Instrumentation Tailing Dams
Instrumentation Tailing Dams
Instrumentation Tailing Dams
Tailings Dams
Luke Clarkson
Bachelor of Engineering: Dual Major in Civil and Geotechnical
Engineering
0000-0003-3696-3282
1
Abstract
The demand for tailings storage, and in turn the associated risk, is increasing exponentially with
time. It has been estimated that the potential risk posed by tailings dams increases by 20-fold
approximately every 30 years, as tailings storages become larger, are built faster and must remain
stable for longer. The social, environmental, and economic consequences of tailings dam failures
are globally significant, and more readily scrutinised and observed by the general public than ever.
As a persistent liability, how does the tailings dam avoid being dismissed as an expenditure until
such time as a serious deterioration or failure occurs?
In hindsight, there have been no tailings dam failures that are unexplainable. Through the
appropriate investigation into their root causes, each and every failure has been able to have its
mechanism described using the data available. Switching this to a proactive perspective, it is
suggested that through access to and the use of the appropriate design and construction, and
operational and monitoring approaches, tailings dam practitioners can be more readily equipped to
understand and foresee the deterioration of their unique tailings dam structure. In turn, this allows
for a safer, collaborative approach between the different stakeholders and provides a vessel for
healthy and informed discussion and decision-making. Today, challenges arise in identifying and
utilising monitoring systems to understand the complex performance and rapid behaviours of these
dams, in turn reflecting on the system’s ability to be able to predict deterioration before failure
occurs.
This thesis developed a comprehensive monitoring strategy for tailings dams. Focusing
predominantly on onsite monitoring techniques, the research explored the role that appropriate
monitoring can play in understanding and assessing the performance of the dam against its expected
behaviour. The research explored the state of practice for monitoring techniques, developed
catalogues of instrumentation types and the systems that these integrate with in practice, described
analysis techniques in relation to risk management processes in combination with the observational
approach and numerical model calibrations, and demonstrated the effectiveness of these techniques
through a number of case studies.
The research found that by appropriately acknowledging the advantages and limitations of different
monitoring methods, tailings dam practitioners have the opportunity to foresee certain types of
failure ahead of time and implement appropriate responses prior to a catastrophic consequence.
ii
Monitoring strategies and technologies need to be simple and user-friendly enough to accommodate
the cost-effective collection of information, which introduces a shared responsibility between
operators, suppliers, and designers. The data collected from different instrumentation should
fundamentally link to potential failure modes and should be used to iteratively update original
design assumptions and the design intent of a safe, stable and non-polluting tailings dam. The
research acknowledged the variability of tailings dam environments, while reiterating a focus on
core mechanisms and the understanding gained from geotechnical theory, lessons learned from
previous failures, and a structured baseline approach allows tailings dam practitioners to spend
more of their time focused on the unique challenges onsite. Collaboration between the global
community of tailings dam practitioners is needed, whereby data are proven to enable greater
understanding through real demonstration of the trends, patterns, and data behaviour.
iii
Declaration by author
This thesis is composed of my original work, and contains no material previously published or
written by another person except where due reference has been made in the text. I have clearly
stated the contribution by others to jointly-authored works that I have included in my thesis.
I have clearly stated the contribution of others to my thesis as a whole, including statistical
assistance, survey design, data analysis, significant technical procedures, professional editorial
advice, financial support and any other original research work used or reported in my thesis. The
content of my thesis is the result of work I have carried out since the commencement of my higher
degree by research candidature and does not include a substantial part of work that has been
submitted to qualify for the award of any other degree or diploma in any university or other tertiary
institution. I have clearly stated which parts of my thesis, if any, have been submitted to qualify for
another award.
I acknowledge that an electronic copy of my thesis must be lodged with the University Library and,
subject to the policy and procedures of The University of Queensland, the thesis be made available
for research and study in accordance with the Copyright Act 1968 unless a period of embargo has
been approved by the Dean of the Graduate School.
I acknowledge that copyright of all material contained in my thesis resides with the copyright
holder(s) of that material. Where appropriate I have obtained copyright permission from the
copyright holder to reproduce material in this thesis and have sought permission from co-authors for
any jointly authored works included in the thesis.
iv
Publications included in this thesis
vi
Other publications during candidature
Conference proceedings
[7] Luke Clarkson, Samuel Carneiro, Breno de Matos Castilho, David Williams, Marc Ruest, and
David Noon, State-of-the-art Monitoring Techniques for Tailings Dams, Slope Stability 2018,
Seville, Spain, XIV Congress. This publication has been incorporated as Appendix E.
[8] Luke Clarkson, Todd Armstrong, and David Williams, Effect of different tailings dam
environments and conditions on phreatic conditions, Tailings and Mine Waste 2020, Keystone,
Colorado. This publication has been incorporated as part of Chapter 6 and added to in the context of
the broader thesis.
[9] Luke Clarkson, David Williams, Pippa Redcliffe, and Freya van der Wal, Trends in Real-time
Instrumentation and Monitoring Techniques for Tailings Dams, Proceedings of Mine Waste and
Tailings Conference 2021, Brisbane, Australia. This publication has been incorporated as Appendix
D.
The following research were undertaken as part of the PhD research topic, with the intent to
leverage global contacts and previously derived knowledge to advance the respective topics.
[10] Freya van der Wal, Satellite Technology for Advanced Monitoring of Tailings Dam Facilities,
4th year thesis, The University of Queensland, Brisbane, Australia.
[11] Pippa Redcliffe, Operating Requirements and Conditions Influencing Real-time Piezometer
Monitoring of Tailings Dams, 4th year thesis, The University of Queensland, Brisbane, Australia.
vii
Contributions by others to the thesis
No contributions by others, outside of those recognised in the contribution to authorship for the
relevant sections.
No works submitted towards another degree have been included in this thesis.
viii
Acknowledgments
I would like to extend my deepest gratitude to my parents, Paul and Linda, for their unwavering
support through this research, my education, and my life. I would not be the person that I am today
without such genuine and honest guidance, advice, and love; I am ever grateful, and I hope to make
you proud.
To my supervisor, David Williams. Thank you for the opportunity and sponsorship to undertake this
research and embrace a global reach of tailings dam practitioners.
To Marc Ruest. Thank you for the encouragement, motivation, and direction through the course of
this research.
To the global community of tailings dam practitioners. Thank you for embracing this research and
its endeavour to achieve a safer mining environment for all stakeholders. I hope to continue this
relationship and culture and am optimistic that through a united approach, we can redefine the status
quo toward a safer and more sustainable future.
ix
Financial support
Financial support was provided by GroundProbe, and was used toward submission to, travel to, and
presentation of the research at conferences outside of the author’s home city.
Keywords
tailings dam, risk, geotechnical, stability, monitoring, design, field instrumentation, safety, tailings
dam failure, numerical modelling
x
Australian and New Zealand Standard Research Classifications
(ANZSRC)
xi
TABLE OF CONTENTS
Abstract ....................................................................................................................................... ii
Statement of parts of the thesis submitted to qualify for the award of another degree ........ viii
Acknowledgments ...................................................................................................................... ix
Keywords ..................................................................................................................................... x
xii
1.1. Background ................................................................................................................... 1
5.1. Catalogue of Real-time Instrumentation and Monitoring Techniques for Tailings Dams
105
5.1.1. Abstract ................................................................................................................. 105
5.1.2. Introduction ........................................................................................................... 106
5.1.3. Method .................................................................................................................. 108
5.1.4. Traditional Instrumentation and Monitoring Techniques ........................................ 108
5.1.5. Real-time Instrumentation ...................................................................................... 110
5.1.6. Online Monitoring Systems.................................................................................... 118
5.1.7. Conclusions ........................................................................................................... 119
5.1.8. Reference List ........................................................................................................ 120
5.1.9. Bibliography - Appendix B Supplementary Research ............................................. 121
5.2. Catalogue of Example Instrumentation and Monitoring Systems for Tailings Dams in
Australia .............................................................................................................................. 126
5.2.1. Abstract ................................................................................................................. 126
5.2.2. Introduction ........................................................................................................... 126
5.2.3. Method .................................................................................................................. 128
5.2.4. Real-Time Tailings Dam Monitoring ..................................................................... 129
5.2.5. Instrumentation Systems ........................................................................................ 130
5.2.6. Conclusion ............................................................................................................. 143
5.2.7. Reference List ........................................................................................................ 143
xiv
Chapter 6 The effect of different tailings dam environments and external conditions on
monitoring response .......................................................................................................... 148
xv
LIST OF FIGURES
Figure 1 (Part) Hierarchy of Failure Modes (adapted from Leonards 1982) ................................. 7
Figure 2 Heave at the toe of an embankment (Pabst et al. 2012) ................................................ 10
Figure 3 Uplift and/or blowout at the toe of an embankment (Pabst et al. 2012) ......................... 11
Figure 4 Influence of foundation permeability on phreatic surface through the embankment (Fell
et al. 2015) 11
Figure 5 Observations during piping through the foundation (Fell et al. 2015) ........................... 12
Figure 6 Factors affecting the initiation of internal erosion after USBR (2015) .......................... 15
Figure 7 Classification of soil erodibility based on flow velocity and applied shear stress against
erosion rate (Briaud 2008) ............................................................................................................. 20
Figure 8 Freeboard criterion (ANCOLD 2012) .......................................................................... 21
Figure 9 Breaching process of a granular embankment dam by overtopping: (a) cross-sectional
view, (b) view from downstream looking upstream (Zhang et al. 2016), (c) Example of a breach
growth in a sand dike (Zhao 2016) ................................................................................................ 21
Figure 10 Breaching process of a cohesive embankment dam by overtopping: (a) cross-
sectional view, (b) view from downstream looking upstream (Zhang et al. 2016), (c) Multi-level
headcut erosion developing process (Zhao 2016)........................................................................... 22
Figure 11 Flow sheet for tailings dam spillway and storage design (ANCOLD 2012) ............. 25
Figure 12 Section showing idealised pore pressure monitoring scheme in an upstream tailings
dam (Martin 2002) ........................................................................................................................ 28
Figure 13 Typical embankment design elements found in a central core design (FEMA 2014) 30
Figure 14 Classification of tailings by degree of dewatering (Davies & Rice 2001) ................. 32
Figure 15 Map showing earthquakes by magnitude since 1898 (adapted from IDV Solutions
2012) against recorded tailings dam failures due to seismicity ....................................................... 33
Figure 16 Flow sheet for seismic stability analysis (ANCOLD 2012) ...................................... 37
Figure 17 Typical undrained shear strain curves (USBR 2015b) ............................................. 38
Figure 18 Generalised drained stress strain curves (USBR 2015b) .......................................... 39
Figure 19 Historical dam stability statistics, country and type (adapted from Tailings Dam
Failures 1915-2016 and Davies, Martin, and Lighthall 2002) ........................................................ 50
Figure 20 Historical dam stability statistics, failures per year by country (adapted from Tailings
Dam Failures 1915-2016 and Davies, Martin, and Lighthall 2002) ................................................ 51
Figure 21 Historical dam stability statistics, cause of failure (adapted from Tailings Dam
Failures 1915-2016 and Davies, Martin, and Lighthall 2002) ........................................................ 51
Figure 22 Historical dam stability statistics, most significant single event release by country
xvi
(adapted from Tailings Dam Failures 1915-2016 and Davies, Martin, and Lighthall 2002) ............ 52
Figure 23 Historical dam stability statistics, recorded deaths and tailings release (adapted from
Tailings Dam Failures 1915-2016 and Davies, Martin, and Lighthall 2002) .................................. 52
Figure 24 Emergency management of dam breaks in time scale (Zhang et al. 2016)................ 74
Figure 25 Baseline estimate of trigger levels as indicated by design analysis........................... 96
Figure 26 Baseline estimate of trigger levels as indicated by back-analysis ............................. 97
Figure 27 Baseline estimate of trigger levels as indicated by regional documentation.............. 98
Figure 28 Different pinout (port) arrangements for transmission types described in this paper
135
Figure 29 Simplified operating band designation for different technologies, from TeraSense
Group (2019)139
Figure 30 Chart showing typical hydraulic conductivity of different soils (adapted from
Domenico & Schwartz 1990) ...................................................................................................... 152
Figure 31 Cross-section of standpipe and vibrating wire piezometer installation components,
respectively, from Clarkson et al. (2020) ..................................................................................... 156
Figure 32 Chart showing approximate response times for various types of piezometer in
homogeneous soils (adapted from Terzaghi and Peck 1967). * with tubing 8ft long, ** without sand
filter surrounding point................................................................................................................ 159
Figure 33 Chart showing variation of degree of equalisation with 𝝁𝑻, representing pore water
pressure measurement in situ (adapted from Gibson 1963) .......................................................... 163
Figure 34 Chart showing variation of degree of equalisation with 𝝁𝟐𝑻, representing pore water
pressure measurement insitu (adapted from Gibson 1963) ........................................................... 164
Figure 35 Location of push-in piezometer installations ......................................................... 169
Figure 36 Total Head Elevation against time for Piezometer 3C ............................................ 176
Figure 37 Total Head Elevation against time for Piezometer 3G ........................................... 177
xvii
LIST OF TABLES
Table 1 Instrumentation and monitoring for foundation failure ................................................. 13
Table 2 Foundation conditions, influence, and control measures (adapted from Fell et al. 2015)
13
Table 3 Effect of soil type on seismic response (Vucetic 1992) ................................................ 35
Table 4 Common monitoring techniques for slope instability ................................................... 41
Table 5 Shear strengths and pore pressures for static design conditions (USACE 2003) ........... 43
Table 6 Reference list of guidelines, acts, and regulations ........................................................ 54
Table 7 Comparison table of key performance/surveillance parameters - instrumentation......... 57
Table 8 Comparison table of key performance/surveillance parameters - visual........................ 59
Table 9 Frequency of reading for different types of instrumentation ......................................... 60
Table 10 Measurable stability indices correlated against geotechnical failure modes for data
aggregation potential ..................................................................................................................... 63
Table 11 Severity level impacts assessment - summary from ANCOLD Consequence Guidelines
(ISSMGE 2004) ............................................................................................................................ 72
Table 12 Recommended consequence category (ISSMGE 2004) ................................................ 73
Table 13 Frequency of reading for different types of instrumentation (adapted from Avella 1993)
80
Table 14 Comparison of common IEEE wireless protocols, from Tang & Cheung (2011) .......... 83
Table 15 Network topology types (adapted from Tang & Cheung 2011, and Mcgrath & Scanaill
2013) 84
Table 16 Summary of land-based dataline and mobile services (adapted from Tang & Cheung
2011) 85
Table 17 Major tailings dam failure influence ............................................................................ 89
Table 18 Estimated cost of instrumentation for different dam structures ..................................... 92
Table 19 Estimated costs for implementation of a real-time instrumentation system at example
dams 93
Table 20 Rating of usefulness of phreatic surface monitoring for identification of different modes
of failure 111
Table 21 Rating of usefulness of pore pressure monitoring for identification of different modes of
failure 112
Table 22 Rating of usefulness of seepage flow monitoring for identification of different modes of
failure 113
Table 23 Rating of usefulness of deformation and movement monitoring for identification of
xviii
different modes of failure ............................................................................................................ 114
Table 24 Rating of usefulness of seismicity monitoring for identification of different modes of
failure 115
Table 25 Rating of usefulness of earth pressure monitoring for identification of different modes of
failure 116
Table 26 Rating of usefulness of climate monitoring for identification of different modes of
failure 117
Table 27 Comparison of household appliance power consumption to sensor network power
consumption................................................................................................................................ 137
Table 28 Typical soil hydraulic conductivity values (adapted from Domenico & Schwartz 1990
and Vick 1990) ........................................................................................................................... 151
Table 29 Hydraulic conductivity correlations from literature (adapted from AQTESOLV 2019)
152
Table 30 Standpipe and vibrating wire piezometer instrumentation details (Clarkson et al. 2020)
154
Table 31 Theoretical concepts of different external conditions on phreatic surface (adapted from
Vick 1990) 158
Table 32 Staged construction and deposition of modelled tailings dam..................................... 167
Table 33 Material parameters adopted in SEEP/W model......................................................... 170
Table 34 External condition scenarios ...................................................................................... 172
Table 35 Instrument response time for scenario presented (based on Penman 1960) ................. 175
xix
LIST OF ABBREVIATIONS
Abbreviation Definition
4G Fourth generation cellular
AEP Annual Exceedance Probability
AHD Australian Height Datum
ALARP As Low as Reasonably Practicable
ANCOLD Australian National Committee on Large Dams
APELL Awareness and Preparedness for Emergencies at Local Level
AS/NZS Australian Standards and New Zealand Standards
AUD Australian Dollar
BEP Backwards erosion piping
CAD Canadian Dollar
CDMA Code Divided Multiple Access
CPT Cone Penetration Test
CRR Cyclic Resistance Ratio
CSR Cyclic Stress Ratio
DEM Digital Elevation Map
DIFF Differential
EDM Electronic Distance Measurement
EM Electromagnetic
EUR Euro
EV-DO Evolution-Data Optimised
FEMA Federal Emergency Management Agency
FMEA Failure Modes and Effects Analysis
FWB Full Wheatstone Bridge
GPR Ground Penetrating Radar
GPRS General Packet Radio Service
GSM Global System for Mobile communications
HSPA High Speed Packets Access
I/O Analog Input and Output
ICOLD International Committee on Large Dams
IEEE Institute of Electrical and Electronics Engineers
IP Ingress Protection
ISO International Standards Organisation
xx
ISSMGE International Society for Soil Mechanics and Geotechnical Engineering
JIGSR Joint Institution Group on Safety Risk
LAN Local Area Network
LoRa Long Range
LTE Long-term Evolution
MDE Maximum Design Earthquake
MRD Mine Residue Deposit
N/A Not Applicable
OBE Operating Basis Earthquake
OCR Overconsolidation Ratio
PAR Population at Risk
PhD Doctor of Philosophy
POT Potentiometer
QLD Queensland (a state of Australia)
RS Recommended Standard
RTU Remote Terminal Unit
SE Single Ended
SMS Short Message Services
TARP Trigger Action Response Plan
TSF Tailings Storage Facility
U Universal
UI User Interface
UNEP United Nations Environment Programme
US EPA United States Environmental Protection Agency
USB Universal Serial Bus
USBR United States Bureau of Reclamation
USD United States Dollar
VWP Vibrating Wire Piezometer
WAN Wide Area Network
xxi
Chapter 1 Introduction
1.1. Background
Tailings dams are typically earthen embankment dams designed to store mining by-products (waste)
that are the processed result of separating ore from the gangue. Transported as a slurry or paste
(depending on the degree of dewatering), the tailings material is deposited in the impoundment
while the embankment is constructed in progressive raises to remain ahead of the mining
operation’s production schedule. The tailings dam is typically a necessary liability to the mine
operation, and unfortunately history has demonstrated the catastrophic result of poor planning and
management of these structures.
Tailings dams are not subject to the same high safety standards as civil water dams, and the
investment in design, construction, and operation of the structure is also lower. However, tailings
dams do not come without their challenges. Their nature as an embankment dam introduces
fundamental risks of foundation failure, internal erosion and piping, overtopping, seepage,
seismicity, and slope instability. Further, the variable nature of tailings material itself introduces an
inconsistent profile of strength, water content, and compressibility. Depending on the material type,
tailings liquefaction can also be a risk.
Failure of the tailings dam can be, and has been catastrophic, with substantial social, environmental,
and economic consequences. For example, in some areas of the world, mining communities are
established at the low-lying base of an ore deposit residing within a mountain or mountain range:
the base being the most convenient location for access to primary income and for economic
development of a town. The simplest option for construction of a tailings dam would be to only
construct external embankments and use the natural mountain valley as the remainder of the
embankment. This in turn places the dam upstream of the mining community, immediately
introducing risk to the persons and infrastructure below.
To avoid a catastrophe, it is the shared responsibility of the many stakeholders of the dam to
continually assess the performance of the dam in actuality, against what was expected. Tailings dam
behaviour that goes outside of what was expected needs to be verified, assessed, understood, and
potentially responded to in order to prevent unfavourable or critical deterioration. To enable this
assessment, data is key. Many of the aforementioned risks occur either internal to the dam or at
relatively small magnitudes compared to the dam’s overall scale; the value of identifying failure
mechanisms with the human eye is limited, and instrumentation and monitoring is required to
supplement the assessment of dam behaviour and performance. Traditionally manual and laborious,
the capability and value of instrumentation and monitoring is progressing within the industry to the
point that real-time, online monitoring systems are being deployed which enable ready
1
measurement of tailings dam conditions. Aligning practitioner understanding of what the data
means is equally important. While this occurs in many cases, it is important to ensure that the
technology and understanding advance together as co-dependent necessities. The ability to select,
install, and monitor instrumentation, and leveraging this to understand the data and trends that are
presented from the data is critical in enabling the ability to better understand and manage tailings
dam structures. “The authors also believe strongly that the data from the instrumentation and the
inspection records must be reviewed by an experienced dam engineer as soon as they are received
from the field inspection team. The whole exercise becomes almost pointless if nobody with the
necessary knowledge looks at this field data” (Fell et al. 2015). In order for real time monitoring to
be effective, the right instrument has to be installed properly, in the right location, and data utilized
in a pre-determined manner by the appropriate experienced professional.
On hindsight review of the information available at the time, there have been no unexplained
tailings dam failures; each and every failure mechanism can be described by the information
contained in pre-existing design, construction, operation, and monitoring records. This context
suggests that the mining industry has an opportunity to improve the way that tailings dams are
managed. Combining an in-depth understanding of the potential failure modes of any particular
tailings dam and their key contributing factors with the type, function, and advantages/limitations of
different instrumentation can be used to target critical locations and parameters to allow a reliable
understanding of tailings dam behaviour to be derived. In turn, the use of monitoring and
instrumentation can allow unfavourable trends to be observed ahead of time in order to forecast
dam failure.
2
• There is insufficient data and case study learnings shared from previous failures or
instabilities. Understandably, due to the sensitive nature of these events, the data is typically
considered confidential to the organisation, however there is much value to be gained in the
sharing of trends, patterns, and observations with geotechnical practitioners. This applies
both to events that result in failure, and those that do not; events that do not result in failure
help to build the “safe” curve within which it is understood that behaviour can occur without
failure.
As described in Summary of Thesis, this thesis is concerned with the use of instrumentation and
monitoring techniques that can be used to help practitioners better understand the performance and
behaviour of their tailings dam structure. By first understanding the potential failure modes and
behaviours that lead to this (Chapter 2), practitioners can learn from previous failures (Chapter 3)
and apply learnings for the betterment of their own dam’s safety. Case studies and advancing
technology allow for a more readily available understanding of tailings dam behaviour through use
of real-time monitoring (Chapter 4, Appendix B, Appendix C and Appendix D) which can be
applied to an assortment of instrumentation and monitoring techniques (Chapter 5) connected
through the appropriate sensor networks (Chapter 5). In turn, the data collected from different
instrumentation should be used to iteratively update original design assumptions to understand any
difference between predicted and actual performance of the structure, and the effect that different
external factors may have on dam performance (Chapter 6). In hindsight, there have been no
tailings dam failures that are unexplainable using the data available. The research demonstrated the
predictive ability of instrumentation through the use of case studies (Appendix E).
4
Chapter 2 Literature Review: An overview of conventional tailings
dam geotechnical failure mechanisms
Luke Clarkson1 & David Williams1
1
Geotechnical Engineering Centre within the School of Civil Engineering, The University of
Queensland, Brisbane, QLD 4072 Australia
This paper has been published in the journal of Mining, Metallurgy and Exploration.
