA charts developed as part of the framework present
the failure modes that are applicable to closure and should not be used to assess dam safety
during construction or operation.

2. FMEA Background
FMEA is a top-down risk assessment tool that allows for the systematic identifica-
tion and analysis of the different failure modes and their associated consequences [8,9].
Minerals 2021, 11, 1234 3 of 35

FMEA aims to identify all potential failure modes, the consequences of these failure modes,
the mechanisms of failure, and how the risks associated with the failures can be avoided
or controlled [10]. The entire process is based on determining what happens if a specific
component or element of the system fails [11]. When extended and applied to a dam,
FMEA focuses on developing a clear picture of the dam, including the various compo-
nents and how they interact in a systematic way [11]. Using this basis, FMEA can be
used to evaluate how component failures can lead to overall system failures, the conse-
quences of component and system failures, and the criticality of various components for
risk control [11].
FMEA is generally site-specific and considers the site data, construction method,
construction materials, and past behaviour of the structure [12]. As a result, it can be a time-
consuming, complex process (occasionally even requiring field or laboratory investigations).
Despite this, it is proven to be an extremely useful tool as it allows risks to be assessed
and managed. As shown by Santos et al., FMEA is effective in demonstrating the potential
failure modes of all components of a tailings dam system which then allows for the
development of a comprehensive dam monitoring and surveillance system [13]. To conduct
an FMEA, the following structure can be used [11]:
1. Define the system, including all components.
2. Based on component interaction, de-aggregate the system into functional sub-systems.
3. Break the sub-systems down into key elements and functions.
4. Analyze the failure modes of the different elements.
5. Assess the failure effects and consequences of the various elements.
6. Summarize the findings.
7. Repeat as necessary.
Steps 4 and 5 involve analyzing the failure mode, effects, and consequences of the
different elements. This is an important step as it allows the effect of component failure
modes on other components of the sub-system and the overall system to be evaluated.
Applying an FMEA to a tailings dam is not a new or revolutionary task; however,
the application of an FMEA to assess the risk of a tailings dam over post closure time
frames is challenging as a major limitation of an FMEA is the inclusion of time dependence
and the depreciation of performance of system components [13]. It is expected that the dam
will evolve significantly over time (i.e., clogging of drains, ageing of sands, etc.) [14,15].
Ultimately, the likelihood and consequences of failure are not static and may increase or
decrease over time [14–16]. Robertson and Shaw note that some risks have a different
likelihood or consequences if they occur during operations or post closure [9]. Risk matrices
aim to include time through defining the time frame in which the probability of failure
(PoF) is valid. This is typically set as an annual PoF, for the duration of the life of the
structure. This concept is complex for evaluating closure as PoF is expected to change over
long periods of time.
Incorporating time into an FMEA requires an evaluation and consideration of how
risk profiles may change over time due to system changes, as shown in Figure 1. Point A
represents the state of a system (dam structure) in terms of its PoF. Point A has an associated
probability of failure that is above the acceptable limit at time zero. To move from point
A to A’ so that the risk meets the acceptable limit would require risk control measures to
be implemented. While this satisfies the conditions at time zero, the question remains:
What will happen as time progresses and the dam undergoes evolutionary processes? As
shown in Figure 1, the risk profile may follow several different trajectories that range from
decreasing over time to increasing over time.
Minerals 2021, 11, x FOR PEER REVIEW 4 of 32
Minerals 2021, 11, 1234 4 of 35

Figure
Figure 1.
1. Changing
Changing probability
probability of
of failure
failure with
with time
time due
due to system changes.
to system changes.

The
The changing
changing risk
risk profiles
profiles over
over time
time are
are influenced
influenced by by the
the closure
closure plan
plan and
and consid-
consid-
erations of the custodial transfer scenario of the local jurisdiction. In cases where the clo-
erations of the custodial transfer scenario of the local jurisdiction. In cases where the
sure risks are intended to be managed with long-term maintenance, consideration may be
closure risks are intended to be managed with long-term maintenance, consideration may
required
be requiredfor for
scenarios where
scenarios wherethe the
maintenance
maintenance maymay
be disrupted
be disruptedfor afor
period of time
a period (for
of time
example, by war
(for example, or another
by war pandemic).
or another pandemic).The developed
The developedG-FMEAG-FMEAconsiders different
considers tem-
different
poral scales
temporal as aasmeans
scales a meansof of
accounting
accounting forforthe
thetime
timedependence,
dependence,depreciation
depreciation ofof system
system
elements, and the associated changing risk profiles over time, as shown
elements, and the associated changing risk profiles over time, as shown in Figure 1. in Figure 1.

3. Risk Matrix Background

3. Risk Matrix Background
FMEAs may be qualitative or quantitative in nature and are commonly combined with
FMEAs may be qualitative or quantitative in nature and are commonly combined with
qualitative risk matrices to estimate the likelihood and consequences of different failure
qualitative risk matrices to estimate the likelihood and consequences of different failure
modes. Risk matrices combine the estimates of the likelihood of a negative outcome with
modes. Risk matrices combine the estimates of the likelihood of a negative outcome with
estimates of the magnitude of consequences to determine a risk level [9,16,17]. The risk
estimates of the magnitude of consequences to determine a risk level [9,16,17]. The risk level
level then determines the level and timing of the required mitigative measures and critical
then determines the level and timing of the required mitigative measures and critical con-
controls to be implemented to reduce the risk level [18]. Risk matrices are often perceived
trols to be implemented to reduce the risk level [18]. Risk matrices are often perceived as
as being simple to interpret and use. However, according to Porter et al. [16], “guidance
being simple
notes that to interpret
explain and
the details use. the
of how However,
matricesaccording to Porter
were developed and et
how al.they
[16],were meant notes
“guidance to be
that explain the details of how the matrices were developed and how they were meant
used for risk assessment, risk communication and risk management are often brief and cryptic”.to be used for
risk assessment,
Ultimately, thisrisk communication
lack of guidanceand canrisk management
lead are often
to problems withbrief
the and cryptic”. Ultimately,
application of the risk
this lack of guidance can lead to problems with the application of
matrix, which is already commonly criticized for being subjective. Risk matricesthe risk matrix,have
which is
been
already commonly criticized for being subjective.
critiqued by a number of authors, as shown in Table 1. Risk matrices have been critiqued by a
number of authors, as shown in Table 1.
Minerals 2021, 11, 1234 5 of 35

Table 1. Risk matrix issues, references, and associated comments related to the G-FMEA.

Issue References Comments related to the G-FMEA

Quantitative data should be used as often as possible with supporting verbal descriptions
and quantitative descriptors (ranges or anchor points to the definition of categories).
Subjective nature (not as simple and transparent as it would seem). [8,19–23]
The risk assessment should be accompanied by a description of the risk assessor, including
their risk tolerances.
The risk matrix does not focus on risk prioritization, but is intended as a screening tool to
Can lead to incorrect risk prioritization. [21,22,24,25]
assess the closure design.
The risk matrix is designed to determine if a closure design is adequate and is a screening
Can shut down conversations about risk instead of opening them up due to the pre-defined
[24] tool. It involves input from stakeholders to determine the consequence ratings and is
colour coding scheme determining risk prioritization.
intended to open up dialogue.
Ranking reversal: where quantitatively smaller risks are assigned qualitatively higher The risk matrix does not focus on risk prioritization, but is intended as a screening tool to
[19,20,26,27]
rating levels than some quantitatively larger risks due to incorrect risk prioritization. assess the closure design.
Does not account for different risk tolerances of the individual conducting the risk
[22,28] The risk assessor‘s background should be listed with the completed risk assessment.
assessment.
Range compression where risks with very different likelihoods and consequences are The number of risk categories should be developed with consideration of range
[19,26,28,29]
grouped together. Risk matrices with too many categories may give false resolutions. compression.
Centering bias can be an issue where individuals have the tendency to avoid extreme An extra category can be added to both sides of the expected range for the consequences
[26,30]
values or statements when presented with a choice. This can exacerbate range compression. and likelihoods, as suggested by Duijm [31].
Category definition bias where different definitions exist for a given likelihood or
[26,31] Clear definitions must be provided for the consequences and likelihoods categories.
consequence descriptor.
Risk matrix can be misleading as it implies that risk is categorical as opposed to a position The risk matrix should be accompanied by a clear definition of risk and associated
[20]
on a risk continuum. discussion on the risk continuum.
Ambiguity of the consequence definition. There are different definitions used in practice for
the consequence which can lead to issues if not clearly defined (worst case, most likely, [31] The consequence category must be clearly defined.
a number of alternate discrete outcomes).
The goal of this risk matrix is to serve as a screening tool and is not intended to provide an
Risk matrices cannot provide aggregate measures of risk (i.e., total risk). [23,27,32]
indication of total risk.
Risk matrices are unable to aggregate risk from multiple consequence dimensions.
This means that different types of consequences should not be directly compared (i.e.,
[23,31,33] A risk rating should be assigned for each consequence category for a hazard.
impact on the environment, human life). In practice, a hazard is often assigned a risk level
based on the most severe consequence. This is misleading.
Corporate-wide risk matrices are intended to be used as a way of standardizing risk
assessment and risk acceptance criteria across a company. This is problematic as risk [31] Corporate-wide risk matrices should not be used with the Generalized FMEA.
tolerance may vary throughout a company.
Minerals 2021, 11, 1234 6 of 35

To aid in mitigating the known issues with risk matrices, a number of authors
have suggested various improvements, including logarithmic scales [19,27]; probability-
consequence diagrams with continuous scales [31]; and following the Cox axioms of weak
consistency, betweenness, and consistent colouring [19]. Despite the issues associated with
risk matrices, some authors suggest that an important element of using risk matrices is
applying them in the appropriate scenarios. Bao et al. suggests that subjectivity is a vital
characteristic of risk matrices as they are effective for assessments where data are insuffi-
cient and quantitative tools cannot be applied [32]. Baybutt notes that risk matrices should
be used to provide “ . . . initial decision guidance, which should be used with caution and the
application of common sense. Risk matrices should not be used in isolation to make decisions” [27].
Considering all of this, the risk matrix developed as part of this research is intended to be
used as a screening tool, where the risks assessed as acceptable require no further analyses
and higher risks (or multiplicity of relatively low risk) could inform where/if quantitative
risk assessment techniques need to be used in the further decision-making stages for the
closure of a facility.

4. Generalized FMEA Framework

A Generalized FMEA framework was developed, based on the current practice with
FMEA, for use by regulators and the industry to assess the long-term risk of the failure of a
tailings dam following closure. In order to conduct an FMEA, the definition of tailings dam
failure must be clearly laid out. For the purposes of this work, a dual definition will be
used for failure, similar to the Ultimate Limit States and Serviceability Limit States used in
foundation engineering, as defined in the Canadian Foundation Engineering Manual [34].
Here, failure is defined as:
• Ultimate failure: the collapse of a tailings dam leading to catastrophic failure as
defined by the Global Tailings Standard [35]:
◦ “A tailings facility failure that results in material disruption to social, environmental
and local economic systems. Such failures are a function of the interaction between
hazard exposure, vulnerability, and the capacity of people and systems to respond.
Catastrophic events typically involve numerous adverse impacts, at different scales
and over different timeframes, including loss of life, damage to physical infrastructure
or natural assets, and disruption to lives, livelihoods, and social order. Operators may
be affected by damage to assets, disruption to operations, financial loss, or negative
impact to reputation. Catastrophic failures exceed the capacity of affected people to
cope using their own resources, triggering the need for outside assistance in emergency
response, restoration and recovery efforts.”
• Serviceability failure: failure to perform as intended.
The goal of performing the G-FMEA is to guide practitioners and regulators through
the process of continuing to ask questions in a disciplined manner to prevent ultimate
failure. The risk management tool is not intended to be a static tool and should be revisited
at specific intervals and triggered by changes in the facility or events of underperformance.
A clear trigger protocol needs to be developed for revisiting the G-FMEA. The completed
G-FMEA should undergo a robust scrutiny, just like the designs, be accompanied by
peer reviews, and assessed by external reviewers or review boards, where applicable.
The development of the G-FMEA is assisted by the following:
1. The interviews presented in Schafer et al. [14,15];
2. Literature reviews related to:
a. Tailings dam characteristics in the oil sands and coal industries;
b. Tailings dam failure modes;
c. Risk assessment tools.
3. The input from the regulator and industry stakeholders.
Schafer et al. [14,15] and a detailed literature review were used to develop the overall
system definition, which included dam characteristics that impact dam behaviour. The sys-
Minerals 2021, 11, x FOR PEER REVIEW 7 of 32

Minerals 2021, 11, 1234 7 of 35

Schafer et al. [14,15] and a detailed literature review were used to develop the overall
system definition, which included dam characteristics that impact dam behaviour. The
system definition
tem definition servesasasthe
serves thefoundation
foundationofofthe thedevelopment
development of of the
the G-FMEA
G-FMEA and and Ayyub
Ayyub
notes
notes that the first step to conducting a system definition is to establish the objectives[8].
that the first step to conducting a system definition is to establish the objectives [8].
This
Thisstep
stepisisessential
essentialto toproviding
providingcontext
contextandandensuring
ensuring that
thatthetheusers
usersofofthe
therisk
riskframework
framework
are
are asking
asking thethe same
same questions
questions whenwhen applying
applying thethe G-FMEA
G-FMEA to to their
their sites.
sites. The
The objective
objective
question
question of the G-FMEA is: How should geotechnical risks associated with an
of the G-FMEA is: How should geotechnical risks associated with an external
external
tailings
tailingsfacility
facilityinin Alberta
Alberta be be managed
managed in in the
the long-term
long-term to to achieve
achieve anan acceptable
acceptable closure
closure
plan,
plan, such
such that
that the
the facility
facility isis able
able to
to be
be deregistered
deregistered as as aa dam?
dam? All All users
users ofof the
the G-FMEA
G-FMEA
must
musthave
havean anunderstanding
understanding of of the
the objective
objective and
and intended
intended use use of
of the
the FMEA.
FMEA.
Following this, the system boundaries can be defined.
Following this, the system boundaries can be defined. The system The system definition is sum-
definition is
marized in Figure 2 and shows the characteristics of the dam that should
summarized in Figure 2 and shows the characteristics of the dam that should be carefully be carefully con-
sidered
consideredand and
defined prior
defined to starting
prior the the
to starting FMEA.
FMEA.TheThe
general
generalcategories
categoriesare are
summarized
summarized in
the flow
in the chart
flow on on
chart thetheleftleft
with details
with on on
details thethe
general categories
general categories in the table
in the on on
table thethe
right.
right.