2.1. Abstract
The intent of this literature review is to provide overview context to allow broader understanding of
conventional tailings dam failure mechanisms. The research leverages global knowledge,
experience, and data collection and interpretation for the safe and controlled management of the
geotechnical stability of a tailings storage facility. The motivation for this review is to facilitate
transparent access to tailings dam background and understanding. This literature review addresses
the core understanding of geotechnical failure mechanisms, and how these eventuate to instability
and failure of tailings dam structures. This research focuses on foundation failure, internal erosion
and piping, overtopping, seepage, seismicity, and slope instability, and provides insight into what
factors contribute to failure, how failure progresses due to such failure, anticipating and monitoring
for the aforementioned failure modes, and designing to mitigate their risk.
2.2. Introduction
A review of a range of guidelines, acts, and regulation has concluded that a misalignment exists in
the standard of practice for tailings dam monitoring and instrumentation, globally (Clarkson and
Williams 2019). While acknowledging that many practitioners are well advanced in the field, the
aim of this paper is to establish the baseline standard of understanding tailings dam failure types.
This is expressed through the contributing factors to failure, progression of failure, ability to
monitor and anticipate failure, and how to integrate these elements into the design and assessment
of a structure. The consolidated information aims to fills the void of “you don’t know what you
don’t know”, encouraging greater global collaboration in safety by extracting best practice
documentation and references in a single paper.
With ongoing catastrophic mine tailings dam failures, the hindsight revelation of poor safety
records, and an increasing prevalence of public scrutiny and attention of mining operations, there is
5
an immediate call for enhanced safety provisions of tailings dams. It is estimated that each 1/3
century the potential risk of tailings dam failure increases by 20 fold (Robertson 2015); to address
increasing demands on waste volume, tailings storage facilities must be bigger, built faster, and to
be longer lasting.
By understanding the extent of what could go wrong, practitioners, operators, designers, suppliers,
and other various stakeholders have an opportunity to significantly improve their safety standard,
beyond what the numbers and analytical procedures may suggest.
For example, the traditional factor of safety method determines the comparative ratio between the
capacity of a system, against the induced loads. A performance-based assessment, on the other
hand, assesses the possible deformability and strength of a soil while still satisfying the
performance requirements of the structure. Hence, the intended design performance can always be
checked by monitoring, integrated with Finite Element Analyses, as well as better understanding the
likelihood of potential events as captured in a risk assessment.
It is stated that the understanding of how components behave and respond under an induced load or
condition change is more important than necessarily meeting specific code clauses (as specified
through traditional methods). The value of performance-based assessment is not in predicting
performance or estimating losses, but in contributing effectively to the reduction of losses and the
improvement of safety (Jantzer 2009).
For structures such as tailings dams, the value of this method is clear: a preceding paper by the
author (Clarkson and Williams 2019) demonstrated significant, unacceptable consequences as a
result of tailings dam failures that were all entirely predictable, in hindsight. It is hypothesized that a
better understanding of and appreciation for the structure and the significance of failure can
improve tailings dam operational practice and in turn, safety.
6
is anticipated that considerations such as long-term and post-closure safety and failure modes
(including bio-intrusion, water and wind erosion, weathering, etc.) would supplement this list for
whole-of-life planning. The ICOLD Failure Cause categories are highlighted in Figure 1 and will be
retained throughout this thesis.
While static liquefaction has become a topic of interest in recent years, the focus in mitigating the
risk associated with this has been on design considerations of potential strength reductions,
operational control and planning, and material characterisation for susceptibility to liquefaction and
to understand pre-consolidation stresses. KCB (2018) stated that ‘much of the risk depends on the
in-situ stress regime, which is difficult to measure and monitor.’ Considering the speed of static
liquefaction occurrence, and the current inability of monitoring systems to identify conditions
7
preceding static liquefaction, real-time monitoring is anticipated to be a beneficial step in the right
direction to significantly increase the frequency of monitoring, more readily understand developing
conditions in the tailings dam structure, and in turn develop further knowledge on the phenomena
and ways that the risk can be mitigated. Until a better understanding is gained and agreed in
industry on how the liquefaction failure mechanism occurs and how/if this can be monitored, as
opposed to only what might lead to liquefaction, it was not deemed appropriate to include this
advice as a certainty in this research.
9
• Pre-sheared foundation materials with low shear strength;
• Inconsistency between in-situ and design values for shear strength;
• Geomembrane liners may form a plane of weakness, particularly if separating the (pervious
rockfill) embankment from the foundation material (Davies, Lighthall, & Martin 2002b);
• Topography of the site and the degree of sloping foundations; and
• Inadequate compaction of materials against design, at construction stage.
Permeability. Impervious foundations, or those with low permeability, can experience excess pore
water pressures from surface loading and TSF activity. “Because the immediate loading is taken by
the water phase in the foundation material, there is no increase in shear strength and the rapid
increase in loading can [induce foundation failure]” (U.S. Environmental Protection Agency (US
EPA) 1994).
On the other hand, “highly permeable foundation materials… can transmit significant flow capable
of eroding material at the base of the embankment and carrying it downstream” (U.S. Bureau of
Reclamation (USBR) 2015). This erosion may result in piping, subsidence, or a reduction in shear
strength, in turn compromising the stability of the overlying embankment.
In a foundation that comprises only cohesionless soils, ‘quick’ conditions exist when the critical
vertical hydraulic gradient is achieved; the pore water pressure equals the submerged unit weight of
the soil, and hence effective stress is zero. Physically, the material may ‘heave’ or ‘boil’, as seen in
Figure 2 (USBR 2015).
In a foundation that comprises a low permeability, confining layer (such as clay) overlying a
pervious layer (such as sand), the potential for failure exists where seepage pressures through the
pervious layer exceed the overburden pressures at the downstream toe. This can cause uplift or
blowout of the confining layer as seen in Figure 3. The location of rupture may be associated with
the location of maximum uplift pressure, or a pre-existing defect in the foundation (USBR 2015).
10
Figure 3 Uplift and/or blowout at the toe of an embankment (Pabst et al. 2012)
Foundation permeability also has an influence on the phreatic surface through the embankment, as
seen in Figure 4. This may influence simplified assumptions taken in slope stability modelling.
Figure 4 Influence of foundation permeability on phreatic surface through the embankment (Fell et al. 2015)
11
of water such as smectite, montmorillonite, vermiculites, and some mixed layer minerals)
should be considered alongside other properties that can encourage the natural expansiveness
of a soil (Mokhtari & Dehghani 2012); and
• “Conversely, [expansive soils] can also become very hard when dry, resulting in shrinking
and cracking of the ground” (Mokhtari & Dehghani 2012).
Settlement and consolidation can progress failure by destabilising overlying and embedded
protection measures, tailings infrastructure, dam instrumentation, and the embankment itself.
2.4.1.3. Anticipating and monitoring foundation failure
Visual inspection criteria for assessment of foundation conditions are outlined by Fell et al. (2015):
• Drainage ditches clogged with vegetation;
• Dam areas, moisture on dry days;
• Flowing water: quantity, location, clarity;
• Boils; and
• Silt accumulations, deltas, cones.
A list of observations noted on failure or incident of the foundation due to piping has been recorded
by Fell et al. (2015), in Figure 5.
Figure 5 Observations during piping through the foundation (Fell et al. 2015)
It is important to monitor the relationship between pore pressures and reservoir level, particularly
on first filling or at historic high reservoir levels. Instrumentation for foundation performance is
recommended in Table 1.
12
Table 1 Instrumentation and monitoring for foundation failure
13
Deeply weathered rock, sometimes High permeability, low Comparatively flatter slope batter angles,
with a lateritic profile strength soil foundation with good under-drainage
Deep alluvium Excess settlement, Good filter design
differential movement and
cracking
Carbonates in impervious core Dissolution of carbonates, Material selection
mineralogy (Davies et al. 2002b) greatly increasing core
permeability
Rodent burrowing (USBR 2015) Shortening of seepage paths Environmental controls, material selection
Low density, fine-grained loess soils or Differential settlement or Site preparation and clean-up: removal or
weakly cemented “desert” soils present hydraulic fracturing through densification
within the foundation (USBR 2015) the material
Variable foundation profile (USBR Differential settlement and Site preparation and clean-up
2015) cracking of embankment
core
Poor clean-up at core-foundation rock Low density or erodible Quality assurance and quality control
surface (USBR 2015) pathway at the contact
Two predominant segments are defined for foundation preparation and clean-up requirements: the
foundation beneath the bulk of the embankment (general foundation), and the foundation
underneath the earthfill core (cutoff foundation) (Fell et al. 2015). The objectives are:
• General foundation: To provide a foundation of adequate strength by removing soft,
compressible materials (Fell et al. 2015); and
• Cutoff foundation: To provide a low permeability, non-erodible foundation consistent with
the foundation drain and filter design (Fell et al. 2015).
The author recommends the reader to the existing research (such as Fell et al. 2015) on preparation
and clean-up, and design detail on the zoning of foundations. The criticality of good record keeping
is reiterated so that “those assessing the safety of the dam in the future can be informed on what was
done” (Fell et al. 2015).
2.4.2. Internal Erosion and Piping
Wherever water is dammed by earthen embankments, the particles are exposed to a hydraulic
gradient (head loss per unit length) and seepage pressures. Seepage is typically expected and does
not necessarily present as a problem so long as it is controlled, uncontaminated, and there is no
associated particle migration (Jantzer 2009). Where the hydraulic gradient and velocity is sufficient
to overcome the geometric fabric and stability of the soil structure, particles may begin to detach,
move and migrate through the embankment, foundation, or close to abutments and infrastructure.
There are different mechanisms of internal erosion, with the most common being piping. Piping
describes the behaviour of the soil structure when internal erosion develops a continuous open
14
seepage path, which acts as an unprotected and preferential flow path for seepage.
“About two-thirds of internal erosion and piping failures and about half of internal erosion and
piping accidents occur on first filling or in the first 5 years of operation. However accidents and
failures do still occur in older dams” (Fell et al. 2015).
2.4.2.1. What factors contribute to internal erosion and piping failure?
Internal erosion initiates when an unfavourable combination of material susceptibility, stress
conditions, and hydraulic load are induced at some location of the dam (USBR 2015). The factors
affecting initiation are described in Figure 6.
Figure 6 Factors affecting the initiation of internal erosion after USBR (2015)
Guidelines for quantifying risk of different conditions are outlined in detail by USBR (2015),and
Fannin and Slangen (2014). USBR (2015) has detailed a list of parameters of influence:
15
Material Properties
• Plasticity: inversely proportional to erosion susceptibility, related to inter-particle bond
strength;
• Gradation and particle size: the greater the particle size, the more energy required to move
particles. Broadly graded soils with a flat tail of fines, particularly if the soil is gap-graded,
are particularly susceptible to poor internal stability (USBR 2015);
• Density: a more dense soil comprises tighter particles, in which case it is harder to dislodge
soil particles and initiate erosion (USBR 2015); and
• Erodibility: highly influenced by plasticity and compacted moisture content.
Stress Conditions
• Influence of Stress Conditions on Internal Stability: stress conditions define whether a
material experiences suffusion or suffosion. In suffusion, mass loss occurs with no change in
volume and an increase in hydraulic conductivity. In suffosion, fine particles transported by
seepage flow induce soil structure collapse (FEMA 2014);
• Low Stress Zones and ‘Arching’: occur in areas of severe differential settlement, potentially
zones of tension; and
• Flaws in the Embankment and Foundation: a primary mechanism for initiation of internal
erosion.
Hydraulic Conditions
• Role of concentrated seepage: discontinuities, defects, or naturally pervious layers in the
embankment or foundation where flows concentrate in preferential paths of least resistance.
It is difficult for seepage modelling to portray these unpredicted anomalies; and
• Gradients: vertical (upward) gradients can encourage heave, uplift, or blowout, and can lead
to unfiltered exits (USBR 2015) or initiation of erosion. Horizontal (internal) gradients
through an embankment and/or foundation are critical considerations for concentrated leak
erosion, backwards erosion piping, or suffusion/suffosion.
2.4.2.2. How does failure progress due to internal erosion and piping?
Failure can develop in response to three general loading scenarios:
• Static/normal operation (poor water management);
• Hydrologic (flood, unpredicted reservoir levels); or
• Seismic (deformation/cracking providing initial pathway for erosion).
“Historically, most internal erosion failures have occurred when the reservoir was within about [1
metre] of the historical maximum level or greater” (USBR 2015). USBR (2015) summarise factors
that have the potential to contribute to the development of internal erosion and piping:
16
• Backward erosion piping (BEP): erosion starts at the seepage exit point and erodes upstream
(backwards). With distance upstream, the process generates shorter seepage paths, higher
hydraulic gradients, more flow, and an increased erosion potential. Four conditions are
identified as coincident characteristics of BEP (USBR 2015):
1. Flow path or source of water;
2. Unprotected or unfiltered exit;
3. Erodible material within the flow path; and
4. Continuous stable roof forms allowing the pipe to form.
• Internal migration (stoping): a void that may stope to the surface as a sinkhole, initiated due to
internal instability/suffusion, or due to open defects in foundations or embedded infrastructure.
The broadly graded, cohesionless soil cannot support a cavity roof, repeatedly collapsing until
embankment structure is compromised;
• Concentrated leak erosion: flow concentration through a pre-existing crack, potentially
caused by desiccation or differential settlement, causing erosion;
• Contact erosion: the selective erosion of fine particles in the embankment or foundation as a
result of seepage flow along the contact between the two; and
• Internal instability – Suffusion, and Suffosion: in suffusion, mass loss occurs with no change
in volume and an increase in hydraulic conductivity. In suffosion, fine particles transported
by seepage flow induce soil structure collapse (Fannin and Slangen 2014).
Unless the eroding forces are mitigated, the passage of erosion will continue and potentially
enlarge. The final phase of internal erosion and piping development is the breach: an
uncontrolled release of material. Four mechanisms are typically considered and can lead to crest
settlement and overtopping erosion (USBR 2015):
• Gross enlargement of a pipe or concentrated leak;
• Sloughing or unravelling of the downstream face;
• Sinkhole development; and
• Slope instability.
2.4.2.3. Anticipating and monitoring internal erosion piping
“Monitoring of seepage, either by visual surveillance, or measurement, is the most common means
of identifying whether internal erosion and piping has occurred” (Fell et al. 2015).
The majority of indicators that are either readily detected, or have a moderate ease of detection are
those that can be visually observed. These comprise: leakage, muddy leakage, sinkholes,
settlements, cracking, whirlpool in reservoir, increase in pore pressure and sand boils (Fell et al.
2015). If identifiers of erosion are visually observed, failure has likely already progressed beyond
17
initiation and would require immediate intervention and remediation. Intervention may include (Fell
et al. 2015):
• Sealing of eroded materials on filters/transitions which satisfy excessive erosion criteria;
• Flow limitation by an upstream dirty rockfill zone; or
• Collapse of the pipe.
Unless the installed instrumentation intersects the failure plane exactly, or the erosion daylights in
the slope face, internal erosion is very difficult to identify. However, this mechanism can be
identified as the cause to a change in pore pressure, seepage, visual appearance, or crest
settlement/slope instability: all unique failure mechanisms discussed in this paper. When
investigating the cause of these changes, thermal or geophysical methods may be employed to help
detect internal degradation. Hence, proactive identification, investigation, and intervention on the
parameter changes discussed can actively lower the consequence of failure.
2.4.2.4. Designing for internal erosion and piping
Dams with properly designed and constructed filters allow an increased likelihood of intervention in
the development of piping (Fell et al. 2015). The controls against internal erosion and piping
formation are closely aligned with those described for seepage. Best practices for design and
construction are detailed by the US Bureau of Reclamation, US Army Corps of Engineers, The
University of New South Wales, and URS (2008), and the Federal Emergency Management Agency
(FEMA 2014). In detailing the requirement for protection against erosion and cracking, (FEMA
2014) references Arthur Casagrande (1969):
“It is not possible to prevent entirely the formation of substantial tension zones and transverse
cracks in the top of the dams in the vicinity of the abutments, no matter what materials we use in the
dam. Therefore, we must defend ourselves against the effects of cracks.”
The issues induced by cracking are remedied by introduction of a filter at locations to protect
vulnerable areas. There are four main classes of filters described by (USBR 2011):
• Class I: Drainage Filters: intercept and carry away main seepage within dam and foundation.
Comprising toe drains, relief wells, and drain fields;
• Class II: Protective Filters: protect base material from eroding into other embankment zones
and to provide some drainage function in order to control pore pressure in the dam.
Comprising downstream chimneys, blankets, and transition zones;
• Class III: Choke Filters: support overlying fill (base material) from moving into pervious or
open work foundations. May be used under upstream impervious blankets (overlying
pervious foundations), or in emergency situations such as to plug whirlpools or sinks.
Comprise foundation filters and sinkhole backfill; and
18
• Class IV: Crack Stoppers: protect against cracks that occur in the embankment core,
especially caused by seismic loading and/or large deformations. Comprise upstream and
downstream chimneys.
The author references USBR (2011) for conditions that may contribute to internal erosion and
piping through an embankment, and potential inhibition through the use of filters.
2.4.3. Overtopping
Overtopping can occur from excessive inflows, malfunctioning spillway or outlet structures,
insufficient spillway capacity, dam settlement, external landslide into reservoir, (ZagonJolli 2007)
or by wind-driven waves running-up and overtopping the dam.
2.4.3.1. What factors contribute to overtopping?
Erosion resistance of the embankment material governs stability if the dam overtops. Zhang et al.
(2016) describe the mechanisms of surface erosion and the characterisation of soil erodibility as the
basis for analysis of overtopping erosion. The author refers the reader to Zhang et al. (2016) for
proposed laboratory and field tests that may be utilised for assessment of soil erodibility.
Surface erosion
Surface erosion is a physical process that occurs when water flow removes soil particles from the
surface of the dam (Zhang et al. 2016).
• Initiation mechanism of surface erosion:
1. Granular soils: Laminar (regular) surface flow attempts to drag surface particles with
dominant viscous (fluid) forces alone; and
2. Cohesive Soils: Three types of erosion are defined: pothole (clay < 30%), line (clay >
30%), and mass (relatively steeper slope angle, clay >30%). Initiation is defined by
individual particle detachment, detachment in ‘thin flakes’, and detachment in ‘chunks’
of sediment, respectively.
• Sediment transport:
1. Bed-load transport: Transport of particles by rolling, sliding, and saltating (leaping) along
the bed surface. This type of transport constitutes “5-10% of the total sediment load”
(Zhang et al. 2016); and
2. Suspended-load transport: Transport of particles suspended in the fluid, sustained against
gravity due to the upward diffusion of turbulence.
19
Characterisation of Soil Erodibility
• Critical erosive shear stress: Ease of initiation of erosion in a particular material;
• Coefficient of erodibility: Soil-specific nature of erodibility encourages site-specific
relationships between coefficient and soil properties:
1. For granular soils: gravitational force, grain size distribution, grain shape, and particle
density;
2. For cohesive soils: clay content, grain size distribution, bulk density, clay type, plasticity
index, dispersion ratio, and water chemical composition;
3. A higher level of compaction, greater cohesion, and optimum moisture content influence
soil behaviour and hence erodibility potential; and
4. Briaud (2008) and Briaud et al. (2008) have proposed a classification of soil erodibility
(Figure 7).
Figure 7 Classification of soil erodibility based on flow velocity and applied shear stress against erosion rate (Briaud 2008)
20
Figure 8 Freeboard criterion (ANCOLD 2012)
Figure 9 Breaching process of a granular embankment dam by overtopping: (a) cross-sectional view, (b) view from downstream
looking upstream (Zhang et al. 2016), (c) Example of a breach growth in a sand dike (Zhao 2016)
The breach initiation phase (Phase 1 & 2, Figure 9-a) is the process whereby overtopping flow
erodes material from the downstream face and crest, until the erosion connects with the upstream
face and begins to accelerate overtopping flows (Zhang et al. 2016). Proposed stages of initiation
21
are summarised as follows:
1. Overtopping occurs;
2. Slope erosion starts;
3. Continuous erosion likely, surface slips induced; and
4. Gradual development of a breach channel by continuous erosion is accelerated through
isolated occurrences of lateral erosion.
The breach development phase (Phase 3 to 7, Figure 9-b) defines the process becoming rapid once
the scour reaches the upstream edge of the dam crest. Failure develops by slope flattening, breach
opening, and downstream slope erosion.
Homogeneous Embankment Dams with Cohesive Materials
Defined primarily by headcut erosion, Figure 10 describes the breaching process of a cohesive
embankment dam by overtopping.
Figure 10 Breaching process of a cohesive embankment dam by overtopping: (a) cross-sectional view, (b) view from
downstream looking upstream (Zhang et al. 2016), (c) Multi-level headcut erosion developing process (Zhao 2016)
In the breach initiation phase (Phase 1 to 4, Figure 10-a), the cohesive nature of the materials could
result in near-vertical side slopes of the breach (Zhang et al. 2016). Proposed stages of initiation are
summarised as follows:
1. Initial overtopping flow progresses surficial erosion on the downstream surface, with
preferential erosion developing a series of cascading overfalls (Zhang et al. 2016);
2. A headcut (or stepped headcuts) develops from the overfall at the downstream face due to
tensile or shear failure on the over-steepened slope;
3. The cascading overfalls combine into a large overfall;
4. Breach flow concentrates pressure and associated erosive forces on the toe of the slope,
creating a reverse roller (Figure 10-c), in turn deepening the erosive effect of the large
overfall and undercutting the toe of the slope (Zhang et al. 2016);
5. As this process continues, headcut erosion retreat accelerates (Zhang et al. 2016);
6. The breach is widened gradually due to the washing away of unstable side slopes; and
7. The headcut reaches the upstream crest.
22
The breach development phase (Phase 5 to 7, Figure 10-b) occurs rapidly once the headcut reaches
the upstream crest. Failure develops by breach enlargement both in depth and laterally, and stops
only once overtopping levels are depleted or the downstream elevation increases.
2.4.3.3. Anticipating and monitoring overtopping
In anticipating and monitoring overtopping, water management is deemed the critical control.
In accordance with the roles and responsibilities outlined in an effective tailings management plan,
direct observations will assist in early detection of faults or abnormalities against anticipated
conditions. Such include (Australian Government Department of Industry, Tourism and Resources
2007):
• Position and size of the decant pond and observations relating to freeboard requirements;
• Status of leak detection systems;
• Status of secondary containment systems;
• Status of automatic flow measurement and fault alarms; and
• Condition of pump and pipeline systems.
Throughout daily operation, it is also important to monitor the weather, wave action, and filling
rate.
The initial breach location is often difficult, if not impossible, to predict prior to failure. Usually
associated with a weak point on the dam crest or downstream slope, the initiation location can be
influenced by a number of factors, including:
• Poor compaction quality of embankment material during construction;
• Unsuitable materials used in embankment construction;
• Presence of internal weak points or discontinuities; and
• Pre-existing damage to the embankment crest or slope.
2.4.3.4. Designing for overtopping
Complexity of the breach formation process is increased based on tailings dam embankment
composition and as a result of altered erosion processes of or around elements such as surface
protection measures, a concrete floodwall on the crest, or a clay core.
23
ANCOLD (2012) detail a flow sheet for tailings dam spillway and storage design (Figure 11).
Considerations to be integrated into the design for tailings storage capacity and water management
include:
Risk Assessment
• Environmental implications of any release;
• Timing of construction: primarily, avoid the wet season and consider the duration of
earthworks required;
• Physical climate;
• Political climate; and
• Importance of maintaining continuous production.
Regional
• Minimum wet season water storage allowance;
• Minimum extreme storage allowance;
• Freeboard allowances (Figure 8);
• Design flood for spillway design;
• Hydrological setting, including site catchment, water inflow/outflow, and characteristic
rainfall and flood events (Australian Government Department of Industry, Tourism and
Resources 2007); and
• Stream management and catchment diversions.