Figure
Figure2.
2.System
Systemdefinition
definitionof
ofoil
oilsand
sandand
andcoal
coalmines
mines in
in Alberta,
Alberta, Canada.
Canada.
 Minerals 2021, 11, x FOR PEER REVIEW 8 of 32
Minerals 2021, 11, 1234 8 of 35

The system
The systemdefinition
definitionserves
servesas as
thethe
minimum
minimum of what should
of what be defined
should for a tailings
be defined for a tail-
ings dam prior to conducting the FMEA. However, these parameters should be com-
dam prior to conducting the FMEA. However, these parameters should be defined defined
prehensively and may
comprehensively and take
may thetakeform
the of a fact
form of sheet. This step
a fact sheet. Thisaidsstepin aids
evaluating and con-and
in evaluating
ducting the FMEA.
conducting the FMEA.
Following the
Following thedevelopment
developmentofofthe thesystem
systemdefinition,
definition, a hazard
a hazard mapping
mapping exercise
exercisewaswas
conducted, which focused on identifying potential failure modes,
conducted, which focused on identifying potential failure modes, mechanisms, triggers, mechanisms, triggers,
and hazards.
and hazards. A A failure
failuremode
modeisisdefined
definedasasthe theoverarching
overarching cause
cause of of
failure. TheThe
failure. failure
failure
mechanism isisthe
mechanism themechanism
mechanism by which
by which a hazard leads leads
a hazard to a failure
to a mode.
failureThis may consist
mode. This may
of a series
consist of aofseries
steps.of Asteps.
triggerAistrigger
definedisas that which
defined initially
as that whichleads to the
initially failure
leads mecha-
to the failure
nism. A hazard is defined as a condition with the potential to cause
mechanism. A hazard is defined as a condition with the potential to cause an undesirable an undesirable conse-
quence. Combined,
consequence. Combined,the system definition
the system and hazard
definition mapping
and hazard contributes
mapping greatlygreatly
contributes to the to
development of the G-FMEA. The steps required to use the
the development of the G-FMEA. The steps required to use the G-FMEA in practiceG-FMEA in practice are out-are
lined in Figure 3. If risk mitigation is required, or the closure plan is altered,
outlined in Figure 3. If risk mitigation is required, or the closure plan is altered, the process the process
may need
may need toto be
be repeated.
repeated.ThisThismaymayrequire
requirethe therisk assessor
risk assessor to to
return
return to to
Step 1 to1 update
Step to update
the system definition if substantial changes are made.
the system definition if substantial changes are made.

Figure 3.
Figure 3. G-FMEA
G-FMEAsteps.
steps.

The G-FMEA
The G-FMEA includes
includesfourfourindividual
individualcharts
chartsforfor
the
thedrainage
drainage system,
system, foundation,
foundation,
dam body, and landform, which are provided in Appendix A. An excerpt of the drainage
dam body, and landform, which are provided in Appendix A. An excerpt of the drainage
system chart
system chart is
is provided
providedin inTable
Table2.2.The
Thedrainage
drainagesystem
system and
andlandform
landform chart
chartconsider
consider
failure modes
failure modesfor fora avariety
varietyof of
different elements
different that that
elements couldcould
be present within within
be present a dam. a The
dam.
foundation
The chart and
foundation chart damandbody
damchart
bodydochart
not provide
do notfailure
providemodes for individual
failure modes forelements,
individual
since mostsince
elements, failure modes
most are applicable
failure modes aretoapplicable
multiple elements.
to multiple Each identifiedEach
elements. foundation
identified
element should
foundation be evaluated
element should forbe every failure
evaluated mode
for everyonfailure
the foundation
mode on list.
theSimilarly,
foundationeachlist.
dam body each
Similarly, element
dam mustbodybe element
evaluatedmust
for every failure mode
be evaluated on the failure
for every dam body mode list.on the dam
bodyFailure
list. modes that are not applicable should be screened out prior to completing the
risk assessment,
Failure modes as discussed
that are notin Step 5. A justification
applicable should bemust be recorded
screened out prior fortoscreening
completing outthe
risk assessment, as discussed in Step 5. A justification must be recorded for screening
out a failure mode and re-evaluated each time the FMEA is revisited. This may include a
Minerals 2021, 11, 1234 9 of 35

justification of whether a failure mode is credible or non-credible as defined by the Global

Industry Standard on Tailings Management [35]. It is important to note that failure modes
that may be credible in the short-term may not be credible in the long-term and vice versa.
While the list of failure modes in Appendix A for the various elements is extensive
and reasonably comprehensive, it is recommended as a pre-list of failure modes as no
list can be completely comprehensive as there are always additional site-specific hazards.
The G-FMEA charts should be applied to a facility using the complete worksheet provided
in Appendix B (Table A5).
Table 2. Excerpt from drainage element G-FMEA chart.

Failure Mode Potential

Element Screening Assessment of Failure Mode Failure Effects
Description Trigger/Cause
Is there erosion protection in place? What
Rise in phreatic surface, increase
Sedimentation, is the slope of the ditch? Is it sufficient to
in seepage, pond on reclamation
sloughing/slope keep particles suspended? What is the
surface, internal erosion, global
Blockage failure of walls, slope of the side slopes? What is the
instability, flooding, blockage of
(partial or full) beaver dam, strength of material? Have there been
drain outlets, toe erosion,
continuous build up failures in this material before? Are there
discharge of process affected
of ice (icing) beavers in the area? Is the mine located in
water to the environment
an area that could experience icing?
Rise in phreatic surface, increase
What is the slope of the side slopes? What in seepage, pond on reclamation
Perimeter Ditch is the strength of material? Have there been surface, internal erosion, global
Sloughing/slope failures in this material before? Will the instability, reduced capacity,
Reduction in
failure of walls, ditch regularly have water running erosion, potential flooding,
cross-sectional area
excessive vegetation through it or will it stay dry for a portion of blockage of drain outlets,
the year? Are there deterrents in place to toe erosion, discharge of process
prevent the growth of vegetation? affected water to the
environment
Is there erosion protection in place? Does
Change to water discharge
the material have the potential to
Erosion, differential velocity, creation of secondary
Change in slope consolidate or settle over time? Is it a cut
settlement channels, localized areas of
into natural ground or is
erosion, instability of dam
the material placed?

4.1. Temporal Scales

It is essential that the FMEA for a closed tailings dam considers the element of time
(i.e., how the dam evolves). Representing this is not straightforward and can be approached
using different methods. The first method would be to complete a series of FMEAs for
different points in time considering the different evolutionary processes that may occur.
Ultimately, this adds a lot of complexity and may be prohibitively time-intensive. Another
route, which is adopted here, is to complete immediate-term, short-term, medium-term,
and long-term risk assessments using a risk matrix for each. The determination of the time
frames requires consideration of what the overall time frame is for closure. An evaluation of
this was carried using the following points: geological context, nuclear repository industry,
Uranium Mill Tailings Remedial Action (UMTRA) surface project, the current databases
that are relied on in the industry (often for less than 100 years), our ability to forecast the
future, and transitions in periods of management.
For the long-term time period, time frames in excess of 10,000 years were deemed
to be inappropriate in the mining industry as it is expected that glacial periods in time
frames in excess of 10,000 years would change the geology and effectively bury the tailings
facilities. The use of these time frames is perhaps better suited for systems such as nuclear
waste repositories in tectonically inactive areas. With the consideration of this and the
expectation for a tailings dam closure lasting 1000 years [5,36], the maximum time frame
selected for the G-FMEA is 1000 years. This is also reflected by the UMTRA project where
the facilities are designed to be effective for 1000 years where reasonably achievable with
a minimum required performance of 200 years [37–40]. All risk assessments conducted
for the 1000-year time frame should include a careful consideration of the uncertainties
Minerals 2021, 11, 1234 10 of 35

associated with the current databases, climate change predictions, and our ability to forecast
the future.
The selection of appropriate time frames for the immediate-term, short-term, and
medium-term risk assessment was completed with consideration of the phases of the
tailings facility following closure. Initially, consideration was given to aligning the different
risk assessments with the ‘active care’ and ‘passive care’ phases commonly used in industry.
This was ultimately discarded as different organizations describe these phases in different
ways, which may ultimately lead to confusion [41–43]. To mitigate this, the performance
monitoring assessment periods recommended by the International Network for Acid Pre-
vention (INAP) were selected. INAP outlines three specific assessment periods, including
the adaptive management period, proactive management period, and reactive management
period, which are outlined in Table 3 [44]. The primary difference between the assessment
periods is the degree of direct management employed [44]. The focus on management,
as opposed to the actions taken during the assessment period, is attractive from a safety
management system perspective.

Table 3. Assessment periods [44].

Assessment Period Definition

Occurs following closure activities when the closed facility may be at its greatest risk of failure prior to reaching
Adaptive Management equilibrium. The operator has the greatest capacity to respond. This period may be defined explicitly by the
regulator, using accumulated knowledge, or using a site-specific scientific basis.
Occurs as personnel and equipment are reduced. Involves a regular fixed frequency monitoring and maintenance
Proactive Management schedule to confirm that the landform is trending along the designed trajectory. It is expected that the frequency will
be less than during the adaptive management period.
Issues are rectified strictly on a reaction basis, once a trigger event occurs. There should be a clear plan in place that
Reactive Management outlines what the trigger events are and how they will be managed. Monitoring may occur in response to events
such as fires, floods, earthquakes, and other extreme events.

Based on the characteristics associated with the adaptive management period, the
immediate-term risk assessment should occur during this phase. The short-term risk as-
sessment may occur during the adaptive management period or the proactive management
period, depending on site-specific characteristics.
The selection of a time frame for the medium-term risk assessment involved the con-
sideration of the reactive management period and the time frames outlined by the UMTRA
project. The medium-term risk assessment should fall within the reactive management
period where direct management efforts (and monitoring) decrease substantially and the
facility largely shows that it meets closure objectives. The recommendations for the time
frames for the various levels of risk assessment are provided in Table 4. The approximate
assessment periods provided are intended as a guideline only and should be based on
site-specific considerations. It is expected that the immediate-term and short-term risk as-
sessments would have the most active forms of risk mitigation employed with risk measures,
slowly transitioning to more passive methods as the risk assessment moves to medium- and
long-term temporal scales. It is also expected that the level of confidence will decrease as the
risk assessment moves from immediate-term to long-term scales for some failure modes.

Table 4. Time frames for various levels of risk assessment.

Risk Assessment Conditions Approximate Assessment Period 1

Should fall within the projected adaptive
Immediate-term 0–10 years
management period
Should fall within the projected adaptive
Short-term 10–50 years
or proactive management period
Should fall within the projected reactive
Medium-term 50–200 years
management period
Long-term - 1000 years
1 Assessment period time frames depend on site-specific considerations. Intended as a guideline.
Minerals 2021, 11, 1234 11 of 35

4.2. Risk Matrix Framework

The development of the risk matrix for the G-FMEA involves a consideration of the
known pitfalls with risk matrices, as outlined in Table 1; recommendations for improvement
from Baybutt and Duijm [25,31]; and the evaluation and consideration of other published
risk matrices, including those provided by Brown, Hadjigeorgiou, the Guidelines for
Mine Waste Dump and Stockpile Design, Mine Environment Neutral Drainage (MEND),
and the Oil Sands Tailings Dam Committee (OSTDC) [17,18,43,45,46]. A number of key
observations can be taken from the reviewed published risk matrices, including:
• Clear descriptions of likelihood and consequence ratings are essential. as provided
by Griffin and MEND [17,46]. It is common for different risk matrices to use the
terms ‘almost certain’, ‘likely’, and ‘possible’ but it is equally common for these to not
be accompanied by clear definitions as shown in Brown and Hadjigeorgiou [18,45].
This is problematic as terms such as these are not universally understood to have the
same meaning amongst practitioners [47]. This aligns with the critiques in Table 1.
• A level-of-confidence scale (or similar) is recommended to go along with the risk
matrix and selected risk rating [17].
• MEND and OSTDC were both developed for closure scenarios and provide useful
analogues for the development of the risk matrix [17,43].
• OSTDC determines the consequence rating based on the loss of function of the struc-
ture, the degree of human intervention required on the structure after an event occurs,
population at risk, and environmental economics [43]. In contrast, a risk matrix
designed to assess the risk for an operational facility may consider elements such
as health and safety, environment, community, reputation, and legal aspects when
assessing the consequences of failure.
While the risk matrix developed by OSTDC provides the best analogue and a use-
ful basis for this work, it cannot be applied directly to the G-FMEA, for the following
reasons [43]:
• The G-FMEA presented in this research assesses the individual elements of a dam,
not the dam as a whole, which is what the OSTDC [43] risk matrix was developed for.
The developed risk matrix must be fit for purpose.
• The OSTDC risk matrix was developed to assess if a facility could be deregistered as
a dam [43]. The consequence and likelihood ratings reflect this. In contrast, the risk
matrix is used here to assess the risk of failure over time and if a closure plan is
adequate to prevent failure to support a facility being deregistered as a dam.
The risk matrix framework is intended as an example of how a qualitative risk assess-
ment method could be combined with the G-FMEA charts to assess a closure plan and
includes a likelihood rating table, consequence rating table, and a risk matrix (including
details on how to colour code the risk matrix and an example of a colour-coded matrix).
The risk matrix to be used for a site should be defined by the industry, regulators, and other
stakeholders with considerations of the technical, social, and economic aspects.