Operational
• Excess capacity contingency: “at least six months excess capacity remaining at the time the
next stage of storage capacity is expected to be available” (ANCOLD 2012);
• Impoundments and their retaining dams need to be able to accommodate extreme hydrologic
events, typically dependent on the consequence of failure of the structure;
• Some minimum allowance for decant storage; and
• Tailings water balance modelling (Australian Government Department of Industry, Tourism
and Resources 2007).
24
Figure 11 Flow sheet for tailings dam spillway and storage design (ANCOLD 2012)
25
In the traditional case, due to a restriction on the discharge of tailings material, tailings dams do not
integrate conventional spillway facilities, but rather rely on service spillways/decanting systems, or
‘emergency’/auxiliary spillways. Overtopping protection can be considered as an auxiliary spillway,
albeit in recognition of the limitations of the approach (FEMA 2014, Gibson and Sandiford 2013).
2.4.4. Seepage
The liquid component of tailings waste comprises dissolved salts, heavy metals, and other residual
chemicals from the mineralogical processes (Fell et al. 2015). There is substantial environmental
and ecological risk in the connectivity of contaminated liquid tailings (liquor) to surrounding
surface water and groundwater. Seepage can influence downstream communities and ecologies to
varying degrees, dependent on the degree of concentration, and the amount of liquor release; to
which regulatory restrictions also apply. Seepage also has the potential to reduce geotechnical
stability.
2.4.4.1. What factors contribute to seepage?
There are a number of factors that have the potential to induce fluctuation of the phreatic surface
and the associated pore water pressures in the embankment, including (from Klohn 1979, Fell et al.
2015, and ANCOLD 2012):
• Elevation difference between decant pond and surrounding ground;
• Reservoir level;
• Rainfall;
• Highly pervious zones;
• Development of cracks in the impervious core;
• Deterioration or cracking of liners and grout curtains;
• Drainage from consolidating slimes;
• Thawing of ice;
• Height of embankment and degree of dissipation by consolidation of pore pressures induced
by the embankment; and
• Dynamic loading.
2.4.4.2. How does failure progress due to seepage failure?
Seepage through tailings embankments can give rise to instability through three primary
mechanisms (Klohn 1979);
• Piping: See Internal Erosion and Piping;
• Slope instability and heaving: an increased pore pressure in the embankment or foundation of
a tailings dam can cause downstream slope instability; and
26
• Excess water losses: occur when the embankment or foundation is pervious. Connectivity of
contaminated seepage water to surrounding surface and groundwater presents substantial
environmental and ecological risks.
2.4.4.3. Anticipating and monitoring seepage
“Seepage data is one of the best indicators of dam performance” (Fell et al. 2015), making
monitoring of seepage an essential part of any tailings management strategy.
A number of the key measurement parameters specifically related to seepage are identified by Fell
et al. (2015) as:
• Seepage flow measurements;
• Pore pressure measurements;
• Measurement of reservoir water level and rainfall;
• Seepage water temperature;
• Seepage water chemistry and pH; and
• Seepage water turbidity measurement.
Complementary visual indicators of seepage in the tailings impoundment may include (Fell et al.
2015):
• Quantity, location, and clarity of seepage discharge water, which may indicate piping of the
embankment or foundation;
• Overgrowth or wet terrain vegetation on dam and within 15 metres beyond toe of dam,
indicating excessive moisture, or seasonal and pond level changes;
• Wet patches, change in local moisture on dam embankment; or
• Damp areas around conduits, outlet works, pumps.
Seepage flow and pore pressure measurements can also be instrumented, primarily through use of a
range of different piezometers. Monitoring any change over time of these instruments can give
information for use in assessment of unexpected pore pressures which may be a precursor to
different failure mechanisms like internal erosion and piping, foundation heave, or slope instability
(Fell et al. 2015). The location of each piezometer in the network is critical in obtaining meaningful
results, noting (Fell et al. 2015):
• An idealised pore pressure monitoring scheme is presented in Figure 12;
• Installing piezometers in the foundations under and downstream of the dam is favoured,
monitoring for change. Instrumentation in zones of different materials, hence different pore
pressure regimes, is also favoured;
• For ‘blow-out’ and ‘heave’ in foundations, piezometric conditions are important to
understand below lower permeability layers which act to confine the seepage flow;
27
• For landslide piezometric conditions, pore pressure on the slide plane is critical yet difficult
to intercept. However, trending phreatic elevations can be monitored;
• For jointed and sheared rock, several piezometers should be installed, intercepting the
discontinuities that groundwater flows along;
• Three elements are introduced in any scheme: observation well standpipes, pressure
piezometers, and standpipe piezometers (Martin 2002);
• Piezometer placement downstream and upstream of a cutoff or drain can give an indication
of the hydraulic gradient behaviour; and
• Caution should be taken when installing piezometers in the core of new dams or in existing
dams, both with the potential to create defects on install.
Figure 12 Section showing idealised pore pressure monitoring scheme in an upstream tailings dam (Martin 2002)
Chemical analysis can be a useful guide to the source of seepage water, for example (Fell et al.
2015):
• A comparison of ions in the reservoir water and seepage may indicate leaching of cement
from grout curtains, or oxidation of sulphides within the foundation or within the
embankment materials;
• Biological analysis can indicate the source relative to the depth in the reservoir; and
• The age of the water as indicated by analysis of tritium can indicate its sources as rainwater
or groundwater.
The monitoring of tailings dams should include surface water and groundwater quality sampling
both upstream and downstream of the facility to check against agreed trigger levels (Australian
Government Department of Industry, Tourism and Resources 2007). Parameters that are monitored
either through continuous sensors or in-field analysis include:
• Water level;
• pH;
• Total dissolved solids;
28
• Turbidity;
• Temperature;
• Dissolved oxygen;
• Conductivity; and
• Heavy metals.
Technology and geophysics represent new developments in geotechnical application. While not
anticipated to replace more conventional seepage measurements and regular inspections, some
techniques are listed as (Fell et al. 2015):
• Self-Potential: detects natural or 'spontaneous' voltages in ground materials as generated by
chemical, thermal or hydraulic processes;
• Electrical Resistivity: utilises direct currents or low frequency alternating currents to investigate
the electrical properties of the subsurface;
• Ground Penetrating Radar (GPR): transmits and receives electromagnetic waves to generate
imagery of the shallow surface profile (Olson Engineering 2005);
• Electromagnetic (EM) profiling: measures the lateral variation of ground conductivity by
inducing an alternating current at a transmitter, which interacts proportional to ground
conductivity and reports to the receiver; and
• Thermal monitoring: temperature measurements provide an indirect measurement of the
presence and behaviour of seepage flow (Fell et al. 2015).
Each method benefits from a comparison against baseline readings.
2.4.4.4. Designing for seepage
The Australian National Committee on Large Dams (ANCOLD 2003) identify key seepage
principles during the design and operation phases as “maximisation of solar drying, minimisation of
water content of tailings, and minimising the volume (and areal extent) of ponded water”. Vick
(1990) notes that even with complex lining systems, full prevention of seepage from a tailings
facility is unlikely to be achievable (Engels 2016). Two overlying systems are employed in tailings
dam design to mitigate seepage concerns (Engels 2016):
• Barrier systems (Seepage reduction): retain or resist the flow of seepage to outside the
tailings impoundment, comprising cutoff walls, upstream blankets, liners or embankment
barriers; or
• Collection systems (Seepage control): intercept and safely focus the seepage as it leaves the
tailings storage facility, comprising embankment toe drains, extraction wells, or ditch
systems.
29
FEMA (2014) proposes typical design elements in an embankment dam which contribute to
seepage control systems, as seen in Figure 13. While it is possible that the seepage control elements
will change orientation and arrangement in different tailings dam geometries, the generalised
concepts and mechanisms are introduced here for information and consideration.
Figure 13 Typical embankment design elements found in a central core design (FEMA 2014)
• Impervious Blanket: Extends the seepage path and increases the head loss zone for dams on
pervious foundations when a cutoff under the dam is impractical. Upstream blankets are
integrated into the core of the dam;
• Riprap and Bedding: Riprap protects the upstream slope of the dam against erosion caused by
reservoir wave action. Bedding under riprap protects against particle movement of the
protected zone after reservoir drawdown;
• Transition Zone: On the interior side of the upstream or downstream shells. Upstream
transition zones can also function as seismic crack stoppers;
• Impervious core: Impervious/low permeability soil that acts as a water barrier;
• Cutoff Trench: To rock or other low permeability strata that is integrated with the overlying
core;
• Cutoff Wall: Vertical water barrier in rock, also known as a grout curtain. Fills all fractures,
joints, and other openings in the rock to prevent seepage flow;
• Chimney Drain: Collects seepage coming through the chimney filter and delivers it to the
blanket drain;
• Chimney Filter: Protects the core from internal erosion and piping;
• Blanket Drain: Provides hydrostatic pressure relief for pervious foundations, outlet for
seepage collected in the chimney, and protects against particle movement in soil foundations;
• Toe Drain: Collects water from the blanket drain as well as any foundation seepage and
safely conveys it away from the embankment;
30
• Relief Well: Collects seepage water in the foundation that cannot be collected by toe drains
due to overlying impervious layers. Typically used to reduce artesian foundation pressures in
confined layers; and
• Drainage Ditch: Open trench downstream of the dam that collects seepage water. Most
effective when extending into a pervious layer.
The seepage control elements required for any particular embankment design depend on site
conditions, availability of materials, loading conditions, and economics (FEMA 2014). To
determine the most appropriate control measures, the designer must understand the behaviour of
seepage through the embankment and/or foundation. Fell et al. (2015) acknowledge parameters
involved in the assessment of seepage potential as:
• Permeability of tailings. The tailings are commonly part of the seepage path and in many
cases control the seepage rates;
• Permeability of the soil and rock underlying and surrounding the storage. This understanding
will demand greater complexity for sites where the flow paths extend beyond the storage
area; and
• Modelling of the seepage, which may involve several section models and/or a plan model.
Seepage assessment at the design stage is necessary in order to (ANCOLD 2003):
• Define pore pressures/phreatic surfaces for use in stability analysis;
• Evaluate restrictions on the rate of rise, if any;
• Determine potential impacts of seepage on the receiving environment; and
• Allow design of drainage and collection systems.
The key parameter for seepage analysis is the material saturated hydraulic conductivity (ANCOLD
2003). It is essential to capture the variation of hydraulic conductivity in different insitu
environments.
The risk of seepage has generated interest in dewatering of tailings, as classified in Figure 14. By
lowering the water content of delivered tailings, the potential seepage can be reduced (Engels
2016). Under the correct and educated considerations, the theory states that with less moisture:
seepage losses, groundwater contamination, costs, and stability are all improved.
31
Figure 14 Classification of tailings by degree of dewatering (Davies & Rice 2001)
2.4.5. Seismicity
Seismicity induces dynamic loading that is short-term, cyclic, and occurs in both horizontal and
vertical directions (Fell et al. 2015). This motion has the potential to destabilise the tailings dam
embankment, propagate internal cracking, or induce pore pressure increase and hence encourage
liquefaction in different materials.
2.4.5.1. What factors contribute to seismicity?
Earthquakes are the most common source of seismicity, where the magnitude of damage caused by
an event depends on (Fell et al. 2015):
• The seismicity of the area;
• Foundation materials and topographic conditions at the dam site;
• The type and construction of the dam;
• The water level in the reservoir at the time of the earthquake; and
• The liquefaction potential of the tailings material (Liang & Elias 2010).
One could be optimistic about the value of education, resilience, and defensive considerations in
earthquake design, where failure of tailings dams due to seismic liquefaction have reduced in
seismic-susceptible Chile “from 14% in pre-2000 cases to zero in post-2000 cases: the 2010 Chilean
earthquake of magnitude 8.8 did not cause any failure” (Azam & Li 2010).
32
Global Seismicity Trends
Figure 15 displays a spatial comparison between 203,186 earthquakes of magnitude four or greater
from 1898 through to 2003, and a by-country scale number of recorded tailings dam failures due to
seismicity. Although earthquakes are most common at tectonic plate boundaries, approximately 2%
(4,063 events, Richardson 2016) of all earthquakes occur in intra-plate regions. The core reason that
intra-plate earthquakes occur has not yet been established. Research into this is deemed important,
as the level of seismic preparedness, monitoring, and retrofitting is minimal in unaccustomed
regions, potentially causing significant damage.
Figure 15 Map showing earthquakes by magnitude since 1898 (adapted from IDV Solutions 2012) against recorded tailings
dam failures due to seismicity
Local Stress
When characterising a site, the stress orientation and magnitude of earth stresses can be aligned
with site geomorphology to assist in anticipation of potential risk. The Seismology Research Centre
(n.d.) provides a generic example, where:
• Dams are usually built in valleys;
• Valleys exist because active erosion is taking place;
• Active erosion implies there has been recent uplift;
• Under compressional tectonic force, reverse or thrust faults produce uplift;
• Reverse or thrust faults dip under the upthrown block; therefore,
• Many dams have an active fault dipping under them.
33
Induced Seismicity
Although less frequent than the natural process, ‘man-made earthquakes’ have been empirically
proven to be caused by mining and exploration activity. Sources of ‘man-made earthquakes’,
aligned with their influence on Coulomb parameters, include (Oancea 2007):
• Decrease in Normal Stress:
• Changes in tectonic forces caused by underground mining, and the associated
groundwater reduction;
• Rock burst in active and inactive underground mines; and
• Oil and gas extraction.
• Increase in Pore Pressure:
• Fluid injection and hydraulic fracturing; and
• Pore pressure increase in active faults (as a result of seepage, monsoons, or induced).
• Increase in Shear Stress:
• Reservoir-induced earthquakes (not driven by the load of the reservoir, but rather "the
increased pore water pressure in faults, leading to a reduction in shear strength over
already stressed faults" (Fell et al. 2015)).
Material Suitability
Analysis of foundation materials is conducted to determine their susceptibility to seismic induced
failure. On either static or dynamic liquefaction of saturated or partially saturated soils, the stiffness
and shear strength of a material is significantly reduced due to rapid increases in loading. The
loading reflects an increase in shear stress, accompanied by the inability for resulting pore pressures
to drain or dissipate in sufficient time. The increased pore pressure then exceeds the contact
pressure between the soil grains, disrupting the soil structure and causing it to ‘flow’. When
referencing the liquefaction characteristics of tailings, Davies, McRoberts, and Martin (2002)
identify that material will have one of four characteristics:
• Brittle strain softening (full liquefaction with the potential for limitless deformation);
contractant behaviour upon shear up to the steady-state condition;
• Limited strain softening (limited liquefaction with limited deformation); some initial
contraction followed by dilation of the tailings skeleton;
• Ductile behaviour with undrained shearing but no significant degree of strain softening (no
liquefaction); and
• Strain hardening (no appreciable liquefaction or deformation); essentially pure dilation.
Divisions defined by strain thresholds are often specified, relative to “changes in cyclic stress-strain
behaviour, stiffness degradation, pore pressure generation, post-cyclic strength, and microscale
34
processes” (Díaz-Rodrigues & López-Molina 2008).
The author references literature for stress and strain losses, and transitional behaviour for different
materials in Boulanger and Idriss (2006), and the challenges in estimating undrained strengths of
materials on interaction with shear-induced pore pressures in Davies, Martin and Lighthall (2002).
Four material states of plasticity and their response to seismic loading are conceptualised by
Vucetic (1992) and presented systematically in Table 3.
Table 3 Effect of soil type on seismic response (Vucetic 1992)
Type of Soil
Non-Plastic Low Plasticity Medium Plasticity High Plasticity
Sands and non-plastic Silty clays, clayey silts, low High plasticity
Medium plasticity clays
silts plasticity clays clays
Effect on Ground Response
Amplification of Ground Motion
Insignificant or none Small, insignificant, or none
Moderate Large
(attenuation possible) (attenuation possible)
Lengthening of Predominant Period
Significant during Likely if larger pore pressures Unlikely or
Unlikely or insignificant
liquefaction process build up impossible
Degradation of Stiffness and Strength; Reduction of Bearing Capacity
Significant for normally
Large, or complete Small, or insignificant for
consolidated or small OCR Insignificant
during full liquefaction overconsolidated soils
soils
Note: Cemented, highly sensitive, and other ‘special soils’ are not included.
2.4.5.2. How does failure progress due to seismicity?
Seismicity can affect different components of a tailings dam, to varying degrees, including (Fell et
al. 2015):
• Settlement and longitudinal and transverse cracking of the embankment, particularly near the
crest of the dam: reduced freeboard, increased potential for overtopping;
• Internal erosion and piping may develop in cracks: seepage and slope stability; and
• Liquefaction or loss of shear strength due to increase in pore pressures in the embankment
and its foundations.
To greater detail, seismic loading may result in (Fell et al. 2015):
• Instability of the upstream and downstream slopes of the dam and large deformations;
• Differential settlements and cracking due to active faults passing through the dam
foundation;
• Development of open cracks [into deep circular sliding], or opening of previously closed
joints in the foundation, close to the core-foundation contact;
35
• Differential movement between the embankment, abutments, and spillway structures leading
to transverse cracks;
• Damage to outlet works passing through the embankment and differential settlements leading
to cracking; and
• Central core zone of rockfill dam exposed at the crest due to settlement and drug of
surrounding shells, causing decrease in lateral constraint and opening of cracks (Narita
2000).
2.4.5.3. Anticipating and monitoring seismicity
It is difficult to predict seismic activity because the two most important factors, the state of stress
and the rock strength at earthquake depths, cannot be measured directly (The Seismology Research
Centre n.d.). As such, seismic monitoring of a dam generally serves two purposes: to provide assess
expected versus actual performance of the dam during a seismic event (Task Committee on
Instrumentation and Monitoring Dam Performance 2000).
Earthquake ground vibration is monitored over a wide range of scales, including global, regional,
local, and microseismic using either surface or downhole instruments (Gibson and Sandiford 2013).
At the larger scales, seismometers are employed, while when addressing earthquakes at a smaller
scale but higher resolution, microseismic monitoring is used.
Against source characteristics and propagation path of waves, local site conditions are suggested to
be the most influential factor on engineering problems (Shoji, Tanii & Kamiyama 2004). Local site
conditions have a strong influence on the maximum amplitude, frequency, and duration of seismic
waves. Event-specific seismic factors to consider include the Annual Exceedance Probability
(AEP), and Design Earthquakes (Operating Basis Earthquake (OBE) and Maximum Design
Earthquake (MDE)).
2.4.5.4. Designing for seismicity
Fell et al. (2015) describe the concept of "defensive design" of embankment design for earthquakes.
Industry best practice is described by (Fell et al. 2015):
• Provide ample freeboard, above normal operating levels, to allow for settlement, slumping,
or fault movements which displace the crest;
• Use well designed and constructed filters downstream of the earthfill core to control erosion.
To ensure effectiveness in the event of large dam settlements, likely associated with
transverse cracking, filters should extend the full height of the embankment;
• Provide ample drainage zones to allow for discharge of flow through possible cracks in the
core. Ensure that at least part of the downstream zone is free draining, or that extra discharge
capacity is provided in the vertical and horizontal drains;
36
• Avoid, densify, drain, or remove potentially liquefiable materials in the foundation or in the
embankment;
• Avoid founding the dam on strain weakening clay soils, completed weathered rock, or weak
rock with the potential to strain weaken; and
• The foundation under the core should so far as practicable be shaped to avoid sharp changes
in profile across the valley, to discourage differential settlement and the associated cracking.
Further, the author refers the reader to Fell et al. (2015) for a number of minor measures that may
also be taken into consideration.
It is noted that these measures are not necessarily applicable in assessment of existing dams, where
an upgrade may be required if unsuitable conditions are present. Appropriate remedial measures and
ground improvement options are described and referenced, as also described in Fell et al. (2015).
The ANCOLD (2012) flow sheet for seismic stability analysis is observed in Figure 16.
37
Material characteristics deemed as the core to slope stability design and analysis comprise (USBR
2015b):
• Shear Strength Selection: potential influences of “sample disturbance, variability in borrow
materials, possible variations in compaction water content and density of fill materials,
anisotropy, loading rate, creep effects, and possibly partial drainage” (U.S. Army Corps of
Engineers (USACE) 2003);
• Critical State Soil Mechanics: “soils loaded under shear reach a critical density/void ratio
where there is no further change in shear stress and no further change in volume” (USBR
2015b). Normally to lightly over-consolidated soils typically contract, while dense to highly
over-consolidated soils typically dilate. On contracting, normally to lightly over-consolidated
soils generate excess pore pressures and undrained strengths are likely governing.
Conversely, dilative soils generate negative pore pressures and, unless pore pressures still do
not dissipate more quickly than loading applied (demanding complex assessment), are likely
to be governed by drained, long term conditions;
• Undrained Strength (Figure 17): fine grained soils that are loaded faster than excess pore
pressures can dissipate are assigned undrained shear strengths (USBR 2015b). Typically, this
occurs at end of construction, flooding and reservoir rise, and very rapid loading such as
during seismicity;
38
• Drained Strength (Figure 18): higher permeability materials such as sands and gravels are
expected to drain and hence dissipate any excess pore pressures near instantaneously. In
clays, loading must be slow enough to allow dissipation and hence consider a non-transient
(steady state) seepage condition;
• Pore Water Pressures and Associated Strengths: increased pore pressure within slopes results
in an increased total stress. Further, the degree of saturation of different materials can cause
the material to behave differently: the most common case is the difference between dry and
saturated material states; and
• Loading Conditions: various loading and analysis conditions should be analysed.
• External influences causing increased shear stress: removal of support (change in slope
angle, unloading, subsidence), surcharge (loading, dam raises), transitory earth stresses
(earthquakes, vibrations/blasts), lateral pressure (freezing of water in cracks, organic
intrusions); or
• Internal influences causing decreased shear strength: weathering (disintegration,
decomposition), or water regime change (increased pore pressure, drawdown).
2.4.6.2. How does failure progress due to slope instability?
Unsatisfactory slope performance for tailings dams can develop by (USACE 2003):
• Shear Failure: sliding of a portion of the embankment along a discrete surface, relative to its
adjacent mass.
• Rotational (circular) Slides: conventional assumption in stability analysis. Shear
movements may occur across zones of appreciable thickness;
39
• Translational Slides: planar sliding along the interface between stratum of differing
stiffness; and
• Compound Failure: a combination of the two sliding mechanisms, where a circular slip
surface encounters stiff stratum and the failure surface occurs along the interface, as
opposed to cutting through.
• Surface Sloughing: a surficial portion of the embankment shears and moves downstream
(USACE 2003);
• Excessive Deformation/Settlement/Subsidence: consolidation of the foundation may result in
a displacement of particles and dislodging of a uniformly constructed slope, in turn
potentially impacting the integrity of the slope. The rate of deformation is more influential
than magnitude, on stability; and
• Creep: continuous mass deformation and/or the progression of numerous small-scale slides.
Slow, long lasting, and difficult to recover.
2.4.6.3. Anticipating and monitoring slope instability
Natural events represent slope instability triggers and encourage increased monitoring due to
associated water inflows, seismicity, or environmental damage. Historical monitoring of each dam
is crucial, as unexpected variation from this may indicate a cause for concern.
Direct observation of the tailings dam embankment, or surrounding slopes, can aid in prediction of
slope instability. Parameters may include (Fell et al. 2015 & Nelson 2013):
• Cracking in the crest or slope face;
• Settlement of the crest;
• Sinkholes;
• Saturated ground in areas that were previously dry;
• Damaged infrastructure and foundations;
• Misaligned fence lines;
• Change in downstream water levels or water quality;
• Sudden decrease in water levels; and
• Unusual sounds, such as trees cracking or rock collisions may indicate moving debris on
surrounding slopes (Nelson 2013).