4.2.1. Likelihood Rating

The developed likelihood rating shown in Table 5 has seven different categories
for likelihood from ‘Close to non-credible’ to ‘Almost certain’, which are defined with
considerations of annual probabilities. Qualitative descriptors were developed to provide
further context on the annual probabilities. The inclusion of seven likelihood categories
is more extensive than many risk matrices, which commonly have five likelihood ratings.
A more extensive likelihood rating list was selected due to the time frames associated with
the closure and to aid in mitigating centering bias (Table 1). It also serves to cover the full
range of likelihoods that may be encountered, which is an essential step to developing
a likelihood table [25]. The risk matrix is intended to be applicable from the time that
closure works have been completed to 1000 years following the completion of closure
works. This means that the risk matrix is not intended to capture failure modes that may
Minerals 2021, 11, 1234 12 of 35

occur during the construction of the closure landscape. The risk matrix is also intended to
be applied to orphaned facilities, which are defined as facilities with “owners that cannot
be found or for which the owner is financially unable or unwilling to carry out the clean
up” [48]. These facilities should not be confused with those that undergo abandonment as
per the Alberta Dam and Canal Safety Directive [4]. It could be argued that the categories of
‘Almost certain’ and ‘Likely’ should not be applied to a closure scenario as the assumption
is that the facility is designed to a standard such that the failure of an element will not
occur within a year. While this may be true for many facilities, it is unlikely to be true
for all facilities (i.e., orphaned facilities). Furthermore, some failure modes may become
more likely when conducting a long-term risk assessment as factors such as degradation
and climate change occur. The likelihood of occurrence is often described in terms of the
probability of failure. While this is an important way of evaluating the likelihood, it is
valuable to describe the likelihood categories using qualitative descriptors to anchor the
probabilities [25]. Further, humans are notoriously poor at estimating the probability of
extreme events [49]. As such, providing qualitative guidance for assessing the likelihood
of a failure mode would be useful. Table 5 describes the likelihood categories in terms
of the annualized probability and provides a qualitative interpretation. The quantitative
interpretation guidance for the example risk matrix considers that each rating is a range of
probabilities. Anchor values can also be adopted, if preferred, as long as it is clear to the
users that the anchor values are a point approximation of this range of probabilities.

Table 5. Likelihood rating.

Quantitative Annualized Probability

Likelihood Rating Qualitative Interpretation Guidance 1
Interpretation Guidance of Occurrence
Almost certain that an incident will occur
given the circumstances. Very high Higher than 10%
Almost certain p ≥ 0.1
probability of one or more occurrences probability in a year
per year.
High likelihood. Commonly observed at Higher than 10%
Likely p ≥ 0.01
similar facilities. probability in 10 years
Has occurred a number of times within
Higher than 1%
Possible the industry and at least once at the site p ≥ 0.001
probability in 10 years
(or similar facilities in the region).
Has occurred before within the industry, Less than a 1% probability
Unlikely p < 0.001
but not at the site. in 10 years
Low likelihood of occurrence, but not
Less than a 1% probability
Rare impossible. Has not occurred at the site p < 0.0001
in 100 years
but has occurred in industry.
Very low likelihood of occurrence,
Less than a 1% probability
Very rare but not impossible. Occurrence cannot be p < 0.00001
in 1000 years
deemed non-credible
Extremely remote likelihood of
occurrence. Although the mechanisms Less than a 1% probability
Close to non-credible p < 0.000001
are technically plausible for the in 10,000 years
occurrence, it is seen as near non-credible.
1 Industry encompasses the mining industry as a whole.

4.2.2. Consequence Rating

The consequence rating shown in Table 6 has five categories that range from ‘Slight’
to ‘Severe’. The consequence rating is selected based on:
• The degree of the consequence of failure of an element on the rest of the system.
◦ This involves an assessment regarding if the failure will result in cascading
consequences to other elements.
Minerals 2021, 11, 1234 13 of 35

• The degree of human intervention (post failure) required.

◦ It is important to note the degree of human intervention required in response
to a failure as the ultimate goal is to reach a state of minimal or no human
intervention. In cases where this is not possible, this may suggest a careful
evaluation of the custodial transfer scenario for that facility.
• Community.
◦ The impact on the community should be assessed using input from the relevant
stakeholders. This is difficult to evaluate in light of the time frames associated
with closure. Our ability to predict downstream and upstream populations in
the future is limited by our short-term knowledge; however, a consideration
can be made of the likelihood that the affected area could be inhabited in the
future (i.e., is the mine located in inhospitable terrain or in an area that is
desirable for development?). This would allow baseline assumptions to be
made about the future community impact. A key step to defining the impact
on a community is identifying the community. As defined by the Impact and
Benefit Agreement (IBA) Community Toolkit [50], this involves answering the
questions: “Who is the community? How is the geographic, ethnic, or scope of
community defined? Who legitimately represents the community? Is it simply
representatives from local community organizations, or is it necessary to reach
out to more diverse groups to ensure all elements are consulted? The definition
of “community” should be inclusive enough to promote equity and avoid
future conflict resulting from lack of inclusion” [50]. The community conse-
quence rating provided in Table 6 is provided as a guideline only and requires
input from the relevant stakeholders to explicitly define the community impact.
This should be accomplished using a meaningful engagement as defined by
the Global Tailings Standard [35]. Defining the consequence rating for the
community may involve a consideration of health (including fatalities), loss of
access/destruction of traditional lands, housing, destruction/damage of farm-
land, harm to livestock, damage to water or soil resources, impacts to trapping
and fishing, the loss of animals, overall cultural impact, and employment. It is
critical that the impact on all valued components (and their condition following
a failure) to the relevant stakeholders are considered. A valued component is
that which is considered important by the community [50]. It should be noted
that Impact and Benefit Agreements developed between the mining companies
and Aboriginal communities may include agreement provisions to account for
catastrophic failures and losses and should be consulted when assessing the
consequences of a dam failure to the community [50].
• Environment.
◦ The environmental impact must be assessed with a consideration of the im-
pacted land (both surrounding the facility and in the facility itself, with respect
to its post mining land use) and waterbodies and the toxicity of the tailings.
The Canadian Dam Association (CDA) developed a Working Group to revamp
their environmental classification system, noting that the existing system lacked
a clear scientific basis and used vague criteria that was open to interpretation
based on personal beliefs and principles [51]. Nikl et al. provides a summary of
the draft environmental consequence classification system [51]. The framework
considers three variables: ecological impact, the intrinsic hazards of contents,
and the duration of the impact, to determine the consequence category (low,
significant, high, and very high) using a matrix and dial combination method
approach [51]. The consequence classification is intended to assess the environ-
mental consequences from a global tailings dam failure (i.e., ultimate failure in
this case). In contrast, the risk matrix and G-FMEA is intended to assess the
failure modes of individual elements; however, the principles and concepts
Minerals 2021, 11, 1234 14 of 35

from Nikl et al. can still be used to support the environmental consequence
category [51]. As such, the methodology used by Nikl et al. is adapted to fit
within the consequence rating framework [51]. The goal of a closure plan is
for the facility to remain safe and sustain a particular land use. Consequently,
the environment category should be assessed with a consideration of the im-
pact on the post mining land use, in addition to the surrounding environment.
It is recognized that post mining land uses may change over time as the closure
plan develops [52]. As the agreed post-mining land use changes, hazards
should be re-evaluated. The most serious consequence associated with the
post mining land use is when the promised land use is destroyed leaving the
land sterilized. In such a situation, the facility may be fenced off to prevent all
access to the site. This situation may also lead to downstream effects where
additional land is sterilized. Ultimately, this would yield a severe consequence
rating.
Fatalities were included in the consequence rating table in the ‘Severe’ consequence
rating in the community column. The inclusion of fatalities in risk decisions can be a
contentious issue as various stakeholders may have different risk tolerances for fatalities,
especially with regard to voluntary versus involuntary risks. It could be argued that fatali-
ties should be included throughout the risk matrix. For example, one fatality could classify
a moderate risk rating, 10–100 fatalities could classify a major risk rating, and greater than
100 fatalities could classify a severe risk rating. Including fatalities only in the ‘Severe’ con-
sequence rating is out of step with the risk tolerances for other industries and fails to reflect
the reality that multiple-fatality events are more consequential than single-fatality events
(e.g., road vs. air accidents). However, this goes against the Global Industry Standard
on Tailings Management, which has a goal of zero harm to people and the environment
with zero tolerance for human fatalities [35]. Considering this and the overall goals of
the closure design, the current consequence rating table classifies one fatality as ‘Severe’.
This may be adapted through a meaningful engagement with stakeholders, if appropriate.
The approximate time frames for the environment and community categories are
provided: short-term (<5 years), medium-term (5–25 years), and long-term (>25 years).
These time frames are intended as guidance and may be amended with the input of
stakeholders. Depending on the risk tolerance of the stakeholders, it may be necessary to
reduce these time frames.
Minerals 2021, 11, 1234 15 of 35

Table 6. Consequence rating.

Consequence of Failure of Degree of Human Intervention

Consequence Rating 1 Environment Community
Element on the Rest of the System Required
Structural integrity maintained.
Failure of element does not have No movement of tailings beyond the structure
Slight No intervention or maintenance No impact on local community.
cascading consequences. footprint.
required.
Released tailings are not toxic 2 , and/or minimal
Failure of element may have Structural integrity maintained. loss of habitat (<5%) of species of special interest
3 , and/or acceptable restoration of water bodies Impact 4 on local community for
Minor cascading consequences that do not Minor or localized intervention or
less than 1 year.
result in global failure. maintenance required. and environment feasible in a short time frame
(<5 years).
Released tailings are not toxic 2 , and/or
Failure of element has cascading Intervention or maintenance moderate loss of habitat (5–20%) of species of
Short-term (<5 years) impact 4 on
Moderate consequences that do not result in required to limit impact of special interest 3 , and/or acceptable restoration
local community.
global failure. cascading consequences. of water bodies and environment feasible in a
short time frame (<5 years).
Released tailings are toxic 2 , and/or significant
Intervention or maintenance loss of habitat (20–50%) of species of special
Global failure of tailings dam with Medium-term (5–25 years) impact 4
Major required to maintain function of interest 3 , and/or acceptable restoration of water
minor release of tailings. on local community.
structure as a whole. bodies and environment feasible in a moderate
time frame (5–25 years).
Released tailings are toxic 2 , and/or very
significant loss of habitat (>50%) of species of
Global failure of tailings dam with Long-term (>25 years) impact 4 on
Severe Structural repair not possible. special interest 3 , and/or acceptable restoration
catastrophic release of tailings. local community. Fatalities.
of water bodies and environment unlikely within
an extended time frame (>25 years).
Notes: 1 Assigned consequence should reflect the most likely outcome. If assigning consequence with consideration of the worst case or a combination of discrete outcomes, this must be declared. 2 Toxicity
assessment of tailings should consider an assessment of the fluids and solids (leaching potential, acidity, radioactivity). 3 Species of special interest is defined as a species that lives in the inundation area that
would be greatly impacted by habitat loss (preferable to select a species that is provincially or federally listed). 4 Community impacts must be determined through meaningful engagement with stakeholders and
may include a consideration of health, loss of access/destruction of traditional lands, housing, destruction/damage of farmland, harm to livestock, damage to water or soil resources, impacts to trapping and
fishing, loss of animals, overall cultural impact, and employment. Reputation, legal aspects, and economics are not considered in this consequence table as they are considered site- and corporation-specific.
It may be necessary to assess these aspects on a site-specific basis.
 Minerals2021,
Minerals 11,1234
2021,11, x FOR PEER REVIEW 17 of 32
16 of 35

4.2.3. Risk Matrix and Rating

4.2.3. Risk Matrix and Rating
The seven likelihood ratings and five consequence ratings can be combined to form
The seven likelihood ratings and five consequence ratings can be combined to form
a risk matrix,asasshown
a risk matrix, shownininFigure
Figure 4. 4.
TheThe challenging
challenging part part of finalizing
of finalizing the therisk risk matrix
matrix lies lies
in determining
in determiningthe theassigned
assigned risk
risk category
category forfor a given
a given likelihood
likelihood andand consequence
consequence rating.
rating.
The risk categories for the G-FMEA are presented in Table 7. The risk categories werewere
The risk categories for the G-FMEA are presented in Table 7. The risk categories
developedwith
developed withthetheconsideration
consideration ofofthethe suitability
suitability of the
of the closure
closure design
design to prevent
to prevent failure
failure
such that a facility could be deregistered as a dam. This resulted in four risk categoriesfrom
such that a facility could be deregistered as a dam. This resulted in four risk categories
from ‘Low’
‘Low’ to ‘Extreme’.
to ‘Extreme’. As theAsrisk
the risk category
category increases,
increases, thethelevellevel of the
of the required
required risk
risk mit-
mitigation
igation increases as the closure plan is assessed as being inappropriate in preventing ser-
increases as the closure plan is assessed as being inappropriate in preventing serviceability
viceability
failure of afailure of a particular
particular element. element.
ItIt isis common
commonpractice
practiceforforrisks
risks totobebemanaged
managed usingusingthethe
ALARPALARP principle: As Low
principle: As Low
As
As Reasonably
ReasonablyPracticable
Practicable(ALARP).
(ALARP). In InALARP,
ALARP, all all
riskriskreduction
reduction measures
measures should be be
should
employed
employedasaslong long as as
thethe
costcost
of implementing
of implementing them is
themreasonably practicable
is reasonably with a con-
practicable with a
sideration
consideration of cost
of effectiveness [53]. In[53].
cost effectiveness TableIn7,Table
the high-risk categorycategory
7, the high-risk is defined iswith con-with
defined
sideration
consideration of the
of principle of ALARP.
the principle of ALARP. In the
In high-risk category,
the high-risk the risks
category, are undesirable
the risks are undesirable
and
andmust
mustbe bereduced
reducedusingusingALARP.
ALARP. If the risk
If the category
risk category cannot be reduced
cannot be reduced usingusing
ALARP,ALARP,
the
the closure
closureplan planshould
shouldbebealtered
altered totoaccommodate
accommodate riskrisk
mitigation.
mitigation.

Figure
Figure4.4.Risk
Riskmatrix.
matrix.