In addition to, or perhaps prompted by visual observation, instrumentation can be utilised on a
variety of different scales to assess and predict slope instability. Potential methods are described in
Table 4. For many instrumentation types, technology has provided the opportunity to automate and
remotely collect monitoring data. Real-time, accessible information have vast benefit in monitoring
practice.
40
Table 4 Common monitoring techniques for slope instability
41
2.4.6.4. Designing for slope instability
Evaluation of slope stability requires (USACE 2003):
• Establish design or loading conditions to which the slope may be subjected during its life;
and
• Performing analyses of stability for each of these conditions. Four conditions must be
considered:
1. During and at the end of construction;
2. Steady state seepage;
3. Rapid drawdown; and
4. Seismic.
Soil stability analysis techniques are generally classified into either limiting analysis approach
(upper and lower bound), limit equilibrium approach (factor of safety), or displacement-based
approach (finite element, boundary element, and discrete element). While the possibilities of
manually calculating slope stability through slice methods or otherwise exists, modern proprietary
software is commonly adopted to improve the speed and accuracy of slope stability model
generation.
For static analyses (load conditions 1-3), the strength envelopes are governed by drained and
undrained conditions (and effective and total stresses, respectively). Material implications for shear
strength are described in Table 5 (adapted from USACE 2003, Table 2-1 "Shear Strengths and Pore
Pressures for Static Design Conditions"). For dynamic analyses, the potential magnitude of
earthquake and manmade earthquakes are the main consideration, as described in Section 2.4.5.
42
Table 5 Shear strengths and pore pressures for static design conditions (USACE 2003)
44
Davies, M.P., Lighthall, P.C., Rice, S & Martin, T.E. 2002b, ‘Design of Tailings Dams and
Impoundments’, Proceedings of the Tailings and Mine Waste Practices, SME, Phoenix,
Arizona.
Davies, M.P. & Rice, S 2001, ‘An alternative to conventional tailings management – “dry stack”
filtered tailings’, Proceedings of Tailings and Mine Waste 2001, Fort Collins, Colorado, pp.
411-420.
Díaz-Rodrigues & López-Molina 2008, ‘Strain thresholds in Soil Dynamics’, Proceedings of the
14th World Conference on Earthquake Engineering, Beijing, China.
Engels, J 2016, Water Management Considerations for Conventional Storage, viewed 16 March
2017, <http://www.tailings.info/technical/water.htm>
Fannin, R.J., & Slangen, P 2014, ‘On the distinct phenomena of suffusion and suffosion’,
Géotechnique, vol. 4, iss. 4, pp. 289-294
Federal Emergency Management Agency (FEMA) 2014, Technical Manual: Overtopping
Protection Alternatives for Dams.
Fell, R, MacGregor, P, Stapledon, D, Bell, G & Foster, M 2015, Geotechnical Engineering of
Dams¸ 2nd edn, CRC Press/Balkema, The Netherlands.
Gibson, G & Sandiford, M 2013, ‘Seismicity and induced earthquakes’, Office of the NSW Chief
Scientist and Engineer.
Holtz R. D., Kovacs W. D., 1981. “An Introduction to Geotechnical Engineering”, Prentice Hall.
Huzjak, R.J. & Prochaska, A.B. 2012, ‘Bedrock Settlement beneath a large Embankment Dam’,
Proceedings of the Biennial Geotechnical Seminar 2012, American Society of Civil
Engineers.
ICOLD 2001, Tailings Dams. Transport. Placement. Decantation. Review and recommendations.
International Commission on Large Dams. Bulletin 101.
IDV Solutions 2012, Earthquakes since 1898, viewed 12 February 2017,
<http://uxblog.idvsolutions.com/2012/06/earthquakes-since-1898.html>
Jantzer, I 2009, ‘Critical hydraulic gradients in tailings dams: Comparison to natural analogies’,
Licentiate Thesis, Luleå University of Technology, Sweden.
Klohn, EJ 1979, ‘Seepage Control for Tailings Dams’, in Tailings and Waste Disposal-Seepage,
Contamination, Regulations, and Control, British Columbia, Canada.
Leonards, G.A. 1982. "Investigation of Failures". Sixteenth Terzaghi Lecture. ASCE Journal of
Geotechnical Engineering, Volume 108, Number GT2, February, pp. 224-283.
Liang, JZ & Elias, D 2010, ‘Seismic Evaluation of Tailings Storage Facility’, Australian
Earthquake Engineering Society, Perth, Western Australia.
Martin, T.E. 2002, Characterisation of pore pressure conditions in upstream tailings dams.
45
Mokhtari, M & Dehghani, M 2012, ‘Swell-Shrink Behavior of Expansive Soils, Damage and
Control’, Electronic Journal of Geotechnical Engineers (EJGE), vol. 17, pp. 2673 – 2682.
Narita, K 2000, ‘Design and Construction of Embankment Dams’, Aichi Institute of Technology,
Japan.
Nelson, S.A. 2013, EENS 3050: Slope Stability, Triggering Events, Mass Movement Hazards,
Online Course Notes, Tulane University, Louisiana, United States.
Oancea, D 2007, Man-Made Earthquakes, viewed 12 February 2017,
<http://technology.infomine.com/articles/1/887/earthquake.geothermal.energy/man-
made.earthquakes.aspx>
Olson Engineering 2005, Geophysical Methods, viewed 19 March 2017
<http://www.olsonengineering.com/methods/geophysical-methods.html>
Pabst, M. W., Engemoen, W. O., Hanneman, D. L., Redlinger, C. G. & Scott, G.A. 2012, ‘Heave,
uplift, and piping at the toe of embankment dams: A new perspective’, Proceedings of Dam
Safety 2012, ASDSO, Denver, CO.
Richardson, E 2016, Plate tectonics and intra-plate earthquakes, viewed 12 February 2017,
<https://www.e-education.psu.edu/earth501/content/p4_p4.html>
Robertson, A.M 2015, FMEA Risk Analysis: Failure Modes and Effects Analysis, PowerPoint
slides.
Shoji, Y, Tanii, K & Kamiyama, M 2004, ‘The Duration and Amplitude Characteristics of
Earthquake Ground Motions with Emphasis on Local Site Effects’, Proceedings of the 13th
World Conference on Earthquake Engineering, Vancouver, B.C., Canada.
Szymanski, M.B. & Davies, M.P. 2004, 'Tailings Dam Design Criteria and Safety Evaluations at
Closure', Proceedings of the British Columbia Reclamation Symposium 2004, British
Columbia.
Task Committee on Instrumentation and Monitoring Dam Performance 2000, Guidelines for
Instrumentation and Measurements for Monitoring Dam Performance, ASCE Publications.
The Seismology Research Centre n.d., Dams & Earthquakes, ESS Earth Sciences, viewed 21 April
2018, <http://www.src.com.au/earthquakes/seismology-101/dams-earthquakes/>
U.S. Army Corps of Engineers (USACE) 2003, ‘Slope Stability’, in Engineer Manuals.
U.S. Bureau of Reclamation, U.S. Army Corps of Engineers, University of New South Wales, and
URS 2008, ‘Risk Analysis for Dam Safety – A Unified Method for Estimating Probabilities
of Failure of Embankment Dams by Internal Erosion and Piping Guidance Document’,
Version D, Issue 2
46
U.S. Bureau of Reclamation (USBR) 2015, ‘Internal Erosion Risks for Embankments and
Foundations’, in Best Practice and Risk Methodology, Security, Safety, and Law
Enforcement Office – Dam Safety.
U.S. Bureau of Reclamation (USBR) 2015b, ‘Embankment Slope Instability’, in Best Practice and
Risk Methodology, Security, Safety, and Law Enforcement Office – Dam Safety.
U.S. Department of Interior Bureau of Reclamation (USBR) 2011, Design Standards No. 13:
Chapter 5 – Protective Filters, DS-13(5)-9.
U.S. Environmental Protection Agency (US EPA) 1994, Design and Evaluation of Tailings Dams,
technical report code EPA 530-R-94, US EPA, Washington, DC.
Vucetic, M 1992, ‘Soil properties and seismic response’, Proceedings of the 10th World Conference
of Earthquake Engineering, Rotterdam, Netherlands, pp. 1199-1204.
Woodward, J 2005, An Introduction to Geotechnical Processes, CRC Press.
Yuen, S.T.S (n.d.) Introduction to Foundation Engineering, The University of Melbourne, viewed
10 April 2017,
<http://people.eng.unimelb.edu.au/stsy/geomechanics_text/Ch13_foundatn.pdf>
ZagonJolli, M 2007, 'Dam Break Modelling, Risk Assessment and Uncertainty Analysis for Flood
Mitigation', PhD thesis, Delft University of Technology, the Netherlands.
Zhao, G 2016, Breach Growth in Cohesive Embankments due to Overtopping, Delft Academic
Press, the Netherlands.
Zhang, L, Peng, M, Chang, D, & Xu, Y 2016, Dam Failure Mechanisms and Risk Assessment, 1st
edn, John Wiley & Sons, Singapore
47
Chapter 3 Critical review of tailings dam monitoring best practice
Luke Clarkson1 & David Williams1
1
Geotechnical Engineering Centre within the School of Civil Engineering, The University of
Queensland, Brisbane, QLD 4072 Australia
This paper has been published in the International Journal of Mining, Reclamation, and
Environment.
3.1. Abstract
The intent of this chapter is to establish the foundation for improvement of monitoring strategies for
the safe and controlled management of the geotechnical stability of a tailings storage facility. This
chapter establishes what currently defines best practice in an effort to realign global practices and in
turn encourage greater collaboration in safety. Best practice is established through a critical review
of guidelines, acts, and regulations, a survey and analysis of 25 global tailings dam practitioners,
and recommendations are made for the development of a global industry best practice,
comprehensive monitoring strategy for tailings dams.
3.2. Introduction
The criticality of good tailings dam management is spotlighted as the complexity of projects
increases alongside a global focus on sustainable development. “Tailings storage facilities are
among the most visible legacies of a mining operation” (Australian Government Department of
Industry, Tourism and Resources 2007). The industry standard, and fundamental benchmark for a
tailings storage facility is to provide “safe, stable, and economical storage of tailings presenting
negligible public health and safety risks and acceptably low social and environmental impacts
during operation and post-closure” (Australian Government Department of Industry, Tourism and
Resources 2007, p.1). From Barker (2010), it is acknowledged that dam safety is governed by an
understanding of the potential failure modes and methods of dealing with these during all project
stages. Aligned with this, the Tailings Management Guidelines (Australian Government Department
of Industry, Tourism and Resources 2007, p.5), recognise the cause of tailings containment wall
failures, in order of prevalence, as:
• Slope instability;
• Earthquake loading;
48
• Overtopping;
• Inadequate foundations; and
• Seepage.
The demand for tailings dam growth, and in turn the associated risk, is increasing exponentially
with time. It is estimated that the potential risk of tailings dams increases by 20-fold approximately
every 30 years (Robertson 2015). To address waste volume requirements, tailings storage facilities
must be:
Bigger: higher stresses, higher strains, higher consequences;
Faster: higher pore pressures, static liquefaction, rushed constructed, less observed time; and
Longer lasting: Mine closure halts operation, but the waste facility still stands. Time-dependent
deterioration must not be excluded from consideration.
These statistics highlight the increasing requirement for good design judgement, understanding, and
execution of tailings storage facility design. With tailings dam instability ongoing in current times,
there is still much to learn and understand. In consideration of the added pressure of increased mine
waste demand and the associated implications, this study promotes the knowledge and
understanding required to ensure the rate of advancement of demand remains aligned with managed
capacity.
Currently, regions operate and demand different standards of tailings dam management and
monitoring. Yet, understanding and management practices of different failure mechanisms are not
aligned. Tailings dam failures, and consequent deaths, are still being observed as a result of
avoidable causes. In developing a comprehensive monitoring strategy for tailings dams, and in
parallel driving collaboration of a global reach of dam operators, designers, consultants, suppliers,
and stakeholders, it is envisioned that social, environmental, and economic disasters can be
mitigated as a result of global alignment of best practice.
Figure 19 Historical dam stability statistics, country and type (adapted from Tailings Dam Failures 1915-2016 and Davies,
Martin, and Lighthall 2002)
A full suite of data was not available for all failures, hence it is acknowledged that there are minor
numerical discrepancies between some categories presented in this data. Figures were created from
data available to June 2018.
3.3.1.1. Characteristics
For all recorded failures since 1915, and for those which the corresponding data is available,
statistics for the characteristics of failure can be determined. From Tailings Dam Failures 1915-
2016, an average annual rate of failure of 1 in 700 to 1 in 1,750, or two to five annual failures for
the roughly 3,500 tailings dams worldwide are described, see Davies, Martin, & Lighthall (2002)
and LePoudre (2015) for more. Comparatively, the estimated probability of failure for a
conventional dam is 1 in 10,000, from Davies, Martin, and Lighthall (2002).
From Luino and Van de Graff (1985), “tailings dam failures only gained recognition as a serious
safety hazard since the late 1960’s and early 1970’s when tailings dam failures became more
frequent than failures of conventional dams”, as is supported by Figure 20. It is theorised in Azam
50
and Li (2010) that the observed increase may be “attributed to increased mining activity
immediately after World War II to address the high global demand for metals, minerals, and raw
materials”, servicing both reconstruction in North America and Europe and new, post-colonialism
development in Asia and Africa.
Figure 20 Historical dam stability statistics, failures per year by country (adapted from Tailings Dam Failures 1915-2016
and Davies, Martin, and Lighthall 2002)
As a result of operational influence, active dam failures are the most prominent type of incident,
representing 58% of all failures (Figure 19-b). Although inactive failures occur less often, their
existence reiterates the need for post-closure monitoring and condition assessment. The 14 inactive
dam failures since the year 2000 contributed a recorded release of 1.5 million cubic metres of waste
to the 8.6 million cubic metre total inactive dam release since 1915.
Earthquakes represent the highest count for cause of failure (Figure 21), strongly weighted by the
number of events that occur in Chile. Earthquakes account for 82% of all tailings dam failures in
Chile, with the La Ligua earthquake in 1965 having the most widespread influence with 17 recorded
failures as a result. Aside from earthquakes, the cause of failure in order of prevalence for these ten
countries comprised overtopping (17%), slope instability (14%), seepage (11%), and foundation
failure (9%).
Figure 21 Historical dam stability statistics, cause of failure (adapted from Tailings Dam Failures 1915-2016 and Davies,
Martin, and Lighthall 2002)
The most significant single event release for each country (Figure 22) does not discriminate against
the type of ore being mined, comprising seven unique ore types between the ten locations. 60% of
the reported events in Figure 22 contributed greater than 70% of their respective country’s total
recorded tailings release since 1915.
51
Figure 22 Historical dam stability statistics, most significant single event release by country (adapted from Tailings Dam
Failures 1915-2016 and Davies, Martin, and Lighthall 2002)
Figure 22 details that some of the events that released a comparatively smaller amount of tailings
were in fact the most significant in terms of number of deaths. On the same note, some of the events
that released the larger amounts of tailings had no fatalities, yet were considered environmental and
cultural disasters. The lack of correlation between the number of deaths and the amount of tailings
released (Figure 23) reiterates that there is no universal template for the management and
categorisation of risk of individual tailings dams.
Figure 23 Historical dam stability statistics, recorded deaths and tailings release (adapted from Tailings Dam Failures
1915-2016 and Davies, Martin, and Lighthall 2002)
The database shows that the height of a tailings dam at failure is not directly proportional to the
amount of material release. Reported data shows that greater than half of the recorded dam failures
were while embankment height was less than 20m.
In Azam and Li (2010), the reasoning is proposed as “unconsolidated materials with high pore
pressures… [present in the dams’] early stages of development are yet to develop adequate shear
strength to counter the resisting forces”. On the other hand, it is suggested that larger dams
comprise “better design and monitoring and surveillance”, from Fell et al. (2015).
With these statistics in mind, a critical requirement for comprehensive risk assessments can be
recognised. Consideration of stakeholder and socio-economic risk with respect to the unique
location, methodology, and characteristics of every tailings dam demands tailored design,
monitoring, and management plans. Regardless of the trends and data manipulation of such an
extensive database, the most important conclusion is identified in Davies, Martin and Lighthall
(2002): there have been no unexplained failure events. Reporting gaps exist in different details, yet
“no need to question fundamental principles of engineering mechanics/hydraulics” (Davies, Martin,
52
& Lighthall 2002) is raised; the tailings dam failures were entirely predictable in hindsight.
Therefore, although every tailings dam demands a unique design, there is identified potential to
prevent failure through effective commitment to and application of correct design and operating
practices.
While different regions typically have standard practice guidelines, acts, and regulations authored
by the respective governing body, a global best standard of practice is the target of this research. It
is acknowledged that advantages and limitations apply to a generalisation across regions, an
understanding of “What could go wrong?” to different degrees and settings will allow a more
thorough understanding of the tailings storage facility structure and the associated variability. A
comparison of existing and representative best practice guidelines and standards at time of this
paper was compiled with reference to the documents listed in Table 6. The comparison results are
presented in Table 7 and Table 8.
53
Table 6 Reference list of guidelines, acts, and regulations
55
3. The criteria against which the organisation will evaluate its environmental performance, and
appropriate indicators;
4. When the monitoring and measuring shall be performed; and
5. When the results from monitoring and measurement shall be analysed and evaluated.”
The organisation shall ensure calibration and verification, evaluation of the effectiveness of the
environmental management system, communicate relevant performance information internally and
externally, and retain documented information.
It is noted that environmental monitoring techniques can also be used to monitor and assess the
geotechnical performance of the dam, including through piezometers, water quality and flow
sensors, and other overlapping opportunities.
ISO 31000 – Risk Management
‘monitoring’: continual checking, supervising, critically observing or determining the status in order
to identify change from the performance level required or expected.
‘review’: activity undertaken to determine the suitability, adequacy and effectiveness of the subject
matter to achieve established objectives.
Both monitoring and review should be a planned part of the risk management process and involved
regular checking or surveillance. It can be periodic or ad hoc.
Responsibilities for monitoring and review should be clearly defined.
The organization's monitoring and review processes should encompass all aspects of the risk
management process for the purposes of:
• Ensuring that controls are effective and efficient in both design and operation;
• Obtaining further information to improve risk assessment;
• Analysing and learning lessons from events (including near-misses), changes, trends,
successes and failures;
• Detecting changes in the external and internal context, including changes to risk criteria and
the risk itself which can require revision of risk treatments and priorities; and
• Identifying emerging risks.
“Selecting the most appropriate risk treatment option involves balancing the costs and efforts of
implementation against the benefits derived, with regard to legal, regulatory, and other requirements
such as social responsibility and the protection of the natural environments”. The information
provided in treatment plans should include:
• The reasons for selection of treatment options, including expected benefits to be gained;
• Those who are accountable for approving the plan and those responsible for implementing
the plan;
56
• Proposed actions;
• Resource requirements including contingencies;
• Performance measures and constraints;
• Reporting and monitoring requirements; and
• Timing and schedule.
Treatment plans should be integrated with the management processes of the organization and
discussed with appropriate stakeholders.
While it is recognised that Standards, Acts and Regulations identify the need for instrumentation
and monitoring strategies, the degree, complexity, and structure of these strategies tends to be
addressed in Guideline documents. To this extent, an assessment of guideline documents unveiled
prescription of instrumentation, as presented in Table 7, but a distinct lack of the "why" or "how" to
accompany this. There is limited value in prescribing instrumentation if the objective is not
described. Is there an inherent assumption that all engineers will know each and every parameter to
monitor for? Integration of content such as that in a recent publication by Fell et al. (2015),
Geotechnical Engineering of Dams, is recommended to contribute significantly to informing this
gap.
This exercise is not to highlight shortcomings or excellence in different guidelines, acts, and
regulations, but rather the author calls for a uniform standard of communication, expectation, and
overall practice by which tailings dams are operated and maintained. There is unique value in a
number of different documents, from which the author has attempted to consolidate, align, describe
with uniform language, and ultimately derive the best potential value from the works to date. For
the purpose of this exercise, and in highlighting the gaps that exist in the referenced documents
(Table 6), the author has not supplemented these with suitable results.
Table 7 Comparison table of key performance/surveillance parameters - instrumentation
57
lateral) measuring points, settlement plates/sensors,
tiltmeters
Internal movement Vertical internal movement devices, baseplate,
settlement sensors, inclinometer, tiltmeter,
No specification.
multi-point borehole extensometer, shear strip,
radiosonde
Seismicity Seismic recording stations; seismograph:
No specification. strong-motion accelerograph, peak recording
accelerograph, seismoscope
Pore pressure Standpipes, pneumatic piezometer, dynamic
No specification.
pore pressure and liquefaction
Uplift pressures Groundwater mounding and
Piezometers, bores, pressure meters and gauges,
outward movement beneath
load cells
and surrounding the facility
Weather Rainfall measurement and
recording daily, with special
Rain gauge, thermometer, wind gauge
inspection post-major, or
prolonged rainfall
Unique parameters acknowledged in different standards/acts/regulations that the author believes identify
valuable points to consider
Ancillary tailings dam Tailings beach development, drainage, density,
interactions Tailings placement and desiccation, process water, evaporation,
capacity measurement of residue transport to site and
surplus water removed
Condition monitoring of
tailings dam infrastructure, not
only to ensure that failure of
infrastructure is not to the Communications, power supply, pipeline flow
detriment of the embankment and pressure,
dam, but also to ensure that
emergency infrastructure is
serviceable.
Temperature No specification. Thermometers, indirect measurement
Chemical analysis Environmental impact -
testing on surface drainage, Chemical analysis versus clean water
seepage, and groundwater
Biological monitoring No specification. No specification.
Dust sampling No specification. No specification.
The significance of understanding the behaviour of different visual parameters, and what constitutes
each of the observations is anticipated to be driven through training and familiarisation. There is no
value in assigning a checklist without the operator understanding the implications and elements
58
involved; you don’t know what you don’t know. Table 8 presents elements consolidated from the
reference documents in Table 6, where measurement values for different parameters are site
specific, and are recommended to be integrated with Trigger Action Response Plans (TARPs),
where possible.
Table 8 Comparison table of key performance/surveillance parameters - visual
During Normal
During Construction Initial Filling
Type of Instrument Operation
Max. Min. Max. Min. Max. Min.
Piezometers 1/week 1/month 1/day 1/week Real-time 2/year
Pore Pressure Cells 2/month 1/day 1/week 5/month 1/month
Total Pressure Cells
(vibrating wire, 1/week 1/month 1/week 2/month 3/month 1/year
pneumatic, other)
Seepage (weirs, flumes,
1/day 1/month 1/day 1/week Real-time 1/month
etc.)
Observation Wells 1/2 weeks 2/month 2/week 1/week 1/6 months
Internal Vertical Each time a unit
Frequently 1/week 1/month 3/month 1/6 years
Movement is completed
Each time a unit
Foundation Settlement 1/week 1/day 1/month 2/month 1/year
is completed
Internal Settlement 1/2 weeks 1/month 1/week 1/month 2/month 1/year
Inclinometer 1/2 weeks 1/2 months 1/week 1/month 6/month 1/year
Extensometer 2/week 1/month 2/week 1/month 3/month 4/year
Tiltmeters 1/2 weeks 1/month 1/week 1/month 6/month 4/year
Measurement Points
(embankment, structural, 1/2 weeks 1/month 2/week 1/month 3/month 1/6 years
surface)
Water Quality 1/2 weeks 1/month 1/day 1/month 6/month 1/year
Real-
Seismic Activity Real-time Real-time On event
time
Geodetic Surveys 1/month 1/week 1/month 4/year 1/10 year
Visual Inspection –
Routine by Trained 1/month 1/day 1/day 1/month
Operators
Visual Inspection –
To
Intermediate by Dam 1/year 1/2 years
occur
Engineer
To
Comprehensive Review 1/5 years 1/10 years
occur
Note: 1/day = one or a set of measurements per day; 1/2 weeks = one or a set of measurements
every 2 weeks
60
3.5. Data Aggregation
On a global scale, limitations on the effectiveness of a tailings dam monitoring system include
capital costs, ongoing maintenance, excessive labour to both maintain and read the instruments, lack
of regulatory mandate, as well as readings local only to the isolated and unique instrument. By
looking at the system through an integrated mindset, the opportunity for optimisation and in turn
counteracting existing limitations is possible.