Table
Table 7.
7. Risk
Riskcategory.
category.
Risk Category
Risk Category Description of Risk
Description Category
of Risk Category
Low
Low Risk minimal. Monitor
Risk minimal. risks.risks.
Monitor Acceptable closure
Acceptable plan.
closure plan.

Moderate RiskRisk tolerable

tolerable withwith controls.
controls. Assess
Assess risk mitigation
risk mitigation optionsoptions and these risks.
and monitor
Moderate monitor these risks.
Minor re-design Minor plan
of closure re-design
may be ofrequired
closure plan may be required
to accommodate to
risk mitigation.
accommodate risk mitigation.
Risk undesirable. Risk mitigation should be employed to ALARP to reduce risk
High
Risk undesirable. Risk mitigation
category. Closure shouldalteration
plan may require be employed to ALARP to
to accommodate reduce
risk risk
mitigation.
High
category.
RiskClosure plan Risk
intolerable. maymitigation
require alteration to accommodate
required immediately risk mitigation.
to reduce risk category.
Extreme
Risk intolerable. Risk
Requires mitigation
more detailed required immediately
risk analysis. to reduce
Closure plan requiresrisk category.
alteration.
Extreme
Requires more detailed risk analysis. Closure plan requires alteration.
Once the risk categories are developed, they can be applied to the risk matrix to
Onceathe
develop risk categories
colour-coded are map’
‘heat developed,
basedthey can combination
on the be applied to the risklikelihood
of the matrix to develop
rating and
aconsequence
colour-codedrating.
‘heat map’ based
The risk on the
matrix is combination of theorlikelihood
not colour-coded populated rating
withandriskconse-
ratings as
quence rating.
this defines Thetolerance.
risk risk matrix is not
This colour-coded
step or populated
should be completed with
with risk
the ratings
input as relevant
of all this
defines risk tolerance.
stakeholders This step
(i.e., industry, should bethe
regulator, completed
public).with the input
However, anofexample
all relevant stake-
of the colour-
holders (i.e.,matrix
coded risk industry, regulator,inthe
is provided public).
Figure However,
5. The an example
colour-coded matrix of presented
the colour-coded
in Figure 5
risk matrix is provided
demonstrates in Figure
major hazard 5. The colour-coded
aversion and presentsmatrix presented
a threshold line in
thatFigure 5 demon-to be
is suggested
strates
used tomajor hazard
trigger aversion
a more and presents
detailed a threshold
quantitative lineThe
analysis. that framework
is suggested for to be used to the
attaining
trigger a morematrix
colour-coded detailed quantitative
is provided analysis.
in this The framework for attaining the colour-
section.
codedMinimal
matrix isguidance
provided isinprovided
this section.in the literature on how to colour code a risk matrix.
Ayyub indicates that colour coding a risk matrix involves shading each box depending
on a “subjectively assessed risk level” [8]. IEC provides more guidance on assigning risk
Minerals 2021, 11, 1234 17 of 35

categories to the risk matrix, indicating that assigning risk categories should depend on the
definitions of the likelihood and consequence tables [23]. Using these definitions, the risk
matrix can have risk categories assigned to provide extra weight to the consequences or
likelihoods, or it may be symmetrical [23]. Oboni suggests that a risk matrix cannot be
symmetrical as this typically indicates an incorrect prioritization of risks [54]. Risk matrices
should be coloured with consideration of what the cells tell the user.
The developed risk matrix for this research helps to assess if the risks associated with
decommissioning an ETF are acceptable or where/if a more detailed analysis is required,
and could be applied throughout a tailings facility’s life. In light of this, an alternative way
to use the developed risk matrix is to work with a “threshold”. Any hazard categorized
with a risk higher than the defined threshold (orange in Figure 5) requires a more detailed
quantitative risk assessment method, which may require a more in-depth investigation,
design, or modelling to be completed. This is an important step to aid in mitigating issues
associated with upper severity limits [25].
The framework for developing the example risk matrix was developed with a consid-
eration of the tool being used as a screening method; guidance provided by Baybutt [25],
Duijm [31], and Levine [29]; and a consideration of the known pitfalls that can be associated
with risk matrices, as discussed in Table 1. The development of the example risk matrix
in Figure 5 was accomplished using the following steps, and may be used as guidance
for colour coding the risk matrix for a specific site (see Figure 4). Appendix C shows how
Figure 5 was developed using the following steps.
1. Evaluate the likelihood ratings (Table 5), consequence ratings (Table 6), and risk
ratings (Table 7).
2. Develop iso-contours of equal risk based on the estimated quantitative consequence
measure and the provided likelihoods (Likelihood = Risk/Consequence). Assuming
that the consequence quantitative measure is an order of magnitude between cate-
gories, the consequence and likelihood iso-contours should be plotted in log-log space.
Plot the iso-contours on the risk matrix. Assign the risk categories to the iso-contours
of equal risk and use these to develop an initial colour-coded risk matrix. This step
serves as a starting point for colour coding the risk matrix. Any available known
acceptance criteria can be used as an initial starting point for this exercise. Decisions
will need to be made about the cells that have an iso-contour cutting through them
(i.e., do you assign the higher or lower risk category?). This exercise demonstrates a
known issue with risk matrices where risk is represented as categorical, rather than
on a continuum, and hazards are binned into these categories.
a. A note from Baybutt [25]: When consequences and likelihood categories differ by
orders of magnitude, they are represented in log-log space. Practically, this means
that high consequence-low likelihood events (negatively correlated) have the most
uncertain risks in these type of rating schemes. This is especially problematic
as these events already have a lot of inherent uncertainty. Ultimately, it is not
possible to increase the precision of these events on a risk matrix. This concept of
uncertainty related to these types of events supports the idea of using a threshold
value on the risk matrix to trigger more extensive risk analysis.
3. Assess the initial colour-coded risk matrix against the Cox axioms for a well-defined
risk matrix [19]. This step is simply a logical ‘check’. Duijm recommends that a
key to using the risk matrix is to recognize that the colouring of the risk matrix is
a risk definition in its own right (and cannot be separately and stringently defined
as Risk = probability × consequence) as it expresses a subjective risk perception
(i.e., major hazard aversion), which is an important element of risk decisions [31].
Duijm notes that, when no reference is made to an external risk definition, then “the
colouring of the matrix itself is the only relevant risk definition, then the axioms of
weak consistency and consistent colouring are trivial” [31]. If a quantitative definition
of risk is desired by stakeholders, then the Cox axioms should be satisfied [19].
Cox axioms can be summarized as [19]:
Minerals 2021, 11, 1234 18 of 35

a. Weak consistency, where each hazard in the red category represents a higher
risk than the hazards in the green category.
b. Betweenness, where every positively sloped line segment that starts in a green
cell and ends in a red cell must pass through an intermediate risk category.
c. Consistent colouring, where hazards with an approximately equal risk have
the same colour.
4. Assess if the Levine lettering scheme is more appropriate for your risk analysis
goals [29]. In the study by Levine, logarithmic axes are used for the consequence
and likelihood axes [29]. Straight line iso-contours of equal risk are drawn. Instead
of using colours, the different areas are labelled A, B, C, D, E, F, and G (each line
represents a new letter). This method results in a matrix that is somewhat unintuitive,
but this prevents risk matrix users from making assumptions about the risks based on
the colouring scheme. When a risk matrix is constructed in this manner, the following
conclusions can be drawn:
a. Risks in one letter category can only be distinguished from risks in another
category if they are more than one letter apart (i.e., C > A, D > B).
b. Risks in categories that are zero or one letter apart are not able to be distin-
guished from one another (i.e., it is not known if C > B or B > C).
5. As noted by Duijm, another way to develop risk scores and colouring is by using
basic arithmetic (multiplication and addition) based on ordinal numbers assigned to
each consequence and probability category [31]. When the categories are essentially
linearly spaced, then the multiplication of the ordinal numbers is an appropriate way
of defining the risk score. When the categories are essentially logarithmically spaced,
the addition of the ordinal numbers is desired. Apply the appropriate mathematical
operations and compare to the risk matrix developed in Step 3.
6. Assess if major hazard aversion is required and apply as necessary. Hazard aversion is
the concept of low-probability–high-consequence events being assigned a higher risk
than a high-probability–low-consequence event, even if the expected loss is mathemat-
ically the same [31]. This concept is used in scenarios where low-probability–high-
consequence events are of greater concern and may require different decisions [25].
7. Conduct “logic checks” by stress testing the risk matrix with different scenarios. Focus
on what the risk matrix is telling the user. Adjust the colour scheme as needed and
repeat. This is an important step as the risk matrix may be developed using one
Minerals 2021, 11, x FOR PEER REVIEW 20 of 32
quantitative measure. As such, it must be assessed to determine if it is applicable
across the different consequence categories.

Figure 5.
Figure 5. Example
Example risk
riskmatrix
matrixconsidering
consideringmajor
major hazard
hazard aversion
aversion andand threshold
threshold for quantitative
for quantitative analysis.
analysis.
4.2.4. Limitations of the Risk Matrix Framework
4.2.4 As
Limitations
noted inofTable
the Risk
1, aMatrix
commonFramework
issue related to risk matrices is their inability to
As noted in Table 1, a common issue related to risk matrices is their inability to aggre-
gate risks, both in general and from multiple consequence categories. With regard to aggre-
gating risks, as noted by IEC “one cannot define that a particular number of low risks or a
low risk identified a particular number of times is equivalent to a medium risk” [23]. This is
an important limitation of risk matrices, and one that risk users must be cognizant of. There-
fore, sound engineering judgement is required for evaluating the results of the matrix where
Minerals 2021, 11, 1234 19 of 35

aggregate risks, both in general and from multiple consequence categories. With regard
to aggregating risks, as noted by IEC “one cannot define that a particular number of low
risks or a low risk identified a particular number of times is equivalent to a medium
risk” [23]. This is an important limitation of risk matrices, and one that risk users must be
cognizant of. Therefore, sound engineering judgement is required for evaluating the results
of the matrix where multiple low-risk failure modes are present. Once the assessment is
complete, a careful evaluation of the results must be completed. All hazards that receive
a risk rating of red or orange need to undergo further analyses and may need additional
controls. If a scenario occurs where all the hazards are green (or green and yellow),
the organization should evaluate if the aggregation of them represents an intolerable
risk. In that case, the quantified descriptions in the likelihood matrix can help with the
aggregation of likelihood.
Next, it is common practice for a hazard to be assigned a risk level based on the most
severe consequence. Caution should be employed with this practice, and if practitioners
choose to use this methodology, it must be clearly declared. Instead, it is recommended
that risk ratings are assigned for each consequence category (i.e., consequence of failure of
an element on the rest of the system, degree of human intervention required, environment,
and community). This is largely due to the inability of risk matrices to aggregate the risk
from multiple consequence dimensions, meaning that the consequences should not be
directly compared (as noted in Table 1). Therefore, a careful consideration is needed when
determining a final risk rating for the hazard.
One final issue that must be addressed is related to the correlated risk. The G-FMEA
involves breaking the system down into its individual components. It is possible that one
hazard may become a trigger for another hazard. Risk matrix users must be aware of the
correlated risks and note them explicitly in the reporting of results, along with how they
were considered in the study.
Given the time frames associated with closure, it is recommended that a level of
confidence regarding the risk estimates is provided. There are uncertainties associated with
assessing the likelihood and consequence ratings, which may be associated with a lack of
data, lack of system understanding, uncertain future operating conditions or uncertain
maintenance, and regional development post closure [9]. As the time frames associated
with the G-FMEA may extend to 1000 years, it is expected that the level of confidence may
decrease as the temporal scale increases. A level of confidence framework is currently
being developed at the University of Alberta to accompany the G-FMEA.

5. Summary and Conclusions

Tailings dams can pose a significant risk to the public and the environment follow-
ing closure. To prevent the failure of these large structures after mining, their long-term
behavior must be understood and incorporated into the closure design, ideally from the
conception of the project and initial design. Challenges lie in the limited information
available regarding how these structures age over time and the forecasting, loading, and en-
vironmental scenarios over long-term periods (weather events, seismicity, human activity,
etc.). A G-FMEA framework was developed to aid in assessing the potential success of a
tailings dam closure strategy for external tailings facilities in Alberta, Canada. The goal of
the G-FMEA was to reduce the future risk of geotechnical failure on the environment and
the public to a degree that is both practical and economical. The G-FMEA can be used to
assess the closure risks at the onset of mining if adopted early in the design process, which
allows meaningful changes to be made to reduce the long-term risks from operational
design decisions and ultimately support the goal of deregistration. In many cases, this is
not possible as the tailings facility may be further into its life cycle (as is the case at many
Alberta tailings facilities).
While this geotechnical risk management tool was developed with Alberta in mind,
it does have global significance and could be easily adapted to other jurisdictions or
organizations (i.e., to account for the different types of mines present in other parts of
Minerals 2021, 11, 1234 20 of 35

Canada or the world). Furthermore, the G-FMEA may be adapted for in-pit facilities
with perimeter dykes (as desired) by updating the G-FMEA tables to account for failure
modes applicable to these types of facilities. The current G-FMEA tables were developed
with a consideration of external tailings facilities at oil sands and coal mines in Alberta.
Differences in physiographic regions, seismic regions, expected climate change, etc., could
lead to other failure modes that should be included in the G-FMEA tables.
The G-FMEA requires that the dam be broken down into the individual elements.
Each element should be analyzed using the G-FMEA framework, which includes four
individual charts for the drainage system, foundation, dam body, and landform. The G-
FMEA is accompanied by a developed risk matrix, including a likelihood rating table,
consequence rating table, and final risk matrix. The risk matrix framework is intended
as an example of how a qualitative risk assessment method could be combined with the
G-FMEA charts to assess a closure plan. It is recommended that the risk matrix to be used
for a site should be defined by industry and regulators with considerations of technical,
social, and economic aspects. Guidelines are provided on how to colour code the risk
matrix. It is recommended that the risk matrix be accompanied by a level of confidence
rating, which is currently being developed at the University of Alberta.
This risk management tool is timely and necessary in Alberta as many tailings facilities
approach closure. It provides a systematic method for assessing the post closure risks asso-
ciated with tailings facilities, so that they can be managed, and, ultimately, aims to develop
actionable tools consistent with the Global Industry Standard on Tailings Management.
The G-FMEA framework aids in satisfying the risk management requirements required by
the Alberta Energy Regulator for the closure of a tailings dam to support these facilities
being deregistered as dams.