It is important to recognise the potential to use different monitoring techniques, albeit primarily
intended to assess different tailings dam management criteria, as indicators of imminent
geotechnical failure. For example, water quality sampling can acknowledge the presence of tailings
downstream, indicating seepage, and hence a path between the two sides of the confinement
structure. This might suggest inadequate disposal, erosion, or even failure of the structure at some
point, emphasising the interconnectivity of the entire tailings management system.
The implementation and use of multiple sensors to monitor for a multitude of behaviours has
historically accepted in tailings dam design and operations. In recent years, the advancement of
technology has seen the rise of sensor networks. From Scaioni and Wang (2016), a two-fold
approach to communication between sensors in a geospatial sensor network is described. The first
describes the “technological infrastructures for data transfer” (Scaioni & Wang 2016), exploring
wireless technology alongside many other solutions. The second details “protocols for smart sensor
communications and interoperability, also through Web-services” (Scaioni & Wang 2016).
While recognising the work that has been undertaken in this field, it is acknowledged in Scaioni and
Wang (2016) that insitu case studies are lacking. Hence, it appears that while technology is
advancing and adequate to facilitate comprehensive monitoring strategies for tailings dams, the
application side in understanding of inputs and physical behaviour of parameters is not aligned.
3.5.1. Data Aggregation Theory
Theoretical modelling for derivation of the factor of safety in tailings dam design is limited by an
inability to consider the randomness of parameters beyond a sensitivity analysis. There is inherent
value in a performance based assessment of tailings dams, not eliminating the need for detailed
design phases and theory, but rather combining these with an understanding gained from observing
in-situ performance. Oftentimes, design limits are carried directly across to the Trigger Action
Response Plan (TARP) as an initial risk consideration; there is value iteratively updating this as
understanding of the behaviour increases (for better or worse against the initial estimate).
Manual techniques and iterations of instrument readings “are based on the already produced macro
deformation, stress and water level of the dam” (Dong et al. 2017). The author disagrees with the
follow-up statement in Dong et al. (2017), which states that “[existent manual or automatic
61
intelligent monitoring methods] do not have the functions to evaluate the stability conditions, where
the future trend is predicted with the monitored information”. In-situ monitoring is an ongoing
update of the safe envelope of conditions within which the tailings dam is operating safely. It has
also been proven that through an understanding of the geotechnical behaviour of the structure, and
through this identifying initiation and progression of failure, the consequence of this failure can be
significantly reduced.
As such, the importance of understanding the trigger values at which different characteristics of the
dam are at risk, and in turn the appropriate response to mitigate the risk, is significant. Further, and
as a parallel theme to this research, the monitoring data collection and the trigger values should not
be observed in isolation. For example:
• Monitoring data collection: water quality sampling can acknowledge the presence of
tailings downstream, indicating seepage, and hence a path between the two sides of the
confinement structure; and
• Trigger values: A high pond water level may not trigger alarms in terms of geotechnical
stability, but when combined with the accelerated deformation of surrounding mountains, the
risk of failure is beyond both if considered individually where rather than being contained,
failure of the mountain could easily induce overtopping.
Both components are deemed critical for operators, consultants, contractors, and suppliers to
understand and embrace: it is entirely possible to understand the complete loop of tailings dam
behaviour through the appropriate risk assessment of the facility.
It is observed in literature such as Fell et al. (2015), Dong et al. (2017), and Hu and Liu (2011), that
the parameters being carried as primary stability indices for tailings dams, identified in this research
so as to retain the validity and application of previous works, are:
• Reservoir water level and rainfall (inclusive of length of dry beach);
• Pore pressures;
• Internal and external, horizontal and vertical deformation; and
• Seepage (including flow, temperature, chemistry, pH, and turbidity).
While these are measurable indices, it is in understanding their meaning and implications in terms
of tailings dam performance where the true value is realised. The stability indices can be correlated
with geotechnical failure modes presented in Fell et al. (2015), examples of which are provided in
Table 10.
62
Table 10 Measurable stability indices correlated against geotechnical failure modes for data aggregation potential
Drivers of Instability
Geotechnical Reservoir water level and rainfall Pore Pressures Internal and external, Seepage Other
Failure Mode horizontal and vertical
deformation
Foundation Associated with increased pore ‘Blow-out’ or ‘heave’ in Deformation indicative of Irregular seepage in the
failure pressures. dam foundations where foundation failure can be foundation under and
seepage forces create a identified at and just beyond downstream of the dam can
zero effective the toe of the embankment by be identified by
stress condition is a survey markers. piezometers.
situation which should be
readily detected by
carefully
positioned and well
monitored piezometers.
Internal Strong evidence to show that failures or Not common to have Initiation of piping is most Concentrated seepage in Thermal measurements
erosion and incidents occur at or above historic high sufficient change in pore common in cracks, high embankment or foundation, can track water flow and
piping reservoir levels (Fell et al. 2015) pressure to identify a permeability zones, or following preferential paths concentrations.
pipe before it is hydraulic fractures (Fell et al. of least resistance.
developed. 2015). Vertical (upward) gradients
Settlement can indicate in potential for heave,
saturation, loss of particle uplift, or blowout, and
mass, however is not horizontal gradients
conclusive. through an embankment
and/or foundation.
Overtopping Excess rainfall on top of an already high N/A Settlement of the dam crest N/A
water level can induce overtopping. can cause overtopping.
63
Uncontrolled reservoir water level can Landslide of surrounding
cause overtopping. terrain into dam can displace
retained tailings/water.
Seepage Critical reservoir level modelled for Increased flow rate could Deformation could be induced N/A. Any inexplicable increase
stability analysis defines allowable limit indicate a higher through seepage carrying in suspended solids, or
of phreatic surface and associated pressure. particles away from their discolouration of the
seepage flow. Abnormal seepage and original location. water, particularly during
wet spots on the first filling, needs to be
embankment face could treated with urgency as it
indicate seepage. may indicate piping
through the embankment.
Seismicity High degrees of saturation of certain Pore pressure increase in Horizontal earthquake loading As per reservoir water
embankment or tailings material could active faults (as a result could increase susceptibility to level.
heighten the susceptibility to of seepage, monsoons, or failure of previously displaced
liquefaction. induced). Fluid injection areas, where the structural
can increase pore integrity is not as-designed or
pressure and cause as-constructed.
induced seismicity.
Slope The reservoir water level drives a higher High pore pressures may Increased rates of deformation Slope instability could be
instability phreatic surface, in addition to be a precursor to slope or settlement in comparison to induced through excess
increasing the pressure on the instability, particularly stable operating conditions. material being carried
embankment. where there are high A limitation exists in strain through seepage,
At historic low reservoir water levels, foundation pore weakening soils, where failure undercutting or degrading
dams with marginal upstream slope pressures. may occur prior to observable the structural integrity of
stability or subject to internal deformation. the slope.
deformations experience abnormal
settlements and deformations.
64
A number of other items to monitor that are not captured in current literature (see Fell et al. 2015),
include, but are not limited to:
• Inspection by trained observers;
• Radar, satellite, and other monitoring techniques for ground movement measurements; and
• Seismic ground motion.
The critical consideration that must be carried when advancing research into the age of Internet of
Things, Cloud Computing, Machine Learning, and other technological advancements, is ensuring
that the content is readily applicable by site staff. By empowering all different levels of site staff
through the ease of use, manipulation, and understanding of solutions, it is anticipated that a dual
benefit can be found in leveraging the advanced site-specific knowledge of these personnel to
improve the model.
“The authors also believe strongly that the data from the instrumentation and the inspection records
must be reviewed by an experienced dam engineer as soon as they are received from the field
inspection team. The whole exercise becomes almost pointless if nobody with the necessary
knowledge looks at this field data” (Fell et al. 2015).
66
safety. Significant comments on this point include:
a. Previous failures have been in compliance with regulations/standards of practice;
b. There is a lack of understanding on brittle undrained failure of low plasticity tailings;
c. Significant variation exists between states, with loose regulation of some existing
facilities; and
d. Worldwide knowledge and fundamental geotechnical understanding should be
incorporated into both guidelines and valuable regulation.
• There was noteworthy interest in collaboration to improve regional and global standard of
tailings dam monitoring: 12 respondents were interested in more case study/knowledge
sharing; 12 respondents for a collaborative effort toward a single, best practice reference
document; and six respondents for a global alignment of standards;
• 67% of respondents do not believe that the frequency of reading mandated through the acts,
regulations, and/or guidelines is sufficient to capture the initiation and progression of
different failure types. Identified areas for improvement on this point include, but are not
limited to:
a. Rigid, prescriptive requirements do not have the flexibility to adequately address the
unique conditions at each site;
b. The frequency of reading must be site specific, not regulation driven. However,
benchmark standards were separately proposed as a potential value-add;
c. Ongoing monitoring and operator refresher/audit to avoid complacency; and
d. The frequency of readings alone is not enough, there needs to be an understanding of the
response of each instrument and integration between them.
Events
• 65% of respondents do not have a secondary structure downstream to control potential runout
in case of a tailings dam breach.
• 28% of respondents acknowledged that tailings dam failures or incidents that have occurred
onsite may have been avoided with different monitoring setups;
• The three primary controls, in hindsight, to avoid tailings dam failures or incidents are:
greater diligence in establishing trigger levels; greater adherence to procedures (QA/QC);
and real-time, telemetry; and
• While 6 respondents acknowledged that no failures or incidents have been identified through
the instrumentation installed, 10 respondents reported that numbers of events in the 1’s to
10’s had been identified.
Open comments on the survey captured the need for professionals to share their experiences. A
67
need was expressed to believe and use new technologies so that they can be improved every day.
Communications and transparency between all stakeholders was deemed critical. A comment was
provided in the industry survey that the author, in consideration of the works presented in this
paper, deemed important to express:
‘In trying to find a strategy one could use in TSF monitoring/surveillance, mandating (whether
through regulation or guidelines) actions like instrument reading frequencies or trigger levels or
providing "checklists" should not be priorities over encouragement of critical thinking and
fundamental understanding of soil mechanics (and physics/engineering). One of the great dangers,
in my opinion, of "checklists" and mandatory requirements is that they often suppress the need to
think critically or gain fundamental understanding, which are the greatest tools we actually have in
creating and maintaining safe structures.’
The author wholeheartedly agrees with this statement. In the age of advancing technologies and
improved capabilities, there is undeniable significance in capturing the fundamentals from the
initial stage and ensuring that these are embedded throughout all processes. There is an opportunity
to capture and leverage the knowledge and experience of many people within industry, collaborate
with the technological age, and drive forward safer systems and processes. If technology is allowed
to advance beyond the parallel time required for adequate diligence in assessing the performance of
systems, the industry could be less safe through complacency and unwarranted reliance.
“The strongest argument of the detractors of mining is that the fields are devastated by mining
operations…further, when the ores are washed, the water used poisons the brooks and streams, and
either destroys the fish or drives them away…thus it is said, it is clear to all that there is greater
detriment from mining than the values of the metals which the mining produces”
Agricola – 1556
68
With increasing publicity given to each successive event, the influence that the public has on project
design decisions has never been higher, as noted in Davies, Martin & Lighthall (2002). The trend of
increasing publicity also reflects an increasing demand on accountability through legal
ramifications for designers, operators, and company executives. Tailings dams do not generate
revenue in the mining process; tailings dams contain contaminated waste and toxins, and at this
stage remain a technically challenging material for reuse, recycling, or reduction. Regardless, there
is a key requirement to understand tailings dams and the associated risks, and equally importantly
be able to convey this understanding to stakeholders including employees, the public, and
regulatory departments. The value of quality communication and an ability to convey technical
information to non-technical or unfamiliar parties in an understandable manner is invaluable.
The stakeholder’s perception of the risk associated with any project is influenced by their values,
needs, assumptions, and concerns. “The more hazardous a stakeholder perceives the risk to be, the
greater the communication and consultation challenge your organisation will face” (Engineers
Australia 2018). Transparency from the organisation throughout the entire project encourages trust,
respect, credibility, and understanding of the risks that exist, and how these are being managed.
From the engineer’s perspective, engaging with stakeholders drives engineering perspective beyond
objective risk management to understand complex attitudes and perceptions, signifying the full
reach of accountability that the engineer has on the project. Mutual understanding increases
stakeholder confidence, and paves the way for progression and advancement.
The Joint Institution Group on Safety Risk (JIGSR 2012) detail five characteristics required to
achieve trust in the engineering domain:
• Competence (those communicating know what they are talking about);
• Objectivity (a view that the source of information is independent);
• Consistency (track record in dealing competently with similar matters);
• Openness (a willingness to disclose information and not to appear secretive); and
• Empathy (willingness to accept the validity of concerns and to listen and consult).
Difficulties in stakeholder engagement arise where these five characteristics are contradicted, the
benefit of the accepting the risk is not perceived to be reasonable, a perception of production over
safety is established, or past occurrences of unpredicted or underestimated consequences in similar
situations, as seen in (JIGSR 2012). Accepting responsibility for the risks of any project is critical;
stakeholder knowledge, influence, and significance in project planning and execution is greater than
ever, and only due to increase. Safety, monitoring, responsibility, and risk never cease to exist.
69
3.8. Acceptable Risk
“Leading tailings management practice recognises potential design limitations and uncertainties by
applying a risk-based management approach throughout the life of the facility: from project
conception, through design, construction, operation, and closure” (ANCOLD 2012).
Effective management of the risks associated with tailings dams during and after their operating
lives should be aligned with the hierarchy of controls. The hierarchy of controls is a universally
accepted standard for engineering design, construction, and risk management, and is prioritised as:
1. Eliminate;
2. Substitute;
3. Isolate;
4. Engineering Controls;
5. Administrative Controls; and
6. Personal Protective Equipment.
Tailings management plans should include iterative monitoring, updating, and review of procedures
“for the purposes of ensuring that controls are effective and efficient in both design and operations”
(ANCOLD 2012). Proposed methodologies for this include:
• Obtaining further information to improve risk assessment;
• Analysing and learning lessons from events (including near misses), changes, trends,
successes, and failures;
• Detecting changes in the external and internal context, including changes to risk criteria and
the risk itself which can require revision of the risk treatments and priorities; and
• Identifying emerging risks.
Risk can only be tolerated once the complexity has been addressed through the appropriate type of
assessment, “leading indicators of potential failures, either of individual, or in combination where a
number of individual issues could combine to result in a failure” (ANCOLD 2012) are clearly
identified, and appropriate risk management techniques are employed. It is recommended that
considerations for risk reduction to As Low as Reasonably Practicable (ALARP) are utilised in
order to establish and agree on an acceptable balance between implementation (money, time,
effort), the benefit gained, and most significantly, the residual risk after controls are employed.
3.8.1. Consequence Category
As part of tailings dam design, it is generally accepted that the quantified risk tolerability of
different design and management requirements of different dams is dependent on the consequence
category of the dam. In line with ANCOLD (2012) guidelines, there are two consequence categories
70
that need to be assessed as part of design: the Dam Failure Consequence Category and the
Environmental Spill Consequence Category.
The Dam Failure Consequence Category considers “the potential failure modes of the facility and
the resulting consequence to the business, the social and natural environment and the potential for
loss of life as described in Guideline on the Consequences of Dam Failure [Table 11]” (ANCOLD
2012). This consequence category should be iteratively updated at each phase of the dam life. A
"dam-break" analysis to determine the inundation area, the depth, and the velocity of potential flows
is also required. This analysis is “evaluated and ranked in accordance with the Population at Risk
(PAR), the nature of the receiving environment and the potential severity of impact in relation to the
nature of the released material” (ANCOLD 2012).
The Environmental Spill Consequence Category can be determined by “considering only the effect
of spilling of water from the dam during a flood event or extreme wet weather period” (ANCOLD
2012) in similar methodologies to that used to determine the Dam Failure Consequence Category,
and normally concentrates on the environmental impacts alone.
The author does, however, questions the current focus in industry on dam break assessments, which
significantly outweigh the focus on monitoring and instrumentation, and understanding/prevention
of the numerous modes of failure. While the author understands the significance in determining
consequence of failure and hence risk of the structure, it is not understood why greater focus is not
placed on preventing the dam from reaching this stage, let alone identifying and triggering
emergency procedures for stakeholders when failure is imminent.
3.8.1.1. Severity Levels Impacts Assessment
In the severity level impacts assessment matrix presented in ANCOLD (2012) and observed in
Table 11, the worst-case influence governs the resulting severity level. For example, should the
impact area exceed 20km2, the dam failure severity will be catastrophic, regardless of any additional
business, environment, or social considerations.
71
Table 11 Severity level impacts assessment - summary from ANCOLD Consequence Guidelines (ISSMGE 2004)
72
3.8.1.2. Population at Risk
The PAR is defined as “all those persons who would be directly exposed to the consequences of
failure of a structure or facility if they did not evacuate” (ISSMGE 2004). The severity level
impacts assessment is integrated with the PAR to determine the consequence category for the dam
(Table 12). In ANCOLD (2012), it is noted that guidelines have developed to recommend higher
design parameters for earthquake and flood than previously, considered appropriate in consideration
that “the majority of tailings dams fall into a high or extreme consequence category when
considering their operational risks and extended (post-closure) design life”.
Table 12 Recommended consequence category (ISSMGE 2004)
73
places; and
• After occurrence. Dependent on action taken and severity of the event.
Figure 24 Emergency management of dam breaks in time scale (Zhang et al. 2016)
In order to realise sufficient lead time prior to an event occurring, the systems in place to observe
and predict detrimental behaviour need to be appropriately designed and implemented. When
disseminating the warning, six factors that have an influence on the effectiveness of delivery as seen
in Zhang et al. (2016) are:
• Disaster characteristics (nature, severity, emergency, possibility of occurrence, and
influenced area);
• Government organisation (government response, organisational effectiveness, and
arrangement of labour and facilities);
• Warning methods (television, radio, internet, face-to-face notice, portable loudspeakers,
sirens, telephones, mail, emails);
• Area characteristics (area usage, population density, communication conditions, and traffic
conditions);
• Time and weather (time of a day, workday or holiday, rainy, foggy, or snowy days); and
• Population characteristics (age, gender, physical state, disaster experience, family members,
and societal relationship).
Decision making and scenario planning should be implemented in the risk assessment process, with
contributions from an appropriate cross-section of the workforce and stakeholders. Establishing a
detailed framework in the right environment eliminates the chance of human error caused by the
urgency and stress of emergency situations.
Without effectiveness of delivery and an appropriate evaluation of the risk, trust, faith, and respect
in the operator to make the right call is compromised. In turn, this has detrimental effects on future
74
risk management efforts in the area and damages relationships that may be critical to safe practice,
as discussed in Perception of Risk.
3.9. Conclusions
This paper establishes the current standard of tailings dam practice, identifies continuity alongside
inconsistencies in the global community and practice, and recommends paths forward by which the
global reach of competent, trained, and experienced practitioners can collaborate toward a safer
future for the tailings dam life cycle. The significance of learning from past and present practices
with the intent to implement and drive safer practices into the future is highlighted, recognising a
need for the advancing technological world to appreciate and leverage fundamental geotechnical
engineering knowledge and ensure that this remains in alignment as practices advance. A critical
flaw identified through this research was the lack of consistent, real-time, and online tailings dam
monitoring.
As a result of the recorded 283 tailings dam failures since 1915, there have been approximately
2300 deaths, > 187 m3 of tailings material release, and > 2000 km of tailings runout beyond the dam
wall. These consequences are unacceptable, yet the question of whether global practices are
improving in alignment must be raised; there have been 500 deaths since the year 2000,
demonstrating no clear reduction in the rate of failure consequences. Further, a critical finding
expressed within this paper is that there have been no unexplained tailings dam failures. The
significance of a Comprehensive Monitoring Strategy integrated with a fundamental understanding
of geotechnical practices and “What could go wrong?” is highlighted. All tailings dam failures to
date were entirely predictable, in hindsight; there is no reason that tailings dam failures should, at
this stage of human development, continue to occur.
This paper described the perspective of 25 global practitioners through an industry survey.
Respondents provided feedback from diverse backgrounds and regions on instrumentation,
guidelines, acts, and regulations, as well as tailings dam incidents and failures. Key conclusions
found from this survey include:
• There was a consistent interest in sharing knowledge about tailings dam monitoring
strategies.
• Only 21% integrate instrument readings through a technology platform, and this process was
identified as something practitioners want to learn more about;
• 67% of respondents do not believe that the frequency of reading mandated through the acts,
regulations, and/or guidelines is sufficient to capture the initiation and progression of
different failure types. Identified areas for improvement on this point include, but are not
limited to:
75
- Rigid, prescriptive requirements do not have the flexibility to adequately address the
unique conditions at each site;
- The frequency of reading must be site specific, not regulation driven. However,
benchmark standards were separately proposed as a potential value-add;
- Ongoing monitoring and operator refresher/audit to avoid complacency; and
- The frequency of readings alone is not enough, there needs to be an understanding of the
response of each instrument and integration between them.
Recommendations for directions forward are described within the paper, with overarching themes
of greater global collaboration in safety, as well as ensuring that fundamental geotechnical
engineering, physics, and science is firstly understood, then maintained and integrated within the
future of operational practices. There is a fear that this is being lost, and the industry is at an optimal
stage to grasp learnings and experience of the past to ensure that repeat errors are not made in the
future.
76
Dong, L, Shu, W, Sun, D, Li, X, & Zhang, L 2017, ‘Pre-alarm system based on real-time
monitoring and numerical simulation using internet of things and cloud computing for
tailings dam in mines’. IEEEAccess 2017, 5, pp. 21080–21089.
Hu, J & Liu, X 2011, ‘Design and implementation of tailings dam security monitoring system’,
Procedia Engineering 2011, 26, pp. 1914–1921.
ISSMGE 2004, Risk assessment – glossary of terms. ISSMGE TC32 – Technical Committee on
Risk Assessment and Management Glossary of Risk Assessment Terms, 1.
Joint Institution Group on Safety Risk [JIGSR] 2012, Risk communication and professional
engineers, The Institute of Engineering and Technology.
LePoudre, DC 2015, Examples, Statistics and Failure modes of tailings dams and consequences of
failure, PowerPoint slides, REMTECH.
Luino, F, & Van De Graff, J 2012, ‘The Stava mudflow of 19 July 1985 (Northern Italy): a disaster
that effective regulation might have prevented’. Natural Hazards and Earth System Sciences
2012, 12, pp. 1029–1044.
Ministry of Water Resources of the People’s Republic of China (MWR). Standard for Classification
of Risk Grade of Landslide Lakes. Standards in Water Industry of the People’s Republic of
China 2009, SL 450–2009, MWR, Beijing.
Prowse, L 2018, Risk Communication – Bringing the Community on the Journey, viewed 8 May
2018, <https://www.engineersaustralia.org.au/portal/event/risk-communication-bringing-
community-journey>
Robertson, A.M 2015, FMEA Risk Analysis: Failure Modes and Effects Analysis. PowerPoint
slides.
Scaioni, M & Wang, J. 2016, ‘Technologies for Dam Deformation Measurement: Recent Trends
and Future Challenges’, Proceedings of the Joint International Symposium on Deformation
Monitoring, Vienna University of Technology.