Author Contributions: Development of G-FMEA framework, including the G-FMEA tables, and risk
matrix methodology, H.L.S.; writing—original draft preparation, H.L.S.; writing—review and editing,
N.A.B. and R.M.; project administration, N.A.B.; supervision, N.A.B. and R.M. All authors have read
and agreed to the published version of the manuscript.
Funding: This research was funded by the Vanier Banting Secretariat and the Alberta Energy
Regulator (Service Agreement: 17SA-OP010).
Data Availability Statement: Not applicable.
Acknowledgments: The authors would like to thank Tim Eaton (Alberta Energy Regulator), Scott
Martens (Canadian Natural Resources Ltd.), and Gord McKenna (McKenna Geotechnical Inc.) for
their ongoing support throughout this research.
Conflicts of Interest: The authors declare no conflict of interest. The funders had no role in the design
of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript,
or in the decision to publish the results.
Minerals 2021, 11, 1234 21 of 35

Appendix A. G-FMEA Charts

Table A1. Drain FMEA.

Failure Mode
Element Potential Trigger/Cause Screening Assessment of Failure Mode Failure Effects
Description
Is there erosion protection in place? What is the slope of the
Sedimentation,
ditch? Is it sufficient to keep particles suspended? What is the
sloughing/slope failure of Rise in phreatic surface, increase in seepage, pond on reclamation surface,
Blockage (partial or slope of the side slopes? What is the strength of the material?
walls, beaver dam, internal erosion, global instability, flooding, blockage of drain outlets,
full) Have there been failures in this material before? Are there
continuous build up of ice toe erosion, discharge of process affected water to the environment
beavers in the area? Is the mine located in an area that could
(icing)
experience icing?
What is the slope of the side slopes? What is the strength of
Rise in phreatic surface, increase in seepage, pond on reclamation surface,
the material? Have there been failures in this material
Reduction in Sloughing/slope failure of internal erosion, global instability, reduced capacity, erosion, potential
before? Will the ditch regularly have water running through
cross-sectional area walls, excessive vegetation flooding, blockage of drain outlets, toe erosion, discharge of process affected
Perimeter ditch it or will it stay dry for a portion of the year? Are there
water to the environment
deterrents in place to prevent the growth of vegetation?
Is there erosion protection in place? Does the material have
Erosion, differential Change to water discharge velocity, creation of secondary channels, localized
Change in slope the potential to consolidate or settle over time? Is it a cut
settlement areas of erosion, instability of dam
into natural ground or is the material placed?
Rise in phreatic surface, increase in seepage, pond on reclamation surface, internal
Sand channel Freezing conditions in Are the drainage channels constructed of sand? Could the
erosion, global instability, reduced capacity, erosion, potential flooding, blockage of
buoyancy channels composed of sand channel experience freezing?
drain outlets, toe erosion, discharge of process affected water to the environment
Lack of control of phreatic surface (potential rise in phreatic surface), increase
Break in pipe, buckling, Is the pipe capable of breaking, buckling, and/or physically
Breakage of pipe in seepage, pond on reclamation surface, internal erosion, global instability,
physical degradation degrading over time?
release of water into downstream shell, erosion on downstream slope
Lack of control of phreatic surface (potential rise in phreatic surface), increase
Biological, chemical,
Pipe clogging Is chemical, biological, or sediment clogging possible? in seepage, pond on reclamation surface, internal erosion, global instability,
particulate clogging
release of water into downstream shell, erosion on downstream slope
Clogging surround Lack of control of phreatic surface (potential rise in phreatic surface), increase
Biological, chemical, Is it possible for the material surrounding the pipe to
Pipes (woven sock or sand in seepage, pond on reclamation surface, internal erosion, global instability,
or particulate clogging become clogged?
(perforated and and gravel bed) release of water into downstream shell, erosion on downstream slope
non perforated)
Blockage from perimeter
Is the pipe outlet close enough to the base of the perimeter Lack of control of phreatic surface (potential rise in phreatic surface), increase
Blockage at outlet channel or other (snow,
channel that it could become blocked? in seepage, pond on reclamation surface, internal erosion, global instability
debris, etc.)
Breakage of connection Lack of control of phreatic surface (potential rise in phreatic surface), increase
Overloading, degradation of
between a drain and How are the drain and outtake connected? in seepage, pond on reclamation surface, internal erosion, global instability,
connection, poor installation
outlet pipe release of water into downstream shell, erosion on downstream slope
Minerals 2021, 11, 1234 22 of 35

Table A1. Cont.

Failure Mode
Element Potential Trigger/Cause Screening Assessment of Failure Mode Failure Effects
Description
Increase in seepage gradients, Decreased capacity of drain resulting in failure, lack of control
Could cracks develop along drains? Is the site known for
Erosion from flowing development of preferential flow paths of phreatic surface (potential rise in phreatic surface), increase
having issues with burrowing animals? Is the material
water adjacent to drain (i.e., burrowing from in seepage, pond on reclamation surface, internal erosion,
erodible?
animals, cracks) global instability, erosion on downstream slope
Lack of control of phreatic surface (potential rise in phreatic
Is chemical, biological, or sediment clogging possible? What surface), increase in seepage, pond on reclamation surface,
Clogging Biological, chemical, particulate clogging
is the grain size distribution? internal erosion, global instability, release of water into
downstream shell, erosion on downstream slope
Change in material properties, including Lack of control of phreatic surface (potential rise in phreatic
Failure to meet drain permeability due to aging, change in To what extent is aging and weathering/degradation of the surface), increase in seepage, pond on reclamation surface,
criteria gradation due to movement with material possible? internal erosion, global instability, release of water into
Granular seepage, weathering/degradation downstream shell, erosion on downstream slope
materials
Lack of control of phreatic surface (potential rise in phreatic
Obstruction of drainage Will an obstruction prevent the drain from performing as surface), increase in seepage, pond on reclamation surface,
Snow and ice blocking outlet
at outlet intended? Is the outlet protected from a blockage? internal erosion, global instability, release of water into
downstream shell, erosion on downstream slope
Lack of control of phreatic surface (potential rise in phreatic
Deformation leading to Slow and continuous deformation under How much deformation is expected over time as the
surface), increase in seepage, pond on reclamation surface,
reduction in drain long-lasting shear and pressure forces materials above the drain consolidate and settle? Has the
internal erosion, global instability, release of water into
capacity from consolidation of overlying material impact on the drain been considered?
downstream shell, erosion on downstream slope
Lack of control of phreatic surface (potential rise in phreatic
How much settlement is expected to occur over time? Was
Crushing/breakage of Overloading drain, settlement of the surface), increase in seepage, pond on reclamation surface,
settlement accounted for in the design? Will the drain
granular drain dam internal erosion, global instability, release of water into
continued to be loaded?
downstream shell, erosion on downstream slope
Brittle rupture of geosynthetic, lack of control of phreatic
Degradation of geosynthetic over time surface (potential rise in phreatic surface), increase in seepage,
Is the geosynthetic capable of aging in the given time frame?
Aging (temperature, oxidation, hydrolytic, pond on reclamation surface, internal erosion, global
Have sufficient tests been performed to investigate this?
chemical, biological, radioactive, etc.) instability, release of water into downstream shell, erosion on
downstream slope
Geosynthetics
Reduction in thickness leading to reduction in drain capacity
or shear failure of drain if in-place deformation reaches a
Slow and continuous deformation under How much deformation is expected over time as the
critical value, lack of control of phreatic surface (potential rise
Creep deformation long-lasting shear and pressure forces materials above the geotextile consolidate and settle? Has the
in phreatic surface), increase in seepage, pond on reclamation
from consolidation of overlying material impact on the geosynthetic been considered?
surface, internal erosion, global instability, release of water
into downstream shell, erosion on downstream slope
Minerals 2021, 11, 1234 23 of 35

Table A1. Cont.

Failure Mode
Element Potential Trigger/Cause Screening Assessment of Failure Mode Failure Effects
Description
Are there dispersive soils present? Are there ferrous soils?
Does the permeant contain oily waters or sludge? Is there
Lack of control of phreatic surface (potential rise in phreatic
turbid water with high suspended solids? Is there potential
surface), increase in seepage, pond on reclamation surface,
Clogging Biological, chemical, particulate clogging for chemical precipitation or biological growth? What is the
internal erosion, global instability, release of water into
end land use (does it involve agriculture or sewage systems
downstream shell, erosion on downstream slope
that could result in clogging)? Is sediment capable of clogging
the drain? What is the grain size distribution?
Intrusion of adjacent materials (i.e., Does the downstream or exit surface of the geosynthetic have Lack of control of phreatic surface (potential rise in phreatic
Geosynthetics geotextile), blockage of downstream or the potential to be blocked and prevent drainage from surface), increase in seepage, pond on reclamation surface,
Blockage
exit surface caused by sedimentation, sediment, vegetation, ice, snow, etc.? Could adjacent internal erosion, global instability, release of water into
vegetation, etc. materials impede movement of water to the geosynthetic? downstream shell, erosion on downstream slope
Blinding where Formation of a filter cake at the interface
Lack of control of phreatic surface (potential rise in phreatic
fine-grained soils are of the geosynthetic from coarse particles
Is the geosynthetic in intimate contact with the soil? Have all surface), increase in seepage, pond on reclamation surface,
prevented from entering being retained by the geotextile and
appropriate filter criterion been followed during design? internal erosion, global instability, release of water into
the geotextile, which intercepting fine particles migrating
downstream shell, erosion on downstream slope
creates a filter cake from the soil
Is there a narrow core? Is there the potential for excessive
Decrease in total stress (i.e., differential differential settlement that could lead to a decrease in total
Hydraulic fracture settlement, arching in narrow cores), stress? Is there the potential for an increase in porewater Cracking, internal erosion, global instability
increase in porewater pressure pressure? Is there an effective downstream filter to prevent
internal erosion of the core?
High hydraulic gradients,
Is the material widely gap-graded or gap-graded non plastic? Global instability, seepage on the downstream slope,
Internal erosion in dam design/construction defect, presence of
Is there an effective downstream filter to prevent internal settlement of the crest, permeability may increase as erosion
from suffusion widely gap-graded or non-plastic
erosion of the core? progresses or decrease if clogging occurs
Low gap-graded soils
permeability
Cracks from vertical deformation in
cores
foundation, starter dyke, or other tailings
Is there a crack or gap that could allow for a concentrated leak
Internal erosion in dam materials or differential settlement;
to develop? Is there an effective downstream filter to prevent Global instability, development of a pipe
from concentrated leak tunnels created by burrowing animals;
internal erosion of the core?
hydraulic fracture; high hydraulic
gradient; design/construction defects
Global instability, static liquefaction, settlement of the crest,
Parallel flow in coarser layer to the Is there a contact between a coarse and fine-grained soil? Is
loss of stability or unravelling, eroded material can clog the
Internal erosion in dam interface between the coarse-grained and there a filter in place? Is there an effective downstream filter
permeable layer and increase the porewater pressure (could
from contact erosion fine-grained soil, high hydraulic to prevent internal erosion of the core? Is there an effective
result in hydraulic fracture and uplift of the downstream toe
gradients, design/construction defects downstream filter to prevent internal erosion of the core?
or a rise in the phreatic surface), development of a pipe
Minerals 2021, 11, 1234 24 of 35

Table A1. Cont.

Failure Mode
Element Potential Trigger/Cause Screening Assessment of Failure Mode Failure Effects
Description
Degradation/weathering, porewater Is there potential for weathering or degradation of materials?
pressure change, change in permeability Does the porewater pressure rely on drain performance?
Shear failure from Slumping of downstream slope, translational slide,
over time, failure of drains, progressive Could drains fail over time? Are the materials strain
changing shear strength rotational slide
failure of strain softening materials, softening or brittle? Is there an effective downstream filter to
brittle failure of contractive materials prevent internal erosion of the core?
Loading/unloading crest, toe, upstream,
Low or downstream; surface erosion of
permeability downstream slope; excessive and Is there potential for anthropogenic contributions (i.e.,
cores Shear failure from uncontrolled seepage through excavations or construction)? Erodibility of material? Is there Slumping of downstream slope, translational slide,
changing shear stress foundation resulting in erosion of toe; an effective downstream filter to prevent internal erosion of rotational slide
subsurface stress changes (geothermal the core?
development, in situ oil or gas
production, wastewater injection, etc.)
Vertical deformation Does the material have the potential to consolidate? How
Release of pore water and loss of height (potential for pond to
(differential or much consolidation has occurred already? How much is
Consolidation/settlement develop on reclamation surface), development of cracks
otherwise) from expected to occur? Is there an effective downstream filter to
above starter dyke, internal erosion, overtopping
consolidation prevent internal erosion of the core?

Table A2. Foundation FMEA.

Failure Mode Description Potential Trigger/Cause Screening Assessment of Failure Mode Failure Effects
What are the current hydraulic gradients and
Heave (seepage forces create zero effective stress Embankment loading, excessive rainfall, maximum possible due to geometry? What are the
Global instability
condition) embankment seepage materials present? Are there cohesionless soils
confined by an overlying lower permeability layer?
Cracking (transverse cracks - perpendicular to dam
Vertical deformation from collapse of karst
Collapse of karst formation Is there karst present in the foundation? crest are larger problems than longitudinal cracks)
formation
in dam, internal erosion in dam, crest subsidence
Will the materials in the foundation consolidate
Cracking (transverse cracks - perpendicular to dam
Vertical deformation caused by settlement of over time? How much consolidation has already
Consolidation crest are larger problems than longitudinal cracks)
material occurred? Does the material have the potential to
in dam, internal erosion in dam, crest subsidence
collapse?
Is there potential for seepage through the
Excessive/uncontrolled seepage through Erosion of downstream toe, increase in porewater
Excessive rainfall foundation? What is the permeability
foundation or foundation/dam contact pressure in dam, global instability
of the materials?
Minerals 2021, 11, 1234 25 of 35

Table A2. Cont.

Failure Mode Description Potential Trigger/Cause Screening Assessment of Failure Mode Failure Effects
Loading/unloading of foundation, earthquake,
Are there pre-existing shear planes? Is there the
Shear failure along pre-existing shear plane from subsurface stress changes (geothermal development, Slumping of downstream slope, translational slide,
potential for anthropogenic loading or unloading
changing shear stress in situ oil or gas production, wastewater injection, rotational slide, static liquefaction
events? Is the material erodible?
etc.)
Loading/unloading of foundation, earthquake,
Shear failure along new shear plane from changing subsurface stress changes (geothermal development, Is there the potential for anthropogenic loading or Slumping of downstream slope, translational slide,
shear stress in situ oil or gas production, wastewater injection, unloading events? Is the material erodible? rotational slide, static liquefaction
etc.)
Degradation/weathering, porewater pressure Are there pre-existing shear planes? Is there the
Shear failure along pre-existing shear plane from Slumping of downstream slope, translational slide,
change, progressive failure of strain softening potential for degradation or weathering of the
changing shear strength rotational slide, static liquefaction
materials, brittle failure of contractive materials material? Is the material strain softening or brittle?
Degradation/weathering, porewater pressure Is there the potential for degradation or weathering
Shear failure along new shear plane from changing Slumping of downstream slope, translational slide,
change, progressive failure of strain softening of the material? Is the material strain softening or
shear strength rotational slide, static liquefaction
materials, brittle failure of contractive materials brittle?
Failure of soil above or around a backward erosion
Static liquefaction, global instability,
Internal erosion in foundation or dam/foundation pipe to hold a roof, heave, high hydraulic gradients,
Are there non-plastic soils in the foundation? unravelling/sloughing of downstream face,
contact from global backward erosion design/construction defect, presence of non-plastic
sub vertical cavities
soils in the foundation
Heave, high hydraulic gradients,
Internal erosion in foundation or dam/foundation Are there non-plastic soils in the foundation and Enlargement of pipe, global instability,
design/construction defect, presence of non-plastic
contact from backward erosion piping soils capable of ’holding a roof’? static liquefaction
soils that are capable of holding a roof
Global instability, static liquefaction, settlement of
Parallel flow in coarser layer to the interface the crest, loss of stability or unravelling, eroded
Is there a contact between a coarse-grained and a
Internal erosion in foundation or dam/foundation between the coarse-grained and fine-grained soil, material can clog the permeable layer and increase
fine-grained soil? Is the geometrical and hydraulic
contact from contact erosion high hydraulic gradients, design/construction the porewater pressure (could result in hydraulic
condition for contact erosion met?
defect fracture and uplift of the downstream toe or a rise
in the phreatic surface), development of a pipe
High hydraulic gradients, design/construction Global instability, seepage on the downstream slope,
Internal erosion in foundation or dam/foundation Is the material widely gap-graded or gap-graded
defect, presence of widely gap-graded or settlement of the crest, permeability may increase as
contact from suffusion non plastic?
non-plastic gap-graded soils erosion progresses or decrease if clogging occurs
Fracture in foundation soil, hydraulic fracture,
high hydraulic gradient, cracks at dam/foundation
Internal erosion in foundation or dam/foundation Is there a crack or gap that could allow for a
contact from vertical deformation in foundation or Global instability, development of a pipe
contact from concentrated leak concentrated leak to develop?
poor construction practices or differential
settlement, design/construction defects
Cracking (transverse cracks - perpendicular to dam
Thawing of foundation permafrost Climate change Is there permafrost in the foundation? crest are larger problems than longitudinal cracks)
in dam, piping in dam, crest subsidence
Minerals 2021, 11, 1234 26 of 35

Table A3. Dam body FMEA.

Failure Mode Description Potential Trigger/Cause Screening Assessment of Failure Mode Failure Effects
Does the resistance to erosion rely on the vegetation? Is the
Suffocation by eroded material, forest fires, pests
area susceptible to forest fires? Are there pests/disease that Increase in surface erosion, instability of
and disease, climate change, large storm event,
could lead to vegetation destruction? What is the downstream downstream slope, global instability, change in
Destruction of vegetation anthropogenic contributions, surface erosion on
slope? Is the area remote? Are there surrounding communities overall evapotranspiration and water balance
downstream slope, evolution of vegetation over
that could lead to destruction of vegetation (i.e., recreational impacts (infiltration versus runoff)
time due to climate change
vehicles)?
Global instability, seepage on the downstream
High hydraulic gradients, design/construction
slope, settlement of the crest, permeability may
Internal erosion in dam from suffusion defect, presence of widely gap-graded or Is the material widely gap-graded or gap-graded non plastic?
increase as erosion progresses or decrease if
non-plastic, gap-graded soils
clogging occurs
Cracks from vertical deformation in foundation,
starter dyke, or other tailings materials or
Internal erosion in dam from Is there a crack or gap that could allow for a concentrated leak
differential settlement; tunnels created by Global instability, development of a pipe
concentrated leak to develop?
burrowing animals; hydraulic fracture; high
hydraulic gradient; design/construction defects
Global instability, static liquefaction, settlement of
Parallel flow in coarser layer to the interface the crest, loss of stability or unravelling, eroded
Internal erosion in dam from between the coarse-grained and fine-grained soil, Is there a contact between a coarse and fine-grained soil? Is material can clog the permeable layer and increase
contact erosion high hydraulic gradients, there a filter in place? the porewater pressure (could result in hydraulic
design/construction defects fracture and uplift of the downstream toe or a rise
in the phreatic surface), development of a pipe
Seismic events in area? Induced seismicity? Density of
Earthquakes, induced seismicity, construction
Dynamic liquefaction material (contractive or dilative)? Saturated or unsaturated? Global instability, local slumps, crest drops
traffic, blasting
Hydraulically placed or compacted?
Degradation/weathering, porewater pressure Is there potential for weathering or degradation of materials?
change, change in permeability over time, failure of Does the porewater pressure rely on drain performance? Slumping of downstream slope, translational slide,
Shear failure from changing shear strength
drains, progressive failure of strain softening Could drains fail over time? Are the materials strain softening rotational slide
materials, brittle failure of contractive materials or brittle?
Loading/unloading crest, toe, upstream,
or downstream; surface erosion of downstream slope;
excessive and uncontrolled seepage through Is there potential for anthropogenic contributions (i.e., Slumping of downstream slope, translational slide,
Shear failure from changing shear stress
foundation resulting in erosion of toe; earthquake; excavations or construction)? Erodibility of material? rotational slide
subsurface stress changes (geothermal development,
in situ oil or gas production, wastewater injection, etc.)
Density of material (contractive or dilative)? Saturated or
Static liquefaction from changing mean Change in pore pressures caused by a phreatic unsaturated? Hydraulically placed or compacted? Does control
Global instability
effective stress surface change (i.e., failure of drainage system) of the phreatic surface rely on drain function? Could drains
become clogged or fail in the future?
Minerals 2021, 11, 1234 27 of 35

Table A3. Cont.

Failure Mode Description Potential Trigger/Cause Screening Assessment of Failure Mode Failure Effects
Loading/unloading; overloading, including
increasing the load, construction activities at the
crest, fill placement at toe; over steepening of
downstream slope or toe (slumping of downstream Density of material (contractive or dilative)? Hydraulically
slope from shear failure), including erosion or placed or compacted? Saturated or unsaturated? Is there a
Static liquefaction from changing
excavation of toe; foundation shear; shear in starter likelihood for anthropogenic contributions in the future (i.e., Global instability
shear stress
dyke; shear in other tailings materials; excessive unexpected construction)? Is the site remote? What is the
and uncontrolled seepage through foundation material of the downstream slope? Is there a nearby river?
resulting in erosion of toe; subsurface stress changes
(geothermal development, in situ oil or gas
production, wastewater injection, etc.)
Density of material (contractive or dilative)? Hydraulically
Static liquefaction from changing shear placed or compacted? Saturated or unsaturated? Weak layers
Lateral extrusion Global instability
stress and mean effective stress interbedded in tailings? Could the tailings ’squish’ out like
toothpaste during loading?
Density of material (contractive or dilative)? Hydraulically
placed or compacted? Saturated or unsaturated? To what
Static liquefaction from long-term change Changing shear strength caused by
extent could weathering or degradation of the material occur?
in material properties resulting in degradation/weathering, progressive failure, Global instability
Will it result in an increase or decrease in strength? Will a
changing shear strength porewater pressure change, failure of drains
change in the phreatic surface impact the strength
of the material?
Slope failures (shallow surficial movement,
Surface erosion from spring sapping Destruction of vegetation, increased seepage on
Is the material susceptible to erosion? Is there a vegetative slumps), change in downstream slope angle,
(headward erosion of gullies due to downstream slope from failure of drainage system,
cover or erosion protection? Will seepage daylight on the blockage in perimeter channel with sediment,
concentration of seepage forces at the locus increased seepage from internal erosion of starter
downstream slope? development of negative drainage, development of
of the gully which accentuates erosion) dyke or tailings deposits)
large erosion scarps
Slope failures (shallow surficial movement,
Destruction of vegetation, rainfall, melting of snow,
Surface erosion from wind and overland slumps), change in downstream slope angle,
wind, increased seepage on downstream slope from Is the material susceptible to erosion? Is there a vegetative
flow resulting in rills, gullies, or sheet blockage in perimeter channel with sediment,
failure of drainage system, increased seepage from cover or erosion protection?
erosion development of negative drainage, development of
internal erosion in embankment
large erosion scarps
Flow action from perimeter ditch or nearby river,
Slope failures (shallow surficial movement,
release of a dam from a beaver, flood event, river
Is the material susceptible to erosion? Is there a vegetative slumps), change in downstream slope angle,
changing course over time, destruction of
Toe erosion cover or erosion protection? Is there a nearby perimeter ditch blockage in perimeter channel with sediment,
vegetation, excessive and uncontrolled seepage
or river? Is there known animal activity in the area? development of negative drainage, development of
through foundation, excessive erosion from internal
large erosion scarps, beaver bafflers
erosion
Release of pore water and loss of height (potential
Does the material have the potential to consolidate? How
Vertical deformation (differential or for pond to develop on reclamation surface),
Consolidation/settlement much consolidation has occurred already? How much is
otherwise) from consolidation development of cracks above starter dyke, internal
expected to occur?
erosion, overtopping
Minerals 2021, 11, 1234 28 of 35

Table A4. Landform FMEA.

Item/Functional
Failure Mode Description Potential Trigger/Cause Screening Assessment of Failure Mode Failure Effects-End Effects
Identification
Formation of ponds on reclamation surface, overtopping, piping
Does the material have the potential to settle (increase in seepage forces and gradients), infiltration of previously
Excessive settlement Consolidation
over time? unsaturated tailings that could increase vulnerability to liquefaction
for materials previously considered "not flowable"
Cap Formation of ponds on reclamation surface, failure to direct surface
water runoff towards drainage channels, development of cracks,
Does the material have the potential to settle over
Consolidation, poor formation of preferential flow paths, localized depressions,
Differential settlement time? Are there areas that have the potential to
construction practices infiltration of previously unsaturated tailings that could increase
settle more that others?
vulnerability to liquefaction for materials previously considered
"not flowable"
Does the material have the potential to Formation of ponds on reclamation surface, overtopping, piping
Excessive settlement Consolidation
settle over time? (increase in seepage forces and gradients)
Infilled material Formation of ponds on reclamation surface, overtopping, piping
Consolidation, Are there areas that have the potential to settle
Differential settlement (increase in seepage forces and gradients), failure of drainage
poor construction practices more than others?
channels to behave as intended, localized depressions
Loading/unloading crest, toe,
Is there potential for anthropogenic contributions
Shear failure from changing slopes; surface erosion; failure of Slumping, translational slide, rotational slide, blockage of
(i.e., excavations or construction)?
shear stress underlying underlying material to drainage channels
Erodibility of material?
support hummock
Degradation/weathering,
porewater pressure change,
change in permeability over time, Is there potential for weathering or degradation of
Shear failure from changing Slumping of slopes, translational slide, rotational slide, blockage of
failure of drains, progressive materials? Are the materials strain
Hummocks shear strength drainage channels
failure of strain softening softening or brittle?
materials, brittle failure of
contractive materials
Slope failures (shallow surficial movement, slumps), change in
Surface erosion from wind
Destruction of vegetation, rainfall, Is the material susceptible to erosion? Is there a downstream slope angle, blockage in drainage channel with
and overland flow resulting in
melting of snow, wind vegetative cover or erosion protection? sediment, development of negative drainage, development of large
rills, gullies, or sheet erosion
erosion scarps
What precipitation event are the channels designed
Precipitation event larger than for? What is the chance of exceedance over Excessive erosion (erosion gullies, etc.), change in slope of drainage
Washout of erosion protection
design events (including extreme 1000 years? How susceptible are the underlying channels, erosion and release of materials underlying
(riprap)
or repeat events) materials to erosion? Was the erosion protection drainage channels
properly designed and constructed?
Drainage channels
Debris, beaver dam, icing, Are there beavers in the area? Is there a chance for
Formation of a pond upstream of the drainage channel, blockage
sedimentation, slumping from a slope failure? Could debris be carried
breakthrough resulting in flooding, overtopping from pond
Blockage (complete or partial) slope failure, ingress of vegetation, downstream and deposited in the channels
formation, revert back to a pond, piping through dam (increase in
slope failure/excessive erosion resulting in a complete or partial blockage? Is the
seepage forces and gradient)
from nearby hummock mine located in an area that could experience icing?
Minerals 2021, 11, 1234 29 of 35

Table A4. Cont.