Tailings Dam Failures 1915-2016, viewed 11 May 2018, <http://www.csp2.org/tsf-failures-1915-
2017>
United Nations Environment Programme (UNEP), Division of Technology, Industry and
Economics 2001, APELL for Mining – Guidance for the Mining Industry in Raising
Awareness and Preparedness for Emergencies at Local Level. Technical Report no. 41,
UNEP, France.
Zhang, L, Peng, M, Chang, D, & Xu, Y 2016, Dam Failure Mechanisms and Risk Assessment, 1st
edn, John Wiley & Sons, Singapore.
77
Chapter 4 Real-time monitoring of tailings dams
Luke Clarkson1, David Williams1, & Jaakko Seppälä2
1
Geotechnical Engineering Centre within the School of Civil Engineering, The University of
Queensland, Brisbane, QLD 4072 Australia
2
EHP Environment Ltd, Oulu, Finland
This paper has been published in Georisk: Assessment and Management of Risk for Engineered
Systems and Geohazards.
4.1. Abstract
Real-time monitoring can improve the performance assessment of tailings dams by reducing the
laborious component of data collection while streamlining the analysis process. When planning and
installing instrumentation, the challenge exists where if nothing goes wrong, the question is asked
whether too much instrumentation is installed, where on the other hand if a failure occurs, the
question is asked as to why more investment was not made to prevent it from occurring. This paper
identifies the monitoring system requirements, assesses the cost of historical tailings dam failures
(Mt Polley and Fundão), assesses the cost of a real-time monitoring system to suit the
instrumentation that was in place at time of failure of those dams (including standpipe and vibrating
wire piezometers, digital inclinometers, and flow meters), and suggests ways to extract more value
from individual instruments to full monitoring system integration.
Keywords: remote sensing, real-time monitoring, tailings dam, risk, geotechnical, mining
4.2. Introduction
Tailings dam operation is becoming an increasingly scrutinised practice as the mining industry
progresses toward a more sustainable future. Failure of these structures can present in many
different forms, and can exhibit significant consequence to the social, environmental, and economic
risks that are linked to its existence.
An assessment of the current standing of the industry in terms of both practice and regulation has
been undertaken by the author (see Chapter 3), with a critical flaw of practice identified as the lack
of consistent, real-time, and online tailings dam monitoring. This presents a major risk in the
understanding of tailings dam performance where the low frequency of manual data collection and
interpretation, is far slower the time taken for dam failure to progress; with respect to identifying
failure, these readings could be considered as redundant measures.
The ability for instrumentation and monitoring to foresee deterioration of slopes is becoming more
78
available with improving technologies. Further, there has not been a single tailings dam failure in
history that, in hindsight, was not proven to have been predictable with the information available
(Davies, Martin, & Lighthall 2002). The severe consequences and risk of tailings dam failure
globally demands improvement in how practitioners understand and monitor the performance of the
dam without complacency and putting people’s lives and the environment at risk. With
opportunities to predict tailings dam failure existing alongside a continued tailings dam failure rate,
the industry has an immediate responsibility to advance their understanding and mitigate the risk of
failure.
As technology is now more readily available, it is important to understand how to effectively and
efficiently apply this in the mining environment. Tailings dams do not contribute to the revenue of
the operation yet can severely deplete this revenue should there be a failure. An evaluation of
effective costs can be used to justify different levels of monitoring programs, identify areas of
optimisation in the instrumentation use (such as overlapping with environmental monitoring), whilst
always appropriately understanding the key geotechnical risk areas.
Total Pressure Cells (vibrating wire, pneumatic, other) 3/month 1/year Yes
80
Observation Wells 1/week 1/6 months Yes
No (survey)
(instruments)
Note: 1/day = one or a set of measurements per day; 1/2 weeks = one or a set of measurements
every 2 weeks.
4.3.1. Instrumentation System
A modern real-time slope instrumentation system generally comprises the following four key
components:
1. Sensor nodes which consist of instruments and a datalogger;
2. Communication network for sensor nodes;
3. Base station with an external communication modem; and
4. Information system for storing and displaying the data.
There are other elements that are anticipated to be useful for application in a tailings dam
application, such as repeaters and aggregators. Repeaters are a form of intermediate ‘connecting’
sensor node not directly connected to an instrument, used for such purposes as avoiding physical
obstacles to the data transmission onsite, seen in Tang and Cheung (2011), and aggregators are a
tool used to “reduce the number of transmissions of sensor nodes, and hence minimising the overall
power consumption” (Al-Karaki, Ul-Mustafa, Kamal 2004). While relevant, the detail of these
components is not discussed in this paper as they contribute more to the logistics of data collection
81
and transmission process onsite, than to the monitoring approach.
4.3.1.1. Sensor Nodes
The items listed in Table 6 define instruments that can be employed, as well as their suitability for
integration into the real-time system. It is noted that “most of the instruments measure the changes
in electrical properties… which are then converted to give ground displacement, distortion, or
groundwater pressure, as appropriate. The changes in these electric properties are usually brought
about by variations in magnetic field, natural frequency, electric resistance or conductivity due to
displacement or elongation of the embedded elements of the instrument” (Tang and Cheung 2011).
It is this electrical property (whether transmitted by analog or digital) that allows the instrument’s
signal to be collected by a datalogger. Together, the instrument and datalogger are connected by
means of a cable in order to make a sensor node. For instruments such as inclinometers or
extensometers, a multiplexing system is used to connect the datalogger to different anchor depths,
detailed by Tang and Cheung (2011).
There are a number of different types of dataloggers, distinguished primarily by their data
transmission method, input format, and power source. Dataloggers can be provided with or without
data transmission, which refers to the WiFi, General Packet Radio Service (GPRS, mobile data,
2G/3G/4G etc.), Long Range (LoRa) technologies, that act as the medium by which information is
transmitted. For tailings dams, cell networks are preferred because they are both cost efficient and
reliable for the purpose. Where cell network is not available (remote mine sites), the practical
alternative is satellite communications, followed by LoRa. However, with the rise of autonomous
and remotely operated vehicles on mine sites, the availability of communication technologies is
increasing with varied opinion at time of this paper as to the preferred method forward.
Dataloggers can either be single sensor input, or multi-channel input. Further, inputs can either be
analog, or digital. Traditionally, instruments would have been set up as analog. For example,
inclinometer tubes would have up to 30 cables to different downhole increments in order to collect
data. As technology advances, the digital solution instead sees a single cable returning the data from
all increments. This makes it easier to handle multiple sensors when the data is digital, hence
gaining more value from both the sensor and datalogger. A digital sensor can record two different
types of reading (e.g. pH and temperature, or conductivity and temperature), or multiple readings of
the same type (e.g. inclinometer).
The power source can vary and is often dependent on the locality of the datalogger, climate/weather
conditions of the site, and redundancy of the instrumentation system (should the power source fail).
Sources include battery, solar, and mains powered dataloggers.
It is anticipated that the majority of instruments can be retrofitted to an online system. This is
dependent on the type of data, type of sensor, which will in turn inform the datalogger that is used.
82
Standard 4-20mA analog sensors can be adapted, and it would be expected that all digital sensors
can be adapted.
4.3.1.2. Sensor Networks
For tailings dams, it should be expected that a number of sensor nodes are installed at different
locations. In order to streamline the collection of data from the different sensor nodes, a sensor
network is set up. This network may link by cabled and wireless connections such as Wi-Fi,
Bluetooth, or ZigBee. A comparison between different connections is presented in Table 14. The
variety of sensors present in a tailings dam system will typically introduce a heterogeneous sensor
network. It is noted that in the context of a tailings dam and operational practicality, the “optimal
operational range” and “battery life” are deemed significant considerations in ensuring a reliable
and connected system.
Table 14 Comparison of common IEEE wireless protocols, from Tang & Cheung (2011)
83
Table 15 Network topology types (adapted from Tang & Cheung 2011, and Mcgrath & Scanaill 2013)
88
Table 17 Major tailings dam failure influence
89
>220,000 m3 of waste water
release from gypsum waste
pond (Yle 2012)
“Nickel and zinc
Community/activist petitions/protests
Talvivaara 94% drop in the following year. concentrations in nearby Snow
against mine reopening, challenging for
Talvivaara 2012 Finland Mining -10.0 Contributing factor to eventual River exceeded the values that
more stringent conditions on mining
Company Plc bankruptcy. are harmful to organisms
permit.
tenfold or even a hundredfold,
uranium concentrations more
than tenfold” (WISE Uranium
Project 2019)
“Several towns flooded, 10 people
killed, approx. 120 people injured, 8
square kilometres flooded” (WISE
Uranium Project 2014).
“MAL Hungarian Aluminium
Private Company.
Production and Trade Company ‘would
MAL Failure occurred on October 4, 2010. The company’s 700,000 m3 of caustic red mud
be prepared to pay 1.5 billion forint
Kolontár 2010 Hungary Hungarian management was taken over by the state on October released (WISE Uranium
($7.6 billion [AUD; $5.5 billion USD])
Aluminium 12, 2010 (WISE 2014) Project 2014).
over five years by way of
In 2013, the company went under liquidation.
compensation for the victims of the
toxic sludge catastrophe two weeks
ago,’ the company's chief, Lajos
Tolnay, said” (WISE Uranium Project
2014).
“Eleven homes were carried away by
the mudflow; at least one person was
Karamken killed” (WISE Uranium Project 2019) “> 1 million m3 of water,
Karamken
Minerals Soviet-era, abandoned mine. Municipal government “Authorities have allocated some 7 150,000 m3 of tailings, and
Gold 2009 Russia
Processing responsible for maintenance. million rubles ($110,000 USD) to 55,000 m3 of dam materials”
Tailings
Plant provide accommodation and (WISE Uranium Project 2019)
compensation for people affected”
(Robinson 2009).
“The wave of ash and mud toppled “4.1M m3 of ashy slurry
power lines, covered Swan Pond Road [released]… The ash slide
Kingston Tennessee
+0.75 and ruptured a gas line. It damaged 12 covered [1.6 square
Fossil 2008 USA Valley +0.3 0.75
USD gain1 homes, and one person had to be kilometres] as deep as [1.83
Plant Authority
rescued, though no one was seriously metres]” (WISE Uranium
hurt” (WISE Uranium Project 2019) Project 2019)
1
Negligible effect observed - price appeared to follow the existing trend
90
Occurrence of tailings dam failures have also been linked, from Davies and Martin (2009), to
the cyclicity of the global copper and gold prices. It is stated that there is a relationship
between the peak in commodity prices and the occurrence of failures, with an anecdotal offset
of approximately two years between the two. As Davies and Martin (2009) describe, this
relationship is ascribed to:
1. Peaking prices drives peaking production, rushing design and construction and
compromising quality/safety standards;
2. Attractive employment market entices employee turnover;
3. Development of resources in high risk areas is seen as viable;
4. Post-boom normalisation of operations induces pressures on cost cutting;
5. There is a use of inappropriate designs adopted from “similar” situations; and
6. Independent review is challenged, suspected to be to avoid the associated time delays
and cost.
With an underlying, dependable monitoring system describing the current performance and
historical reliability of the tailings dam structure, the influence of external factors can be
mitigated. Modelling and design can be readily validated using real data, and performance-
based monitoring and response is readily available.
It is the funding allocated to identify and mitigate risks in the tailings facility that presents
challenges; often, justification for expenditure on these structures that do not generate a profit
for the organisation is challenged, delayed, and reduced. The industry need not segregate
safety from finances, entirely; however, investment is required in ensuring that the systems in
place are functional, practical, and reporting on the true performance of the dam structure.
Hence, research and development into optimising systems, increasing understanding,
and empowering integrated site teams to be intrinsically linked to the performance of
their structure is essential.
4.4.2. Cost of Instrumentation
New technologies and systems are making real-time, online monitoring systems economically
viable. A collaborator in this research, EHP Environment Ltd., provides modern solutions for
real-time online monitoring. An overlapping opportunity was identified with a number of their
solutions, such as pore pressure monitoring, water flow, quality, and level, as well as the
directly related inclinometer for both horizontal and vertical measurements.
For a single Ground Monitoring Station, which comprises a data logger with a mobile (4G)
modem, and solar power system costs approximately $5,000 AUD (€3,000 EUR, $3,400 USD
91
at time of this research). 1-4 sensors (piezometers/inclinometers) are typically connected to
this datalogger.
An estimated cost to fully monitor the instruments described in the Fundão and Mt Polley
investigation reports is presented in Table 18. An approximate, typical cost for instruments
and dataloggers is presented as below. This is provided based on typical instrument cost in the
UK and Europe:
• Standpipe Piezometers, $1,000 - $1,150 AUD (€600 – €700 Euro);
• Vibrating Wire Piezometers, $1150 - $1350 AUD (€700 – €800 Euro, including
datalogger but not including portable readout unit);
• Digital Inclinometers, $2550 - $10,050 AUD (€300 per metre, plus assembly); and
• Flow Meters, $500 - $650 AUD (€300 – €400, excl. ultrasonic).
It is important to note that this exercise is not undertaken to suggest that a real-time
monitoring program would have changed historical events, but rather to utilise the data and
information available from existing mines and case examples. It is suggested that the
usefulness of real-time data in predicting events is not yet known, however without exploring
the option it shall remain as an unknown.
Data for the quantity of instrumentation is sourced for Fundão Dam and Mt Polley from
Appendix E of Cleary Gottlieb Steen and Hamilton LLP et al. (2016) and Appendix F of
Province of British Columbia (2019), respectively.
Table 18 Estimated cost of instrumentation for different dam structures
# of Piezometers
Tailings Dam
Total # of
# of Flow
92
Table 19 Estimated costs for implementation of a real-time instrumentation system at example dams
4.5. Implementation
In both literature and practice regarding tailings dam monitoring, it is clear that there is no
“one size fits all” solution for dam instrumentation quantity and location. The objectives of
monitoring, however, are consistent throughout (and described further in ANCOLD 1976 and
Fell et al. 2015):
• To provide confirmation of design assumptions and prediction of performance during
the construction phase and initial filling of the reservoir;
• To provide during the operation phase of the life of the dam an early warning of the
development of unusual and potentially unsafe trends in behaviour;
• To provide data on behaviour of dams which may not conform with accepted modern
criteria and warrant continuous and close monitoring as a guide to the urgency for
introduction of remedial/stabilising works or other measures;
• During raising or remedial/stabilizing works, which may need to be carried out with the
storage full, close monitoring of structural/seepage behaviour is warranted to ensure
that the additional loading introduced by the new works is applied in a manner which
will not adversely affect the safety of the dam;
• To satisfy legal obligations of the duty of care; and
• To provide data to allow developments in dam engineering: through better
measurement of properties.
93
In order to achieve these objectives, a systematic approach should be followed when
implementing the monitoring system to ensure that every instrument’s purpose, tying
seamlessly into alert levels and action plans, is realised (from Marr 2014):
1. Identify what questions need answering;
2. Identify what measurements can and should be made;
3. Design appropriate monitoring system;
4. Plan installation, calibration, maintenance, and data management;
5. Prepare and update budget;
6. Procure, test, install and verify instruments;
7. Calibrate and maintain instruments and readouts;
8. Collect, process and evaluate data;
9. Interpret and report results quickly; and
10. Take action when required.
It is emphasised that monitoring is only useful if well planned, designed, installed, and used
appropriately.
Real-time monitoring’s predominant geotechnical benefit is from two perspectives. Looking
forward, the provision of real-time data presents an improved opportunity to identify
potentially unsafe trends in behaviour, early. This information can then inform a response to
repair or evacuate, in either case improving the opportunity to reduce the consequence of
deterioration or failure. Looking backward, the record of data which is centrally stored and
near void of human error in data entry/plot generation presents an improved opportunity for
back-analysis to understand the current performance of the dam in relation to the design, and
also represents a dataset of measurements that are within the performance limits of the dam:
the “safe and stable” envelope.
4.5.1. Establishing Baseline Triggers for Real-time Monitoring
For new mining projects or when an existing mine introduces a new monitoring
technology/technique, it is necessary to establish reliable targets to monitor performance
against. Importantly, “the monitoring program must be developed with a clear sense of
purpose, which will dictate how the accumulated data are interpreted and reported” (Hawley
and Cunning 2017). “Interpretation of data from monitoring systems typically involves
looking for changes from the expected norms” (Hawley and Cunning 2017). The challenge is
then presented of what the ‘expected norm’ appears as for each unique mine; what is
acceptable, and what is not?
94
Challenged by the lack of existing case studies at the mine, the approach is recommended to
be generalised and conservative until a sufficient set of data is gathered and can inform the
original estimates. There are three approaches that can be utilised to gain a better
understanding of the ‘expected norm’, prior to gathering any data:
• Design analysis – modelling.
• Design analysis – back-analysis of past collapses; and
• Generic advice and regionally documented TARP levels.
4.5.1.1. Design Analysis – Modelling
It is rare that a tailings dam will not have been appropriately planned and assessed using
geotechnical software to theoretically indicate the safety and stability of the structure, prior to
construction. Slope stability and seepage models are generated, with the structure modified in
both short and long term cases to achieve an acceptable Factor of Safety or Probability of
Failure.
Depending on the software used, static or dynamic analysis can also be undertaken in
undesirable scenarios to understand how the tailings dam will perform in extreme or adverse
circumstances. For example, seismic analyses can be undertaken by applying a horizontal
acceleration to the entire structure to ensure that the Factor of Safety remains above
(typically) 1.1. The magnitude of horizontal acceleration is based on the seismicity of the
region and applies in the short term, however, in sensitivity testing the seismic analysis: what
is the magnitude of horizontal acceleration that sees the Factor of Safety drop below drop
below 1.1, or even below equilibrium?
Should an anomalous event occur (anecdotally it is often the anomalous events, or a
combination of these events, that precedes failure), what magnitude of seismicity triggers
different actions? The modelling can inform this, in the first case.
Similarly for simulation of slope instability and phreatic surface rise/fall: modelling can
assess the rate of movement and the water level (including cases such as high water, gradual
drawdown, and rapid drawdown), to inform actions that should be taken when slope and
piezometer monitoring, respectively, reach these levels.
In theory, a number of trigger levels can then be established based on design analysis to
present a baseline estimate as presented in Figure 25.
95
Figure 25 Baseline estimate of trigger levels as indicated by design analysis
96
stand, however opportunity exists to refine the slope movement velocity trigger. Where visual
observations identify deterioration at a lower movement velocity, a more conservative yet
reasonable approach would be to lower the Level 1 trigger (which often prompts inspection/a
higher degree of monitoring in the area) to the level at which signs of deterioration were
observed. This is appropriate as a starting point until a better understanding of behaviour can
be ascertained.
97
foundation settlement, and water quality, the value of shared knowledge in the tailings
community is reiterated. The broader sample space of case studies available improves the
ability for the community to assess potential risks to their unique sites and learn from things
that have gone well/not so well in the past. The uniqueness of different sites is always
acknowledged, albeit the opportunity to educate based on the experience of others (and to
repay the favour) inevitably reduces the risk associated with experiencing an event onsite for
the first time and either being unprepared or missing the indicators of failure entirely.
The hypothetical is continued in Figure 27, where a similar slope was documented and
published to have experienced substantial deterioration at a slope movement velocity lower
than that indicated by design analysis (8mm/hr).
4.6. Conclusion
As with all technology and systems that see rapid development, the accessibility and cost of
the solution improves in parallel. The appropriate next step for tailings dam monitoring is to
minimise the need for laborious manual readouts of instrumentation, instead transitioning to
an automated, real-time solution. This works for the betterment of cost, time and resource
availability, which can instead be redirected to visual observations of the dam, interpretation
of pre-generated data representations, and updating/activation of risk management plans.
The technology is seen through the four primary components of the instrumentation system:
sensor nodes (instruments and datalogger), communication network between sensor nodes,
base station with an external communication modem, and an information system for storing
and displaying the data. For each component, there are multiple layers of solutions available
to tailor to site specific needs, including protocols and topology related to optimal operational
range, battery life, and system contingency, amongst others. It is not necessary for operators
to understand the complexities of the systems to meticulous degrees; it can be argued that
technology design focused on functionality, usability and user experience should be a priority
in a field where rapid decision making and safety critical response is required.
The value of this response in mitigating the consequence of failure was discussed. The
monetary cost of recent failures, quantified in terms of their stock price, social impact
(compensation), and environmental impact (remediation), were observed to be in the range of
$750M to $56B, or otherwise contributing to eventual bankruptcy of the responsible
organisation. For the case studies available, the instrumentation that was installed in the dam,
retrofitted with a hypothetical real-time monitoring system saw systems worth approximately
$400,000 to $500,000. When comparing the two costs, the real-time monitoring system
equated to 0.0008% to 0.0477% of the total cost of failure. While ignorant to assume that this
is a direct comparison, the ability for a real-time monitoring system to provide sufficient
warning to save lives, environmental damage, and reduce other consequential losses still by
far prevails on the cost-benefit scale.
When looking to implement a real-time monitoring system, it is necessary to establish reliable
targets to monitor performance against. Without previous data, this can be difficult, hence
100
three methods are proposed as a preliminary option until further data and hence understanding
can be gained. These include design analysis in the form of either modelling or back-analysis
of past failures/stable slopes, or making reference to generic advice and regionally
documented TARP levels.
Opportunities also exist to extract more value from individual instruments, to full monitoring
system integration. In relatively undocumented areas to date, but theorised to hold immense
potential, centralising monitoring data and investigating the value of cause-and-effect
instrumentation both take a holistic approach to the monitoring system. The two approaches
seek to gain multiple benefits from a single instrument, multiplied again when considering a
system of instruments, in turn improving the cost-benefit of the geotechnical monitoring
system.
By not only considering these elements, but also ensuring to document and share information
throughout the learning and development process, the community of tailings dam practitioners
can progress together toward a safer future.
101
<http://csp2.org/files/reports/Bowker%20%26%20Chambers%20-%20Risk-
Public%20Liability-
Economics%20of%20Tailings%20Storage%20Facility%20Failures%20%E2%80%93
%2023Jul15.pdf>
Calleja, NP 2013, Philex agrees to pay in full P1.034B fine for mine spill, viewed 2 June
2019, <https://business.inquirer.net/107843/philex-agrees-to-pay-in-full-p1-3-b-fine-
for-mine-spill>
Clarkson, L & Williams, D. 2019, ‘Critical Review of Tailings Dam Monitoring Best
Practice’. International Journal of Mining, Reclamation and Environment, vol. 34, iss.
2, pp. 119-148.
Cleary Gottlieb Steen & Hamilton LLP, Vale S.A., BHP Billiton Brasil Ltda. & Samarco
Mineração S.A. 2016, ‘The Fundão Tailings Dam Investigation’, Appendix E –
Samarco Field Monitoring Data.
Davies, M & Martin, T, 2009, ‘Mining market cycles and tailings dam incidents’.
Proceedings of the 13th International Conference on Tailings and Mine Waste, Banff,
Alberta.
Davies, M, Martin, D & Lighthall, P 2002a, 'Mine Tailings Dams - When Things go Wrong'
Proceedings of Tailings Dams 2000, Association of State Dam Safety Officials, U.S.
Committee on Large Dams, Las Vegas, Nevada: 261-273.
EHP Environment 2019, EHP-Dataservice, viewed 17 June 2019
<https://www.ehpenvironment.com/en/dataservice/>
Fell R, MacGregor P, Stapledon D, Bell G, Foster M 2015, Geotechnical Engineering of
Dams, 2nd ed., CRC Press/Balkema, The Netherlands.
GroundProbe 2019, Geoexplorer, viewed 17 June 2019
<https://www.groundprobe.com/product/geoexplorer/>
Hawley, M and Cunning, J 2017, Guidelines for Mine Waste Dump and Stockpile Design.
Australia, CSIRO Publishing.
Hui, S, Charlebois, L, & Sun, C 2017, ‘Real-time monitoring for structural health, public
safety, and risk management of mine tailings dams’. Canadian Journal of Earth
Sciences. 55(3):221-229.
Klohn Crippen Berger (KCB) 2018, Static Liquefaction and Strength Loss in Tailings Dams,
viewed 30 November 2019 <https://www.klohn.com/blog/static-liquefaction-strength-
loss-tailings-dams/>
102
Marr, WA 2014. “Geotechnical Instrumentation and Monitoring.” Presented at the Iowa
Section of ASCE and SEDC Geotechnical Conference, Iowa, April 10.
McGrath M.J., Scanaill C.N. 2013, ‘Sensor Network Topologies and Design Considerations’,
Sensor Technologies, Apress, Berkeley, CA
Province of British Columbia, 2019, Report on Mount Polley Tailings Storage Facility
Breach, Appendix F – Instrumentation and Monitoring.
Robinson, R. 2009, Karamken Dam Break Information, viewed 2 June 2019
<http://www.sric.org/enr/docs/2009-09-07_KaramkenDamBreak.pdf>
Roche, C, Thygesen, K, & Baker, E, 2017. Mine Tailings Storage: Safety is No Accident. A
Rapid Response Assessment., UN Environment, GRID-Arendal.
Samarco n.d., Samarco, viewed 18 May 2019 <https://www.samarco.com/en/a-samarco/>
Tang, CSC, & Cheung, SPY 2011. Review of Real-time Data Transmission Systems for
Slope Instrumentation. GEO Report No. 262. Geotechnical Engineering Office, The
Government of the Hong Kong Special Administrative Region.
The Catholic Bishops Conference of the Philippines – National Secretariat for Social Action,
Justice and Peace [CBCP-NASSA], Climate Change Congress of the Philippines
[CCCP], Philippine Misereor Partnership Inc. [PMPI] and it Northern Luzon Cluster,
Peace Foundation, Inc., Pambansang Kaisahan ng mga Magbubukid ng Pilipinas
[PKMP], Katribu Ingigenous Peoples’ Partylist, Cordillera Peopels Alliance [CPA],
Caritas Baguio, Community Volunteer Missioners [CVM] 2012, THE PHILEX MINE
TAILINGS SPILL OF 2012: AN INDEPENDENT FACT-FINDING MISSION
REPORT, viewed 1 June 2019
<https://file.ejatlas.org/docs/ATM_cFinal_FFM_Philex_Report_2Oct_2012FINAL_R
EPORT.pdf>
WISE Uranium Project 2019, Chronology of major tailings dam failures, viewed 2 June 2019
<http://www.wise-uranium.org/mdaf.html>
WISE Uranium Project 2014, The Kolontár red mud dam failure (Hungary), viewed 2 June
2019 <http://www.wise-uranium.org/mdafko.html>
WISE Uranium Project 2018, The Mount Polley tailings dam failure (Canada), viewed 1 June
2019 <http://www.wise-
uranium.org/mdafmp.html?_sm_au_=iVVZT1n0kS71QHTH>
103
Yle 2012, Talvivaara waste water leaks into environment, viewed 1 June 2019
<https://yle.fi/uutiset/osasto/news/talvivaara_waste_water_leaks_into_environment/63
63459>
104
Chapter 5 Catalogued instrumentation, monitoring, and systems
To enable greater accessibility to the instrumentation and systems available for tailings dam
monitoring, this research provides a single point of reference for the different available types.
The advantages, disadvantages, and specifications of each are described in two primary
sections:
• Real-time instrumentation and monitoring techniques, describing the instruments
suitable for installation within, or situated remote to the tailings dams. With a particular
focus on techniques capable of real-time monitoring, the utilisation of online
monitoring systems is also explored; and
• Example instrumentation and monitoring systems, describing the components involved
in instrumentation and monitoring systems for tailings dams, translating traditionally
electrical and systems engineering terminology into a reference base suitable for the
broad range of tailings dam practitioners.
The summary catalogues describing the techniques and systems are presented in Appendix B
and Appendix C, respectively. A snapshot and insight into the trends, patterns, and common
considerations identified in the current standing of real-time tailings dam monitoring practice
is described in Appendix D.
5.1.1. Abstract
With ongoing catastrophic mine tailings dam failures, the hindsight revelation of poor safety
records, and an increasing prevalence of public scrutiny and attention of mining operations,
there is an immediate call for enhanced safety provisions of tailings dams. Today, challenges
arise in identifying and utilising the ability of monitoring systems to understand the complex
performance and rapid behaviours of these dams, in turn reflecting on the system’s ability to
be able to predict deterioration before failure occurs. New literature, mining regulators,
105
insurance firms, and mining practitioners are calling for increased diligence in the form of
real-time monitoring: but what can the industry offer in response? This research establishes a
centralisation of real-time monitoring instruments suitable for tailings dams, discussing the
specifications, advantages, and disadvantages of each. An understanding of traditional
instrumentation was detailed, progressing to online monitoring systems and the value of
centralised monitoring. Collaboration with suppliers discovered advanced and innovative
systems which enable monitoring of different failure modes and mechanisms.
108
• Increased time between readings, potentially missing the development of deterioration
trends, or falling out of compliance based on a single, missed reading;
• Increased labour demand to read and process the measurements;
• Possibility for human error at the multiple touch-points in the process, including
reading the instrument, documenting the reading, processing the reading, or during
other data handling activities;
• Remote monitoring is limited to traditional survey, which either means that additional
intrusive instrumentation needs to be installed, increased survey frequency is required,
and the extent of surface monitored is limited to discrete points (with potential
interpolation between these points);
• Frequent installation of instrumentation to understand different areas can increase
costs;
• Less opportunity to identify when instrumentation is faulty or out of calibration. This
may take several readings (at a reduced relative frequency) to identify the error;
• Training of a higher number of field personnel to check, understand, and read
instrumentation. Also trusting the professional interest of personnel undertaking the
reading for reliable and accurate measurement; and
• May be impractical for remote sites where personnel do not frequent.
Noting these shortfalls, there are a number of traditional instruments that are not suited to be
upgraded to real-time monitoring. This is typically as a result of their installed configuration
or the method by which the measurement is undertaken. Such examples include:
• Water level gauge, limited by its nature as a static measurement attached to
infrastructure;
• Observation well, limited by its nature as an open hole unless combined with a
piezometer;
• Traditional survey, limited by the requirement to have an operator undertaking and
manoeuvring the survey;
• Catch containers, limited by its nature as a coarse water collection and measurement
tool;
• Magnetic extensometers, based on their measurement method relying on the sound of a
buzzer on a tape reel prompting the taking of a measurement; and
• Thermotic/self-potential geophysical surveys, limited by the requirement to have an
operator undertaking and manoeuvring the survey.
109
It is noted that in planning and establishing a monitoring system, the objective should remain
as using the best tools for the intended purpose. If trained personnel readily frequent an area
of interest, there is an inherent practicality on relying on visual observation, validated by
instrumentation (potentially manually read, tailored to the risk of failure), as opposed to full
reliance on a dense grid of automated instrumentation. Hence, it is suggested that a
combination of data collection methods can be used, yet emphasised that the availability of
newer (and increasingly more affordable) technologies can supersede many of the traditional
instrumentation and monitoring techniques.
5.1.5. Real-time Instrumentation
A detailed insight into the instruments that currently have real-time capability are summarised
in Appendix B. The key performance metrics addressed by this instrumentation are extracted
from Clarkson et al. (2019) as:
• Phreatic Surface and Water Level;
• Pore Water Pressure;
• Seepage Flow;
• Deformation and Movement (surface and internal);
• Seismicity;
• Earth Pressures; and
• Climate.
These performance metrics are addressed due to their role in identifying and understanding
the most prevalent modes of failure of tailings dams. These failure modes are described in the
following sections, alongside a generalised rating of the performance metrics’ usefulness in
identifying the different modes.
It is important to note that the focus on automated instrumentation does not negate the
importance of retaining the interpretation of collected data: the responsibility of which should
remain with the engineer. The real-time automation of instrumentation is encouraged in the
collection and processing of raw data to provide useful information, ready for interpretation.
5.1.5.1. Phreatic Surface/Water Level
The pond water level of a tailings storage facility typically has a significant influence on the
behaviour and trend of the phreatic surface. By measuring and understanding the structure’s
water balance, monitoring the pond water level as a recharge point, and the trend of the
phreatic surface through the structure as a driven by the pond water, saturated tailings
deposition, and environmental conditions, two loading scenarios can also be better
110
understood: static/ normal operation (poor water management) and hydrologic (flooding,
unpredicted reservoir water levels).
5.1.5.1.1. Applicability
A generalised rating of the usefulness of direct monitoring of the phreatic surface for
identification of different modes of failure is presented in Table 20.
Table 20 Rating of usefulness of phreatic surface monitoring for identification of different modes of failure
Internal
Failure Foundation Slope
Erosion Overtopping Seepage Seismicity
Mode Failure Instability
and Piping
111
5.1.5.2. Pore Water Pressure
Pore water pressure describes the pressure of water in voids between soil particles, or in
discontinuities in rock. The phreatic surface generally defines zero (atmospheric) pore
pressure, with the hydrostatic water pressure increasing linearly with depth below this.
However, soil suction in fine grained soils can also cause capillary rise and negative pore
pressures above the water table can be present. A change in pressure can cause an imbalance
in the driving and supporting forces of a structure, resulting in destabilisation of varying
scales (from particle transport to slope slumping).
5.1.5.2.1. Applicability
A generalised rating of the usefulness of direct monitoring of pore pressure for identification
of different modes of failure is presented in Table 21.
Table 21 Rating of usefulness of pore pressure monitoring for identification of different modes of failure
Internal
Failure Foundation Slope
Erosion Overtopping Seepage Seismicity
Mode Failure Instability
and Piping
112
5.1.5.2.2. Instrumentation
In order to understand the behaviour of the pore pressure, a number of instruments are
available with real-time measurement capability. These are described on Page 2 and Page 3 of
Appendix B.
5.1.5.3. Seepage Flow
“Seepage data is one of the best indicators of dam performance” (Fell et al. 2015). Seepage
processes describe the flow of water through the embankment, potentially giving rise to
instability through piping (material transport), slope instability and foundation heaving
(increased pore pressure), or excess water losses (environmental damage). While the drivers
are inherently linked to the behaviour of the phreatic surface and pore water pressures,
additional approaches related to measurement of the flow of seepage can contribute to an
understanding of the failure mechanism.
5.1.5.3.1. Applicability
A generalised rating of the usefulness of direct monitoring of seepage flow for identification
of different modes of failure is presented in Table 22.
Table 22 Rating of usefulness of seepage flow monitoring for identification of different modes of failure
Internal
Foundation Slope
Failure Mode Erosion Overtopping Seepage Seismicity
Failure Instability
and Piping
113
Supporting Indicators
• Seismicity: Increased seepage flow could heighten saturation of embankment, which in
turn could heighten susceptibility to liquefaction.
5.1.5.3.2. Instrumentation
In order to understand the behaviour of the seepage flow, a number of instruments are
available with real-time measurement capability. These are described on Page 4 of Appendix
B.
5.1.5.4. Deformation and Movement
Deformation and movement can be of varying scale and behaviour. Each observation deserves
assessment and investigation into the root cause, considering that movement is not typically
within the plan and design of the structure. Deformation can either be on the surface, or
internal to the embankment, with behaviour typically any of vertical, horizontal, rotational, or
translational.
5.1.5.4.1. Applicability
A generalised rating of the usefulness of direct monitoring of deformation and movement for
identification of different modes of failure is presented in Table 23.
Table 23 Rating of usefulness of deformation and movement monitoring for identification of different modes of failure
Internal
Failure Foundation Slope
Erosion Overtopping Seepage Seismicity
Mode Failure Instability
and Piping
114
encourage initiation of piping;
• Overtopping: Settlement of the dam crest can reduce freeboard and encourage
overtopping. Landslide of surrounding terrain can decrease storage capacity;
• Seepage: Material carried by water flow can deform embankment/foundation
geometry; and
• Seismicity: Can cause settlement or lateral spreading.
5.1.5.4.2. Instrumentation
In order to understand the behaviour of the deformation and movement, a number of
instruments are available with real-time measurement capability. These are described on Page
5 and Page 6 of Appendix B.
5.1.5.5. Seismicity
Seismicity can present itself naturally or be induced by mining activities. Naturally through
earthquakes, the regional susceptibility to seismic behaviour is often understood and
accounted for in the design. Regardless, monitoring techniques are employed to help
understand the magnitude, distance to source, and the potential influence that these natural
events may have on current activities. Mining activities have also been empirically proven to
cause induced seismicity, from triggers such as underground rock burst, oil and gas
extraction, fluid injection and hydraulic fracturing, and pore pressure increase in faults.
5.1.5.5.1. Applicability
A generalised rating of the usefulness of direct monitoring of seismicity for identification of
different modes of failure is presented in Table 24.
Table 24 Rating of usefulness of seismicity monitoring for identification of different modes of failure
Internal
Failure Foundation Slope
Erosion Overtopping Seepage Seismicity
Mode Failure Instability
and Piping
115
For direct and supporting indicators, justification is provided as:
Direct Indicators
• Slope Instability: Can cause settlement or lateral spreading; and
• Internal Erosion and Piping: Resultant cracking within the embankment can initiate
internal preferential erosion and piping.
Supporting Indicators
• Foundation Failure: Depending on foundation materials, either static or dynamic
liquefaction of saturated or partially saturated soils can occur. The stiffness and shear
strength of a material is significantly reduced due to rapid increases in loading; and
• Overtopping: Settlement as a result of particle rearrangement can reduce allowed
freeboard.
5.1.5.5.2. Instrumentation
In order to understand the behaviour of the seismicity, a number of instruments are available
with real-time measurement capability. These are described on Page 7 of Appendix B.
5.1.5.6. Earth Pressures
Earth pressures within tailings dams can provide an indication of the magnitude and direction
of stresses, the percentage contribution of water and soil to total pressures (when combined
with a piezometer), and in turn a comparison of design/expected conditions against actual. If
the stresses are different than anticipated, an anomaly may exist in the structure which may
not have been accounted for in design, and hence in assessment and management of any
associated risk.
5.1.5.6.1. Applicability
A generalised rating of the usefulness of direct monitoring of earth pressures for identification
of different modes of failure is presented in Table 25.
Table 25 Rating of usefulness of earth pressure monitoring for identification of different modes of failure
Internal
Failure Foundation Slope
Erosion Overtopping Seepage Seismicity
Mode Failure Instability
and Piping
116
For direct and supporting indicators, justification is provided as:
Direct Indicators
• Foundation Failure: Earth pressures can increase in the foundation indicative of uplift,
heave or blowout, or increased/excess loading from the embankment, tailings, and
water regime combination; and
• Slope Instability: A change in lateral earth pressures, increasing through driving forces
such as water pressure build-up or decreasing through relaxation or separation in the
embankment body can indicate instability.
5.1.5.6.2. Instrumentation
In order to understand the behaviour of the earth pressure, a number of instruments are
available with real-time measurement capability. These are described on Page 7 of Appendix
B.
5.1.5.7. Climate
Climate variability has primary influence on the water balance of the tailings storage facility.
In the geotechnical sense, water can influence each of the different modes of failure to some
degree. It is important to understand the amount and rate of change to compare against
designed state and anticipate upcoming challenges.
5.1.5.7.1. Applicability
A generalised rating of the usefulness of direct monitoring of the climate for identification of
different modes of failure is presented in Table 26.
Table 26 Rating of usefulness of climate monitoring for identification of different modes of failure
Internal
Failure Foundation Slope
Erosion Overtopping Seepage Seismicity
Mode Failure Instability
and Piping
117
Supporting Indicators
• For all: An increase in the anticipated water recharge (increased saturation) can raise
the phreatic surface, induce additional pore pressures, alter material parameters and
strength states, and in turn initiate any of the different modes of failure. A decrease in
the anticipated water recharge (drying) can induce cracking, shrink reactive soils, and
alter the structure of the soil materials used to form the embankment.
5.1.5.7.2. Instrumentation
In order to understand the influence of the climate, a number of instruments are available with
real-time measurement capability. These are described on Page 8 of Appendix B.
5.1.6. Online Monitoring Systems
For many tailings dam structures, the size and complexity of the monitoring system warrants
consideration of an integrated geotechnical monitoring system. These systems enable (adapted
from Hawley and Cunning 2017, GroundProbe 2019, and EHP Environment 2019):
• Real-time data acquisition and processing of multiple instrumentation systems;
• The ability to issue alerts and alarms, including threshold breaches and data upload
failures;
• Monitoring data can be stored in a central location and queried by authorised users at
any time, from anywhere in the world;
• The saving and backup of measurement data on secured servers;
• Automatic operational control over monitoring stations;
• Increased monitoring program reliability while reducing data acquisition and
processing costs;
• Measurement and data transmission interval changes remotely;
• Current and historical data interpretation, allowing interaction of any data along the
timescale; and
• Customisable reports comparing data from different instrumentation types.
With the tools available to digitise and streamline monitoring data, operators also have the
ability to not only reduce the time taken to manage and interpret the data, but also increase the
time allowed for critical analysis and understanding of the performance of their structure.
The software architecture of online monitoring systems is being fundamentally designed to
accommodate thousands of sensors, of different varieties. Automation of data processing
allows automatic alarm systems to be implemented: trigger levels aligned with the site Trigger
118
Action Response Plan (TARP), can automatically notify the appropriate persons of potentially
problematic conditions. This notification can be in the form of text message, email, pop-up
display, or other alerts that remove the need for 24/7 eyes on data trends.
When the engineer responsible for monitoring dam performance interrogates the data,
however, the user interface provides a number of options to facilitate data interpretation,
including:
• Customised dashboards suitable for individual sites;
• Plotted data visualisations, combining multiple instruments and aggregations for totals,
average, maximum and minimum;
• Scalable data presentation for different time periods;
• Fully customisable reporting for effective communication;
• Real-time displays overlain to 2-dimensional cross-sections, or 3-dimensional mine
surveys;
• Measurement contour plot for wide-area interrogation; and
• Input of complex equations to derive calculated values, as suitable.
There are two main challenges to highlight when considering online monitoring systems. The
first is internal and external data security. Data security is being addressed through software
inclusions such as fine grained user permissions, allowing in-house managers to designate
user control and viewing levels, and secure communication protocols. The second challenge
to highlight is ensuring that processed data should not be mistaken for interpreted data. “The
delicate task of fine interpretation belongs to the engineer” (Fell et al. 2015). The
measurements need to be applied to the tolerable levels at the instrument location to translate
the data into an understanding of dam performance. An example of this data collection,
management, and interpretation process as linked to site TARPs is presented in Appendix E.
5.1.7. Conclusions
This research established a catalogue of real-time instrumentation and monitoring techniques
for tailings dams, focusing on the key performance parameters for identification of prevalent
failure modes and mechanisms. The advancement of technology to be able to present
measured data in real-time mitigates many of the risks associated with manual data collection.
Through mitigation of these risks, the responsible engineer can more readily and reliably
understand the performance of the tailings dam.
Through collaboration with a global selection of instrumentation suppliers and a generalised
approach (at this stage of the wider project), it is anticipated that this catalogue is applicable
119
to tailings dams across the globe. By transparently listing the advantages, disadvantages,
specifications, and considerations that are pertinent to each tailored instrument type,
practitioners have the opportunity to be better informed when engaging with consultants and
suppliers, but most importantly are presented with a tool that can help toward a better
understanding and management of their tailings dam.
5.1.8. Reference List
Clarkson, L & Williams, D 2019, ‘Critical Review of Tailings Dam Monitoring Best
Practice’. International Journal of Mining, Reclamation and Environment, vol. 34, iss.
2, pp. 119-148.
Clarkson, L & Williams, D 2021, ‘Catalogue of Example Instrumentation and Monitoring
Systems for Tailings Dams in Australia’, Mining Technology.
Clarkson, L, Williams, D & Seppälä, J 2020, ‘Real-time monitoring of tailings dams’.
Georisk: Assessment and Management of Risk for Engineered Systems and
Geohazards.
EHP Environment 2019, EHP-Dataservice, viewed 17 June 2019,
<https://www.ehpenvironment.com/en/dataservice/>
Fell, R, MacGregor, P, Stapledon, D, Bell, G & Foster, M 2015, Geotechnical Engineering of
Dams¸ 2nd edn, CRC Press/Balkema, The Netherlands.
GroundProbe 2019, Geoexplorer, viewed 17 June 2019,
<https://www.groundprobe.com/product/geoexplorer/>
Hawley, M and Cunning, J. 2017, Guidelines for Mine Waste Dump and Stockpile Design,
CSIRO Publishing, Australia.
ICOLD 1983, Deterioration of dams and reservoirs, examples and their analysis.
International Commission on Large Dams, Paris.
ICOLD 2001, Tailings Dams—risk of dangerous occurrences, lessons learnt from practical
experiences, United Nations Environmental Programme (UNEP), Division of
Technology, Industry and Economics (DTIE) and International Commission on Large
Dams (ICOLD), Paris, France, Bulletin 121
Klohn Crippen Berger (KCB) 2018, Static Liquefaction and Strength Loss in Tailings Dams,
viewed 30 November 2019, <https://www.klohn.com/blog/static-liquefaction-
strength-loss-tailings-dams/>
120
5.1.9. Bibliography - Appendix B Supplementary Research
Avella, S 1993, ‘An analysis of a worldwide status for monitoring and analysis of dam
deformation’, Masters Thesis, University of New Brunswick, Fredericton, New
Brunswick, Canada.
Avnet Abacus 2019, Piezoresistive pressure sensors, viewed 11 January 2020,
<https://www.avnet.com/wps/portal/abacus/solutions/technologies/sensors/pressure-
sensors/core-technologies/piezoresistive-strain-gauge/>
Báťková, K., Matula, S., Miháliková, M. 2013. Multimedial Study Guide of Field
Hydropedological Measurements. (2nd revised edition). English version. Czech
University of Life Sciences Prague. Prague, Czech Republic. No pagination. Available
at: http://hydropedologie.agrobiologie.cz/en-index.html
Bengtson, H 2020, Use a Rectangular Weir to Measure Water Flow Rate in an Open Channel,
viewed 11 January 2020, <https://www.brighthubengineering.com/hydraulics-civil-
engineering/65880-open-channel-flow-measurement-5-the-rectangular-weir/>
Boyes, W 2013, Understanding how ultrasonic continuous level measurement works, viewed
11 January 2020, <https://www.controlglobal.com/articles/2013/automation-
technology-ultrasonic-continuous-measurement/>
Campbell Scientific 2020, Automated Weather Stations, viewed 11 January 2020,
<https://www.campbellsci.com/automated-weather-stations>
Campbell Scientific Australia 2020, 229-L Heat Dissipation Matric Potential Sensor, viewed
11 January 2020, <https://www.campbellsci.com.au/229-l>
Campbell Scientific Europe 2020, AVW200 – A New Vibrating Wire Interface, viewed 11
January 2020, <https://www.campbellsci.eu/vibrating-wire>
Canary Systems 2019, Vibrating Wire Piezometer Deployment Guidelines, viewed 11
January 2020, <http://www.canarysystems.com/nsupport/canary_san24.pdf>
Chipkin Automation Systems 2017, Velocity Meters, viewed 11 January 2020,
<http://www.chipkin.com/velocity-meters/>
Clayton, CRI, Matthews, MC, and Simons, NE 1995, ‘Basic field instrumentation for site
investigation’, in Site Investigation (2nd ed.), Department of Civil Engineering,
University of Surrey, Chapter 10.
Corry, CE, DeMoully, GT, and Gerety, MT 1982, ‘Field Procedure Manual for Self-Potential
Surveys’, Zonge Engineering & Research Organisation, Tucson, Arizona.