Item/Functional
Failure Mode Description Potential Trigger/Cause Screening Assessment of Failure Mode Failure Effects-End Effects
Identification
Freezing conditions in channels Are the drainage channels constructed of sand?
Sand channel buoyancy Flooding
composed of sand Could the channel experience freezing?
Drainage channels
Do the drainage channels rely on erosion control Excessive erosion (erosion gullies, etc.), change in slope of drainage
Improper design/construction,
Erosion control failure for stability? Is there the chance for differential channel, erosion and release of materials underlying drainage
differential settlement
settlement of the channel? channel, formation of secondary channel
Precipitation event larger than What precipitation event is the outlet designed for?
design event (including extreme or What is the chance of exceedance over 1000 years?
Washout of erosion Excessive erosion (erosion gullies, etc.), change in slope of outlet,
repeat events), flood following How susceptible are the underlying materials to
protection (riprap) erosion and release of materials underlying outlet
sand channel buoyancy event in erosion? Was the erosion protection properly
drainage channel designed and constructed?
Debris, beaver dam, icing,
Are there beavers in the area? Is there a chance for
sedimentation, slumping from Formation of a pond upstream of the outlet, blockage breakthrough
a slope failure? Could debris be carried
slope failure, ingress of vegetation, resulting in flooding, overtopping from pond formation, revert back
Outlet Blockage (complete or partial) downstream and deposited in the outlet resulting
increase in depositional material to a pond, piping through dam (increase in seepage
in a complete or partial blockage? Is the mine
due to failure of erosion protection forces and gradient)
located in an area that could experience icing?
in drainage channels upstream
Freezing conditions in channels Is the outlet constructed of sand? Could the
Sand channel buoyancy Flooding
composed of sand channel experience freezing?
Does the outlet rely on erosion control for stability? Excessive erosion (erosion gullies, etc.), change in slope of outlet,
Improper design/construction,
Erosion control failure Is there the chance for differential settlement erosion and release of materials underlying outlet, formation of
differential settlement
of the channel? secondary channel
Does the resistance to erosion rely on the
vegetation? Is the area susceptible to forest fires? Increase in surface erosion, deposition of material in drainage
Suffocation by eroded material,
Are there pests/disease that could lead to channels that could lead to a blockage via sedimentation,
forest fires, pests and disease,
Vegetative cover Destruction of vegetation vegetation destruction? Is the area remote? Are development of negative drainage on reclamation surface, ponding
climate change, large storm event,
there surrounding communities that could lead to of water near dam crest, internal erosion (increase in seepage forces
anthropogenic contributions
destruction of vegetation and hydraulic gradient)
(i.e., recreational vehicles)?
Does the material have the potential to Formation of ponds on reclamation surface, overtopping, piping
Excessive settlement Consolidation
settle over time? (increase in seepage forces and gradients)
Tailings Consolidation, different material Formation of ponds on reclamation surface, overtopping, piping
Are there areas that have the potential to settle
Differential settlement properties/infilling (increase in seepage forces and gradients), failure of drainage
more than others?
techniques, etc. channels to behave as intended, localized depressions
Minerals 2021, 11, 1234 30 of 35

Appendix B. FMEA Worksheet

Table A5. G-FMEA worksheet.

Immediate-term/Short-term/Medium-term/Long-term Assessment *
Level of
Consequences Risk Rating
Confidence

Element failure consequence

Element failure consequence

Element failure consequence

If yes, is there

Human intervention

Human intervention

Human intervention
Screening sufficient data

Likelihood
Is this failure
Failure mode Failure mode Potential trig- assessment to evaluate Failure

Environment

Environment

Environment
mode

Community

Community

Community
Element description ger/cause the risk?
identification of failure applicable? effects
List any Controls Remarks
mode
resource gaps.

* Likelihood rating, consequence rating, risk rating, level of confidence, and controls must be determined for each failure mode for the short-term assessment, medium-term assessment, and long-term assessment.
Minerals 2021, 11, x FOR PEER REVIEW 29 of 32
Minerals 2021, 11, 1234 31 of 35

Appendix
Appendix C. C. Risk
Risk Matrix
Matrix Development
Development
The
The steps to colour code the
steps to colour code the risk
risk matrix
matrix outlined
outlined in in Section
Section 4.2.3
4.2.3 are
are described
described here
here
with the associated development of the example risk matrix in Figure 5 used for illustration
with the associated development of the example risk matrix in Figure 5 used for illustra-
purposes.
tion purposes.
1. The
1. The likelihood
likelihoodratingsratings (Table 5), consequence
(Table 5), consequence ratings (Table(Table
ratings 6), and6),
riskand
ratings
risk (Table
ratings7)
were
(Tableevaluated.
7) were evaluated.
2. In
2. In order
order to develop the
to develop the iso-contours
iso-contours of of equal
equal risk,
risk, quantitative
quantitative values
values from
from 0.01
0.01 to
to
10,000 were
were assigned
assignedto tothe
theconsequence
consequencecategories
categoriesand andassumed
assumedtoto have an order
have an orderof
magnitude
of magnitude increase
increase between
betweenthe the
categories. For For
categories. individual projects,
individual site-specific
projects, con-
site-specific
sequences
consequences could be considered
could be considered herehere
where therethere
where are known
are known magnitudes
magnitudes of the
ofcon-
the
sequences
consequences (i.e.,(i.e.,
financial
financialimpacts
impacts ofofenvironmental
environmentalconsequences).
consequences). Iso-contours
Iso-contours of of
equal risk were developed based on the estimated quantitative consequence measure
equal risk were developed based on the estimated quantitative consequence measure
and
and the
the provided likelihoods using
provided likelihoods using the
the definition
definition(Likelihood
(Likelihood==Risk/Consequence).
Risk/Consequence).
The iso-contours are shown in Figure A1, which show
The iso-contours are shown in Figure A1, which show thethe annualized
annualized probability
probability
plotted against the
plotted against theconsequences.
consequences.ItItisisimportant
important toto remember
remember that
that thisthis is estimation
is an an estima-
technique
tion techniqueonlyonly andand serves as aasfirst-order
serves a first-orderstep
stepforfor
colour
colour coding
codingthe thematrix.
matrix.ItIt is
is
desirable to use quantitative measures of the consequences that
desirable to use quantitative measures of the consequences that extend across the full extend across the
full range
range of categories.
of categories. Risk Risk categories
categories were were then assigned
then assigned to thetoiso-contours
the iso-contours of
of equal
equal risk (Table A6). This was used to colour code the initial
risk (Table A6). This was used to colour code the initial risk matrix shown in Figurerisk matrix shown in
Figure A2. Cells that had an iso-contour cutting through them
A2. Cells that had an iso-contour cutting through them were assigned to the higher were assigned to the
higher
risk risk category.
category.

Figure A1.
Figure A1. Iso-contours
Iso-contours of
of equal
equal risk.
risk.

TableA6.
Table A6. Assigned
Assignedrisk
riskcategories
categoriesto
toiso-contour
iso-contourrisk
risklevels.
levels.

Risk Category
Risk Category Risk Level
Risk Level
Red
Red >1 × 10−−11
>1 × 10
Orange
Orange >1×× 10
>1 10−−33
Yellow
Yellow >1×× 10
>1 10−−55
−7
Green
Green >1×× 10
>1 10−7
Minerals 2021, 11, x FOR PEER REVIEW 30 of 32
Minerals 2021,
Minerals 11, 1234
2021, 11, x FOR PEER REVIEW 30 of
32 of 35
32

FigureA2.
A2. Initial
Initial colour-coded
colour-codedrisk
riskmatrix
matrix withconsideration
consideration ofiso-contours.
iso-contours.
Figure
Figure A2. Initial colour-coded risk matrix with
with consideration of
of iso-contours.

3.
3.
3.
The initial colour-coded risk matrix was assessed
The initial colour-coded risk matrix was assessed assessed to to determine
to determine if
determine ifif itit satisfied
satisfiedthe
it satisfied theCox
the Cox
Cox
axioms
axioms (weak
(weak consistency,
consistency, betweenness,
betweenness, and
and
axioms (weak consistency, betweenness, and consistent colouring). consistent
consistent colouring).
colouring).
4.
4.
4. The risk
The
The risk matrix
risk matrix
matrix was was assessed
assessed using
was assessed
using the
using the
the Levine
Levine lettering
lettering scheme
Levine lettering
scheme [29].
scheme [29].
[29]. For Forthis,
this,loga-
For this,
log-
loga-
rithmicaxes
arithmic
rithmic axeswere
axes wereused
were usedfor
used forthe
fortheconsequence
consequenceand
the consequence andlikelihood
likelihoodaxes
and likelihood axesandandstraight-line
axes straight-lineiso-
and straight-lineiso-
contours
iso-contoursof equal
of equal risk were
risk were drawn
drawn (similar
(similartotoFigure
FigureA1).
contours of equal risk were drawn (similar to Figure A1). Following this, each area
A1). Following
Following this, each
this, each area
area
was labelled
was
was labelledwith
labelled withaaaletter
with letterasas
letter asopposed
opposed
opposed
to to
to aa colour,
colour,
a colour, as shown
as shown
as shown in Figure
in Figure
in Figure A3 (each
A3 (each
A3 (each line
line rep-
line
represents
resents
represents
a newa new letter).
letter).
a new letter). Levines
Levines
Levines
method method
method results
results in a
results in in
matrix a matrix
that is
a matrix that that
somewhatis somewhat
is somewhat unin-
unintuitive,
unin-
but this but
tuitive, prevents
this risk matrix
prevents riskusers
matrix from making
users from assumptions
making about
assumptions risksabout
based on
risks
tuitive,
the but this
colouring prevents
scheme. When risk
a matrix
risk matrixusers
is from making
constructed in assumptions
this manner, about
the risks
following
basedon
based onthe
thecolouring
colouringscheme.scheme.When Whenaarisk riskmatrix
matrixis isconstructed
constructedin inthisthismanner,
manner,the the
conclusions
following can be drawn:
conclusions can be drawn:
following conclusions can be drawn:
a. Risks
a.
a. Risks
Risksin one
in one
in oneletter category
letter
letter
category
category cancan
can only
only be distinguished
only
be distinguished
be distinguished fromfrom
from risksrisks
risks in another
in another cate-
in another
cate-
gory if they
category ifare
theymoreare than
moreone thanletter apart
one letter
gory if they are more than one letter apart (i.e., C > A, D > B). (i.e.,
apart C > A,
(i.e., CD>>A, B). D > B).
b. RisksRisksinin categories
categoriesthat thatarearezero zeroororoneoneletter
letterdistant
distant areare not
not able able to be distin-
distin-
b. Risks in categories that are zero or one letter distant are not able to be distin-
guished from
guished
guished from another
from another (i.e.,
another (i.e., itit is
(i.e., is not
it not known
is known ifif CC >> BB or
not known if C > or BB >> C).
B or B C).
> C).

FigureA3.
A3. Risk
Risk matrix
matrix based
based on
on Levine
Levine[29].
[29].
Figure
Figure A3. Risk matrix based on Levine [29].

5.
5.
5. As noted
As
As notedby
noted byDuijm,
by Duijm,another
Duijm, another
another way
way
way to develop
to to develop
develop risk
riskrisk scores
scores
scores andandand colouring
colouring
colouring is by
is by is by using
using using
basic
basic arithmetic
arithmetic (multiplication
(multiplication and and addition)
addition) based based
on on
ordinal ordinal
basic arithmetic (multiplication and addition) based on ordinal numbers assigned numbers
numbers assigned
assigned to to
to
each
eachconsequence
consequence
consequence
each andandand probability
probability category
category
probability category [31].
[31]. AsAs
[31]. As
thethe
the categorieswere
categories
categories werelogarithmically
were logarithmically
logarithmically
spaced,
spaced, the
the addition
addition ofof the
the ordinal
ordinal numbers
numbers was was used,
used, as
spaced, the addition of the ordinal numbers was used, as shown in Figure as shown
shown in in
FigureFigure A4.this
A4. A4.
For For
For
this example,
example, this this results
results in a in a
risk risk
matrixmatrix
that that
is is colour-coded
colour-coded in in
the the
same same
way
this example, this results in a risk matrix that is colour-coded in the same way as Figure way
as as Figure
Figure A2.
A2.
A2.
Minerals 2021,11,
Minerals2021, 11,xxFOR
FORPEER
PEERREVIEW
REVIEW 31
31of
of32
32
Minerals 2021,
Minerals 11, 1234
2021, 11, x FOR PEER REVIEW 31 of
33 of 35
32

Figure A4.
FigureA4.
Figure Colour-coded
A4.Colour-coded risk
Colour-codedrisk matrix
riskmatrix based
matrixbased on
basedon addition
onaddition of
additionof ordinal
ofordinal pairs.
ordinalpairs.
pairs.
Figure A4. Colour-coded risk matrix based on addition of ordinal pairs.
6. Major
Major hazard
hazard aversion was applied to the risk matrix in Figure A2 to assign a higher
6.6. Majorhazard
Major hazardaversion
aversionwas
wasapplied
appliedtotothe
therisk
riskmatrix
matrixinin FigureA2
events, Figure
A2totoassign
assignaahigher
higher
risk
risk rating
risk
rating to
rating
to high-probability–low-consequence
to
high-probability–low-consequence
high-probability–low-consequence events,
events,as
as these
as
these events
these
events
events were
were consid-
were
consid-
consid-
risk
eredrating to greater
high-probability–low-consequenceFigureevents, as these events were consid-
ered
ered to
to be
to
be of
be
of
of greater
greaterconcern,
concern, as
concern,
as shown
as
shown in
shown
in Figure
in Figure A5.
A5.
A5.
ered to be of greater concern, as shown in Figure A5.

Figure A5.
FigureA5. Colour-coded
A5.Colour-coded risk
Colour-codedrisk matrix
riskmatrix following
matrixfollowing application
followingapplication of
applicationof hazard
ofhazard aversion.
hazardaversion.
aversion.
Figure
Figure A5. Colour-coded risk matrix following application of hazard aversion.
7.7.
7. Following
Followingthe
Following theapplication
the applicationof
application ofmajor
of majorhazard
major hazardaversion,
hazard the
aversion,the
aversion, risk
therisk matrix
riskmatrix was
matrixwas stress
wasstress tested
stresstested
tested
7. Following the application of major hazard aversion, the risk matrix was stress tested
by
by
by evaluating
evaluating its
its
itsperformance
performance
performance in
in different
indifferent scenarios
different scenarios
scenariosand
and evaluating
andevaluating
evaluatingwhat
what the
the
what risk
risk
the ma-
ma-
risk
by evaluating its performance in different scenarios and evaluating what the risk ma-
trix
trixtold
matrixtold the
the
told user.
user. The
The
the user. evaluation
evaluation showed
showed
The evaluation that
that
showed the
the yielded
yielded
that risk
riskratings
the yieldedratings from
riskfrom Figure
Figure
ratings A5
fromA5
trix told the user. The evaluation showed that the yielded risk ratings from Figure A5
for
forthe
the
Figure ‘Slight’
‘Slight’
A5 and
and
for the ‘Minor’
‘Minor’
‘Slight’ consequence
consequence
and rating
rating
‘Minor’ consequence columns
columns were
were
rating too
toohigh.
high.
columns Some
Some
were amend-
tooamend-
high.
for the ‘Slight’ and ‘Minor’ consequence rating columns were too high. Some amend-
ments
ments were
Some were made
made to
amendments the
the risk
towere risk
madematrix
matrix and
to theand itit was
risk was stress
stress
matrix tested
andtested again.
it wasagain. This
This
stress resulted
resulted
tested in
again.in
ments
the were
final made
example to
risk the risk
matrix matrix
in Figure and
A6.it was stress tested again. This resulted in
thefinal
This
the finalexample
examplerisk
resulted in theriskmatrix
matrixin
final exampleinFigure
FigureA6.
risk A6.
matrix in Figure A6.

Figure A6.
FigureA6. Example
A6.Example colour-coded
Examplecolour-coded risk
colour-codedrisk matrix.
riskmatrix.
matrix.
Figure
Figure A6. Example colour-coded risk matrix.
Minerals 2021, 11, 1234 34 of 35

References
1. Baker, F.; Cook, D.; Deem, J.; Letient, H.; Walsh, H. Matachewan Mines Tailings Dam from Failure to Rehabilitation. In Proceedings
of the 1996 CDA Annual Conference, Niagra Falls, ON, Canada, 6–10 October 1996.
2. Provincial Court of Alberta. Agreed Statement of Facts, Docket No, Agreed Statement of Facts, Docket No. 151258456P1 &160061354P1;
Provincial Court of Alberta, 2017; Available online: http://www1.aer.ca/compliancedashboard/enforcement/201706-04_
AgreedStatementofFacts_PrairieMines_Obed_2013-006.pdf (accessed on 6 November 2021).
3. LDI. Mining with the End in Mind: Landform Design for Sustainable Mining. Position Paper 2021-01; Landform Design Institute (LDI):
Delta, BC, Canada, 2021.
4. Government of Alberta. Alberta Dam and Canal Safety Directive, AEP Water Conservation 2018 No.3; Government of Alberta:
Edmonton, AB, Canada, 2018.
5. AER. Manual 019 Decommissioning, Closure, and Abandonment of Dams at Energy Projects; Alberta Energy Regulator (AER): Calgary,
AB, Canada, 2020.
6. Ghahramani, N.; Mitchell, A.; Rana, N.M.; McDougall, S.; Evans, S.G.; Take, W.A. Tailings-flow runout analysis: Examining the
applicability of a semi-physical area–volume relationship using a novel database. Nat. Hazards Earth Syst. Sci. 2020, 20, 3425–3438.
[CrossRef]
7. Al-Mamun, M.; Small, A. Revision of Guidance of Landforms in CDA Minings Dams Bulletin-A Companion Paper with
Additional Details. In Proceedings of the 2018 CDA Annual Conference, Quebec City, QC, Canada, 13–19 October 2018.
8. Ayyub, B.M. Risk Analysis in Engineering and Economics, 2nd ed.; CRC Press: Boca Raton, FL, USA, 2014; ISBN 978-1-4665-1826-1.
9. Robertson, A.; Shaw, S. Mine Closure; Info Mine E-Book: Vancouver, BC, Canada, 2006.
10. Valis, D.; Koucky, M. Selected Overview of Risk Assessment Techniques. Probl. Eksploat. 2009, 4, 19–32.
11. Hartford, D.N.D.; Baecher, G.B. Risk and Uncertainty in Dam Safety; Thomas Telford Publishing: London, UK, 2004;
ISBN 9780727740397.
12. Kupper, A.G.; Hurndall, B.; Morgenstern, N.; Sobkowicz, J.; Abeda, W.; Baldwin, G.; Cameron, R.; Eenkooren, N.; Grapel,
C.; Martens, S.; et al. De-Licensing Oil Sands Tailings Dams-When Is a Dam No Longer a Dam? In Proceedings of the 17th
International Conference on Tailings and Mine Waste, Banff, AB, Canada, 3–6 November 2013.
13. Dos Santos, R.N.C.; Caldeira, L.M.M.S.; Serra, J.P.B.; Dos Santos, R.C. FMEA of a tailings dam. Georisk: Assess. Manag. Risk Eng.
Syst. Geohazards 2012, 6, 89–104. [CrossRef]
14. Schafer, H.L.; Macciotta, R.; Beier, N.A. Tailings dam closure scenarios, risk communication, monitoring, and surveillance in
Alberta. CIM J. 2020, 11, 80–90. [CrossRef]
15. Schafer, H.L.; Beier, N.A.; Macciotta, R. Closure and the Long-Term Behaviour of Tailings Dams: Using Industry Experience to
Fill in the Gaps. In Proceedings of the 72nd Canadian Geotechnical Conference, St. John’s, NL, Canada, 29 September–2 October
2019.
16. Porter, M.; Lato, M.; Quinn, P.; Whittall, J. Challenges with Use of Risk Matrices for Geohazard Risk Management for Resource
Development Projects. In Proceedings of the First International Conference on Mining Geomechanical Risk, Perth, WA, Australia,
9–11 April 2019; Australian Centre for Geomechanics: Crawley, WA, Australian; pp. 71–84.
17. MEND. Cold Regions Cover System Design Technical Guidance Document, MEND Report 1.61.5c; Mine Environment Neutral
Drainage (MEND) Program: 2012. Available online: http://mend-nedem.org/wp-content/uploads/2013/01/1.61.5c.pdf
(accessed on 6 November 2021).
18. Brown, B.S. What Are the Real Risks for Tailings Facilities? In Proceedings of the First International Conference on Mining
Geomechanical Risk, Perth, WA, Australia, 9–11 April 2019; Australian Centre for Geomechanics: Crawley, WA, Australia;
pp. 21–30.
19. (Tony)Cox, L.A. What’s Wrong with Risk Matrices? Risk Anal. 2008, 28, 497–512. [CrossRef]
20. Pickering, A.; Cowley, S.P. Risk Matrices: Implied Accuracy and False Assumptions. J. Health Saf. Res. Pract. 2010, 2, 11–18.
21. Ball, D.J.; Watt, J. Further Thoughts on the Utility of Risk Matrices. Risk Anal. 2013, 33, 2068–2078. [CrossRef]
22. Monat, J.P.; Doremus, S. Alternatives to Heat Map Risk Matrices for Project Risk Prioritization. J. Mod. Proj. Manag. 2018, 6,
104–113. [CrossRef]
23. IEC. IEC/ISO 31010 Risk Management-Risk Assessment Techniques; International Electrotechnical Commission (IEC): Geneva,
Switzerland, 2009.
24. Center for Security Studies. Visualizing Risk: The Use of Graphical Elements in Risk Analysis and Communications; Center for Security
Studies: Zurich, Switzerland, 2012.
25. Baybutt, P. Guidelines for designing risk matrices. Process. Saf. Prog. 2017, 37, 49–55. [CrossRef]
26. Thomas, P.; Bratvold, R.B.; Eric, B.J. The Risk of Using Risk Matrices. SPE Econ. Manag. 2014, 6, 56–66. [CrossRef]
27. Baybutt, P. Designing risk matrices to avoid risk ranking reversal errors. Process. Saf. Prog. 2015, 35, 41–46. [CrossRef]
28. Oboni, C.; Oboni, F. Is It True that PIGs Fly When Evaluating Risks of Tailings Management Systems? In Proceedings of the 16th
International Conference on Tailings and Mine Waste, Keystone, CO, USA, 14–17 October 2012; pp. 95–105.
29. Levine, E. Improving risk matrices: The advantages of logarithmically scaled axes. J. Risk Res. 2012, 15, 209–222. [CrossRef]
30. Smith, E.D.; Siefert, W.T.; Drain, D. Risk matrix input data biases. Syst. Eng. 2008, 12, 344–360. [CrossRef]
31. Duijm, N.J. Recommendations on the use and design of risk matrices. Saf. Sci. 2015, 76, 21–31. [CrossRef]
Minerals 2021, 11, 1234 35 of 35

32. Bao, C.; Li, J.; Wu, D. A fuzzy mapping framework for risk aggregation based on risk matrices. J. Risk Res. 2016, 21, 539–561.
[CrossRef]
33. Oboni, F.; Oboni, C. Tailings Dam Management for the Twenty-First Century; Springer Nature Switzerland AG: Cham, Switzerland,
2020; ISBN 978-3-030-19447-5.
34. Canadian Geotechnical Society. Canadian Foundation Engineering Manual, 4th ed.; Canadian Geotechnical Society: Richmond, BC,
Canada, 2006; ISBN 9780920505281.
35. Global Tailings Review. Global Industry Standard on Tailings Management; International Council on Mining and Metals: London,
UK; United Nations Environment Programme: Nairobi, Kenya; Principles of Responsible Investment: London, UK, 2020.
36. ICOLD. Internal Erosion of Existing Dams, Levees and Dikes, and Their Foundations; International Commission on Large Dams
(ICOLD): Paris, France, 2017; Volume Bulletin 164.
37. Task Committee on Low-Level Radioactive Waste Management of the Technical Committee on Nuclear Effects Management of
Inactive Uranium Mill Tailings. J. Environ. Eng. 1986, 112, 490–537. [CrossRef]
38. DOE Lakeview Oregon Final Completion Report Volume 1; Department of Energy (DOE) Albuquerque Operations Office:
Albuquerque, NM, USA, 1991.
39. Mathes, D.E. Lessons-Learned from the 20-Year Uranium Mill Tailings Remedial Action Surface Project. In Proceedings of the
Waste Management, Tuscon, AZ, USA, 28 February–4 March 1999.
40. IAEA. The Long Term Stabilization of Uranium Mill Tailings; International Atomic Energy Agency (IAEA): Vienna, Austria, 2004.
41. ICOLD. Tailings Dam Committee Sustainable Design and Post-Closure Performance of Tailings Dams; International Committee on Large
Dams (ICOLD): Paris, France, 2013; Volume Bulletin 153.
42. CDA. Application of Dam Safety Guidelines to Mining Dams; Canadian Dam Association (CDA): Markham, ON, Canada, 2014.
43. OSTDC. De-Licensing of Oil Sands Tailings Dams. In Technical Guidance Document; Oil Sands Tailings Dam Committee (OSTDC).
2014. Available online: https://era.library.ualberta.ca/items/8131d11c-4402-4bed-b0ab-0d3833b95af6/view/7165e9fb-e003-4
ce6-8a92-efaf8355ec2b/2014-2003-2028-20--20Report-20--20Oil-20Sands-20Dams-20De-Licensing-20Technical-20Guide.pdf
(accessed on 6 November 2021).
44. INAP. Global Cover System Design Technical Guidance Document; International Network for Acid Prevention (INAP). 2017.
Available online: https://www.inap.com.au/wp-content/uploads/global-cover-system-design.pdf (accessed on 6 November
2021).
45. Hadjigeorgiou, J. Understanding, Managing and Communicating Geomechanical Mining Risk. In Proceedings of the First
International Conference on Mining Geomechanical Risk, Australian Centre for Geomechanics, Perth, WA, Australia, 9–11 April
2019; pp. 3–20.
46. Griffin, B. Risk Assessment. In Guidelines for Mine Waste Dump and Stockpile Design; Hawley, M., Cunning, J., Eds.; CSIRO
Publishing: Clayton South, VIC, Australia, 2017; pp. 187–195.
47. Reagan, R.T.; Mosteller, F.; Youtz, C. Quantitative meanings of verbal probability expressions. J. Appl. Psychol. 1989, 74, 433–442.
[CrossRef] [PubMed]
48. NOAMI. The Policy Framework in Canada for Mine Closure and Management of Long-Term Liabilities: A Guidance Document; National
Orphaned/Abandoned Mines Initiative (NOAMI): Ottawa, ON, Canada, 2010.
49. Slovic, P.; Weber, E.U. Perception of Risk Posed by Extreme Events. In Proceedings of the Risk Management Strategies in an
Uncertain World, Palisades, New York, NY, USA, 12–13 April 2002.
50. Gibson, G.; O’Faircheallaigh, C. IBA Community Toolkit: Negotiation and Implementation of Impact and Benefit Agreements; Walter &
Duncan Gordon Foundation: Toronto, ON, Canada, 2015.
51. Nikl, L.; Steele, J.; Robertson, L.; Davidson, S.; Schryburt, R. Environmental Consequence Classification of Mine Dam Failures:
Updating the Canadian Classification System. In Proceedings of the ASDSO 2018 Dam Safety Conference, Association of State
Dam Safety Officials, Seattle, Washington, DC, USA, 9–13 September 2018.
52. Department of Mines and Petroleum. Environmental Protection Authority Guidelines for Preparing Mine Closure Plans; Government
of Western Australia: Perth, WA, Australia, 2015.
53. Aven, T. Risk assessment and risk management: Review of recent advances on their foundation. Eur. J. Oper. Res. 2016, 253, 1–13.
[CrossRef]
54. Oboni, F.; Oboni, C. The Long Shadow of Human-Generated Geohazards: Risks and Crises. In Geohazards Caused by Human
Activity; Faris, A., Ed.; InTech: Rijeka, Croatia, 2016; pp. 107–140. ISBN 978-953-51-2802-1.