121
Revil, A, and Jardani A 2013, ‘Fundamentals of the self-potential method’ in The Self-
Potential Method, Theory and Applications in Environmental Geosciences, Cambridge
University Press, excerpt.
Durham Geo-Enterprises, Inc 2019, Pneumatic Piezometers, viewed 11 January 2020,
<https://durhamgeo.com/product/pneumatic-piezometers/>
Durham Geo-Enterprises, Inc 2019, Serial HD In-Place Inclinometers, viewed 11 January
2020, <https://durhamgeo.com/product/serial-hd-in-place-inclinometers/>
Durham Geo-Enterprises, Inc 2019, Temperature Sensors, viewed 11 January 2020,
<https://durhamgeo.com/product/temperature-sensors/>
Durham Geo Slope Indicator 2004, Guide to Geotechnical Instrumentation, viewed 11
January 2020, <https://durhamgeo.com/pdf/documents/course%20material/guide-to-
instrumentation.pdf>
Edaphic Scientific 2019, water flow /velocity meters, viewed 11 January 2020,
<https://www.edaphic.com.au/products/water/swoffer-water-flow-meters/>
El-Hakim, M 2009. “Instrumentation and Overall Evaluation of Perpetual and Conventional
Flexible Pavement Designs.” Masters Thesis, Department of Civil Engineering,
University of Waterloo, Waterloo, Ontario, Canada.
Elprocus 2020, All You Know About LIDAR Systems and Applications, viewed 11 January
2020, <https://www.elprocus.com/lidar-light-detection-and-ranging-working-
application/>
Encardio Rite 2020, Piezometers: Types, Functions & How it Works?, viewed 11 January
2020, <https://www.encardio.com/blog/piezometers-types-functions-how-it-works/>
Federal Energy Regulatory Commission (FERC) 2018, ‘Instrumentation and Monitoring’, in
Office of Energy Projects, Engineering Guidelines for the Evaluation of Hydropower
Projects, FERC, Chapter 9, pp. 9-1 to 9-56.
Froehlich, D, and Dhawan, AK 2017, Guidelines for Instrumentation of Large Dams, Central
Water Commission, Ministry of Water Resources, River Development and Ganga
Rejuvenation, Government of India.
Geofabrics Australasia Pty Ltd.n.d., Bidim C Range – Conductive Geotextile for
Geomembrane Liner Leak Detection, viewed 11 January 2020,
<https://www.geofabrics.co/products/bidim-c-range>
Geokon 2019, MEMS In-Place Inclinometer Systems, viewed 11 January 2020,
<https://www.geokon.com/content/datasheets/6150F-Series-MEMS-IPI.pdf>
122
Geokon 2019, Settlement Plate, viewed 11 January 2020,
<https://www.geokon.com/content/datasheets/4625_Settlement_Plate.pdf>
Geokon 2019, Vibrating Wire Tiltmeter, viewed 11 January 2020,
<https://www.geokon.com/content/datasheets/6350_Vibrating_Wire_Tiltmeter.pdf>
Geokon 2019, Vibrating Wire Weir Monitor, viewed 11 January 2020,
<https://www.geokon.com/content/datasheets/4675LV_Weir_Monitor.pdf>
Geokon 2019, VW Piezometers & Pressure Transducers, viewed 11 January 2020,
<https://www.geokon.com/content/datasheets/4500_Series_VW_Piezometers.pdf>
Geoscience Australia n.d., Interferometric Synthetic Aperture Radar, viewed 11 January 2020,
<https://www.ga.gov.au/scientific-topics/positioning-navigation/geodesy/geodetic-
techniques/interferometric-synthetic-aperture-radar>
Gommenginger, CP, Clarizia, MP, and Unwin, M n.d., ‘Towards Long-Term Sustainable
Observations of Ocean Wind and Waves with GNSS Signals of Opportunity’,
ResearchGate.
GroundProbe 2015, Ground Breaking News – Introducing SSR-SARx, viewed 11 January
2020, <https://www.groundprobe.com/wp-content/uploads/2017/06/English-
Introducing-SSR-SARx.pdf>
Instrumentationtoolbox.com 2019, Bubbler Tube System for Level Measurement – Operating
Principle, viewed 11 January 2020,
<https://www.instrumentationtoolbox.com/2014/06/bubbler-tube-system-for-
level.html>
Interstate Technology & Regulatory Council (ITRC) 2013, Appendix D. Weirs and Flume
Size and Flow Calculations, viewed 11 January 2020, <https://itrcweb.org/bcr-
1/Content/Appendix%20D%20Weirs%20Flume%20Size.htm>
Jewell Instruments 2020, GEOTECHNICAL MEMS TILTMETER SERIES, viewed 11
January 2020, <http://www.jewellinstruments.com/geotechnical-mems-tiltmeter-
series/>
Mashari, O 2019, 2D Borehole Sonic Logging Simulations, viewed 11 January 2020,
<https://support.onscale.com/hc/en-us/articles/360002276277-2D-Borehole-Sonic-
Logging-Simulations>
METER Group 2018, User Manual T5/T5x Pressure Transducer Tensiometer, viewed 11
January 2020, <http://library.metergroup.com/Manuals/UMS/T5_Manual.pdf>
123
Openchannelflow 2020, Discharge Tables, viewed 11 January 2020,
<https://www.openchannelflow.com/flumes/parshall-flumes/discharge-tables>
Openchannelflow 2020, Weirs, viewed 11 January 2020,
<https://www.openchannelflow.com/weirs>
Patil, BN 2015, A Review of Various Soil Moisture Measurement Techniques, PowerPoint
slides, RadiusVision 2018, Solar Powered Security Camera Systems, viewed 11
January 2020, <https://radiusvision.com/complete-systems/solar-powered-ip-camera-
system/>
RdF Corporation n.d., Platinum RTD Probe Construction, viewed 11 January 2020,
<http://www.rdfcorp.com/anotes/pa-r/pa-r_01.shtml>
REF TEK n.d., Dam, Tailing Dam & Reservoir Monitoring, viewed 11 January 2020,
<https://www.reftek.com/dam-reservoir-monitoring/>
ReportBuyer 2017, Seismometers Market: By Type (Teleseismometers, Strong-Motion
Seismometer, Others); By Range (50 to 750 V/m, 1500 V/m, 20,000 V/m); By
Varieties (Short Period, Long Period, Broadband); By Output (Analog, Digital) & By
Region-Forecast (2016-2022), viewed 11 January 2020,
<https://www.prnewswire.com/news-releases/seismometers-market-by-type-
teleseismometers-strong-motion-seismometer-others-by-range-50-to-750-vm-1500-
vm-20000-vm-by-varieties-short-period-long-period-broadband-by-output-analog-
digital--by-region-forec-300404462.html>
Ridley, AM 2015, 'Soil suction — what it is and how to successfully measure it', Proceedings
of the Ninth Symposium on Field Measurements in Geomechanics, Australian Centre
for Geomechanics, Perth, pp. 27-46
RST Instruments n.d., Digital Bus Vertical In-place MEMS Inclinometer System, viewed 11
January 2020, <https://www.rstinstruments.com/content/brochures/Vertical-In-place-
MEMS-Inclinometer-ICB0044D.pdf>
RST Instruments n.d., Monitoring Instruments for Earthfill Dams, viewed 11 January 2020,
<https://rstinstruments.com/applications/earthfill-dams>.
RST Instruments n.d., Vibrating Wire Liquid Settlement System, viewed 11 January 2020,
<https://www.rstinstruments.com/content/brochures/Vibrating-Wire-Liquid-
Settlement-System-SSB0005O.pdf>
Shrama Sadhana Bombay Trust’s College of Engineering and Technology, Bambhori,
Jalgaon.
124
Specto Technology 2020, Piezometers, viewed 11 January 2020,
<https://www.spectotechnology.com/product/piezometer/>
Splinter, R 2017, Illustrated Encyclopedia of Applied and Engineering Physics, Three-Volume
Set, CRC Press.
The G3 Group 2010, Time Domain Reflectometry (TDR), viewed 11 January 2020,
<http://www.geo3group.com/time-domain-reflectometry-tdr/>
Total Control Systems 2020, TCS Literature Downloads (Flow Meters), viewed 11 January
2020, <http://www.tcsmeters.com/product-literature-and-documentation.html>
TRICOR Coriolis Technology 2020, TRICOR Coriolis Mass Flow Meters, viewed 11 January
2020, <https://tricorflow.com/products/>
Vuidart, I, Hadadou, R, Hanocq, P, and Semmelbeck, L 2015, ‘Post-mining management in a
major French mining area – example of the Lorraine iron ore basin’, Proceedings of
the 9th International Conference on Mine Closure, Mine Closure 2014, Johannesburg,
South Africa.
WatElectronics.com 2019, Liquid Level Sensor – Different Types of Level Sensors and their
Workings, viewed 11 January 2020, <https://www.watelectronics.com/liquid-level-
sensor-and-types-of-level-sensors/>
Yokogawa Australia Pty Ltd. 2020, Rotameter – Variable Area Flowmeter, viewed 11
January 2020, <https://www.yokogawa.com/au/solutions/products-platforms/field-
instruments/flow-meters/variable-area-flow-meter-rotameters/>
125
5.2. Catalogue of Example Instrumentation and Monitoring
Systems for Tailings Dams in Australia
Luke Clarkson1 & David Williams1
1
Geotechnical Engineering Centre within the School of Civil Engineering, The University of
Queensland, Brisbane, QLD 4072 Australia
5.2.1. Abstract
New literature, mining regulators, insurance firms, and mining practitioners are calling for
increased diligence in the form of real-time monitoring: but what can the industry offer in
response? This research establishes a centralisation of instrumentation systems suitable for
tailings dams, discussing the specifications, advantages, and disadvantages of each. This
paper describes an understanding of dataloggers and sensor nodes, sensor networks, and
online monitoring systems critical to the transmission and reception of sensor data across vast
and remote site areas. Collaboration with suppliers discovered systems which enable reliable,
efficient, and real-time transmission of instrumentation data, ready for interpretation. This
paper aims to translate traditionally electrical and systems engineering terminology into a
reference base suitable for the broad range of tailings dam practitioners. This reference base is
anticipated to help facilitate informed discussions and encourage deployment of appropriate
instrumentation systems that are suitable to the practitioner’s short- and long-term intent.
5.2.2. Introduction
Tailings dam and mine site environments demand versatile, rugged, and reliable
instrumentation systems. The different nodes of these systems cannot always be connected by
cable, either due to risks posed by ongoing construction/development or due to the distance
that these cables would need to span. In traditional monitoring methods, this then meant that
the instruments were limited to manual readings at the location of the instrument, and the
frequency of data collection dependent on the accessibility and schedule of mine personnel.
With the advancement of technology in many different sectors (within and beyond mining),
the ability to automate the transmission of data from sensor to office is more feasible and
126
streamlined than ever before. In turn, this inherently relates the understanding of
instrumentation and monitoring systems to achieving real-time data collection in the tailings
dam environment.
Reliable data allows for prompt and informed decisions to be made in terms of health and
safety, maintenance, compliance, and productivity. However, to achieve reliability requires
due consideration of individual site requirements and in turn assignment of individual
monitoring system components. The selection and procurement of monitoring products can
include (ALS Limited 2020):
• Data loggers;
• Measurement and control devices;
• Telemetry systems;
• Sensors and interfaces;
• Software;
• Power supplies; and
• Enclosures, masts, and brackets.
Further, considering the requirements of different mine sites encourages the system to be
tailored to:
• Power availability (on-grid/off-grid), and in turn whether instruments and systems need
to consider different power supplies such as battery or solar;
• Proximity (distance between datalogger and office), and the allowable operating range
of different systems as well as data storage capability for less frequent visits;
• Obstructions, and whether this has implications on line-of-sight transmission;
• Climate, and in turn harsh weather conditions including heat, freeze, wind, lightning,
and rain;
• Commodity, and as such exposure of instruments and systems to chemical aggressivity,
moisture, or other characteristics of the tailings residue itself;
• Pre-existing instruments, and if these are to be retrofitted to the new monitoring
system; and
• Protection, in terms of animal damage (livestock rubbing) and theft.
It is clear to see that a tailored approach is required, and there is likely no one-size-fits-all
solution considering the vast differences that can be experienced even between neighbouring
mine sites. A focus on improving traditional, manually monitored instrumentation systems is
observed in a number of elements critical to safety of the tailings dam structure and in turn the
127
stakeholders to its failure. Such elements include:
• Interrogation of more frequent data points, leading to less assumption when needing to
interpolate between two discrete data points read 1 month apart, for example. Also
allows for more feasible integration of rate-based Trigger Actions Response Plans (for
example, mm/hr movement triggers) on different instruments;
• Optimisation of instrumentation, both in terms of selective installation of future
instrumentation based on a better understanding of current conditions, as well as the
potential for overlapping instrumentation use (e.g. seepage and environmental
monitoring);
• Better informed understanding of geotechnical mechanisms, through mitigation of data
limitations clouding the interpretation; and
• Design response, where tailings dam constructive is a progressive and iterative process,
original designs are often revisited to ensure life-of-mine plans for the structure remain
valid. Through a better understanding of the structure performance and condition based
on available instrumentation data, the design can be more readily tailored to actual
conditions with fewer assumptions.
It is important to note that there has not been a single tailings dam failure in history that, in
hindsight, was not proven to have been predictable with the information available. The severe
consequences and risk of tailings dam failure globally demands improvement in how
practitioners understand and monitor the performance of the dam without complacency and
putting people’s lives and the environment at risk. There is speculation (for or against) as to
the ability of instrumentation to forecast tailings dam deterioration leading to failure.
However, to date there have not been sufficient case examples to prove this true or false.
Hence, it is suggested that the usefulness of real-time data in predicting events is not yet
known, however without exploring the option it shall remain as an unknown and risk
mitigation of tailings dams may not focus on key contributing factors, forward.
5.2.3. Method
This paper forms one part of a wider research project by the authors to develop a
comprehensive monitoring strategy for tailings dams. This paper aims to identify and
catalogue the present-day range of systems and networks required to implement real-time
instrumentation and technologies in tailings dams. Supplementary to this is a separate paper
(Clarkson and Williams 2021) which describes the instrumentation and technologies.
To complement the research and enhance applicability in the industry context, a number of
128
suppliers were engaged with nine suppliers providing input to this research paper. A
scrutinised compilation of the feedback provided is presented within this paper. The
contributing suppliers are acknowledged on Page 4 of Appendix C of this paper.
5.2.4. Real-Time Tailings Dam Monitoring
A parallel paper by the author (Clarkson, Williams, & Seppälä 2020) describes the monitoring
requirements and different components as part of a real-time instrumentation system. “A
modern real-time slope instrumentation system generally comprises the following four key
components” (Clarkson, Williams, & Seppälä 2020):
1. Sensor nodes which consist of instruments and a datalogger;
2. Communication network for sensor nodes;
3. Base station with an external communication modem. This resides at the computer site
and serves as a link between field stations and repeater stations; and
4. Information system for storing and displaying the data.
In achieving a real-time instrumentation system, there are multiple layers of solutions
requiring consideration, “including protocols and topology related to optimal operational
range, battery life, and system contingency, amongst others” (Clarkson, Williams, & Seppälä
2020). “It is not necessary for operators to understand the complexities of the systems to
meticulous degrees; it can be argued that technology design focused on functionality, usability
and user experience should be a priority in a field where rapid decision making and safety
critical response is required.” (Clarkson, Williams, & Seppälä 2020).
It is suggested that a relatively increased value can be extracted from individual instruments
when considered as part of an overall system, as opposed to simply interrogating these in
isolation. However, to progress an instrumentation system beyond manual readings and
isolated instruments inevitably requires a network of sensors, transmitters, and receivers
which rely on other discrete elements to perform effectively. In the hypothetical scenario that
a critical real-time instability event were occurring, it would be prudent that the
instrumentation system is fully functional both in terms of practical transmission of data
(network is intact) and any lag by which the data suffers prior to being available for
interpretation (network performance is sufficient).
Considering this case, this paper aims to describe the network components that can be utilised
to achieve an effective and appropriate system, and also to simplify terminology and
accelerate understanding of the broad range of tailings dam practitioners to allow informed
and productive discussion.
129
5.2.5. Instrumentation Systems
A detailed insight into the systems that can facilitate the capacity for real-time, or improved
monitoring of various instruments associated with tailings dam monitoring are described in
Appendix C. This is proposed to be read in parallel to detail of the instruments themselves
(Clarkson & Williams 2020) with the intent to understand the requirements of instruments as
a system, as opposed to isolated nodes. Viewing instrumentation as a system emphasises the
significance of each element in the chain until ultimately the data is presented for
interpretation.
Considering this, it is important to note that the focus on instrumentation systems does not
negate the importance the interpretation of collected data: the responsibility of which should
remain with the engineer.
5.2.5.1. Sensor Nodes
Most “instruments measure the changes in electrical properties… which are then converted to
give ground displacement, distortion, or groundwater pressure, as appropriate. The changes in
these electric properties are usually brought about by variations in magnetic field, natural
frequency, electric resistance or conductivity due to displacement or elongation of the
embedded elements of the instrument” (Tang & Cheung 2011). It is this electrical property
(whether transmitted by analog or digital) that allows the instrument’s signal to be collected
by a datalogger. Together, the instrument and datalogger are typically connected by means of
a cable or wireless communication in order to make a sensor node.
The “Measurement Devices” described in Appendix C detail dataloggers and different data
acquisition systems that retrieve data from the instruments and sensors themselves. It is
intended that the datasheets presented in Appendix C can be read in conjunction with this
section to allow rapid familiarisation with electrical and systems engineering approaches, and
how these might apply to practitioners responsible for the monitoring of tailings dams.
For the purpose of this paper, the author refers the reader to parallel works (Clarkson &
Williams 2020) for instrument details, where this paper focuses on the datalogger component
of the sensor node.
5.2.5.1.1. Acronyms and Definitions
Within the feedback that was received from suppliers, it was evident that the language used in
instrumentation systems is vastly different from that used in standard civil, geotechnical, and
mining engineering applications. This is predominantly due to the electrical and systems
engineering focus of instrumentation systems, however for the purpose of bridging this gap
130
and allowing increased collaboration and understanding between disciplines, a high-level
description of key “Measurement Devices” terms relating to this research are described by
section, as:
5.2.5.1.1.1. Analog Inputs
“An analog signal is a kind of signal that is continuously [or infinitely] variable [to represent
any small amount of change], as opposed to having a limited number of steps along its range
(called digital)” (Kuphaldt 2000). Analog inputs convert an electrical signal into a digital
value that can be stored and processed in a computer. Different types of input and
measurement include:
• Analog Input and Output (I/O): “the signals from sensors that measure natural factors
such as temperature, pressure, and flow rate are often analog signals” (CONTEC n.d.).
An analog I/O interface bridges the analog signal and the digital signal handled by the
computer;
• Differential (DIFF): “differential input uses two signal wires and a ground wire for a
total of three wires for measuring the voltage of a signal source” (CONTEC n.d.). Has
the advantage over SE inputs that DIFF is less susceptible to noise. Has the
disadvantage against SE inputs that three wires are required for one signal source,
making one half the number of channels that can be used;
• Single Ended (SE): the most common input method for analog inputs, a single-ended
input uses a signal wire and a ground wire for a two-wire connection, measuring the
voltage of a signal source using the potential difference from ground;
• Universal (U): “universal inputs/outputs are for any mix of temperature, pressure, flow,
status points, and similar point types” (Schneider Electric 2017). Universal input also
carries a power source input range and frequency operating range that allows use
anywhere in the world without adjusting for input power; and
• Configurable: supporting both analog and digital signals.
Analog inputs, in the instrumentation case, are commonly sourced from signals in the form of
voltage or current. Voltage signals are used in many applications because of their inherent
design simplicity. The signal standard defines the range of measurement (e.g. 0-10 volts is 0%
of measurement to 100% of measurement). Commonly, a ‘live zero’ concept can be applied
which describes a way of scaling a signal so that an indication of 0% measurement can be any
number (such as 4 – 20mA). When the baseline value is a non-zero value, this allows a more
ready assessment of system malfunction (which would be indicated by a zero signal value).
131
Voltage is measured in instruments in different ways, many of which abide by the
fundamental Ohm’s Law (Voltage = Current x Resistance):
• Current Loop: the 4-20 mA loop standard is considered to be the dominant standard for
analog signal that can be used to transmit process information. Following Ohm's Law,
the power supply provides voltage to drive the current through the current loop.
Current is the same in all places throughout the loop, and this is regulated between 4-
20mA by the transducer (the magnitude of current is dependent on the input coming
from the instrument sensor). A data acquisition device then measures the magnitude of
regulated current and converts this to a measurement value. Advantaged by using
current rather than voltage, in that “the inherent disadvantage of using voltage to
transmit signals in an industrial environment is voltage drops over long cable lengths
due to wire resistance” (National Instruments 2019);
• Data Measurement Method:
• Period Average Reading: repeating a measurement several times and averaging the
results, to account for uncertainty and fluctuation. “This average is generally the
best estimate of the "true" value” (Advanced Instructional Systems, Inc &
University of North Carolina 2011); and
• Static Reading: “the static characteristics of instruments are related with steady
state response… [and defines] the relationship between the output and the input
when the input does not change, or the input is changing at a slow rate”
(Engineering Institute of Technology 2019).
• Full Wheatstone Bridge (FWB): “wheatstone bridge strain gauge circuits are used
extensively inside transducer to convert a mechanical strain into an electrical output
signal… A Wheatstone bridge strain gauge circuit is created by mounting a pair of strain
gauges on a material that will be stressed, so that when a force is applied, they will stretch
along their width” (SensorsONE Ltd 2020);
• Potentiometer (POT): the output voltage value of a potentiometer is changed or varied by
the movement of a mechanical contact, by some external action. The physical position of
the mechanical contact along a fixed resistive element defines a ratio between known
(input) voltage (end-to-end) and measured (output) voltage (end-to-mechanical contact)
(Electronics Tutorials n.d.);
• PT100 Resistance Thermometers: resistance thermometers measure temperature through
the fundamental operating principle of measuring the resistance of a platinum element.
132
“The most common type (PT100) has a resistance of 100 ohms at 0°C and 138.4 ohms at
100°C” (Pico Technology n.d.). Assuming a linear relationship between temperature and
resistance over this range presents a 0.4°C error at 50°C, however more precision
measurement is possible;
• Thermistor: thermistors are semiconductor devices with known resistance properties, used
to measure temperature. The most common measurement type is Negative Temperature
Coefficients (NTC thermistors), where the thermistor’s resistance decreases as
temperature increases. By understanding the properties of the materials that make up the
thermistor, the resulting resistance-temperature relationship provides an indication of
temperature measurement;
• Thermocouple: a thermocouple is a sensor that measures temperature. It consists of two
different types of metals, joined together at one end. When the junction of the two metals
is heated or cooled, a voltage is created that can be correlated back to the temperature;
• Ratiometric Bridge: “in ratiometric measurements, the measured quantity is proportional
to a ratio of two voltages rather than an absolute voltage” (Campbell Scientific 2001). It is
suggested that using ratios allows an assessment independent of the reference voltage (and
hence a stable reference point is not critical in achieving accurate measurements).
Ratiometric measurements eliminate inaccuracy caused by the temperature drift of the
measurement voltage reference.
5.2.5.1.1.2. Data Transmission
There are a number of different types of dataloggers, distinguished primarily by their data
transmission method, input format, and power source. The data transmission method
predominantly influences the rate of data transfer and power requirements. Different types of
data transmission methods include:
• Ethernet (see Figure 28): “ethernet is the traditional technology for connecting devices
in a wired local area network (LAN) or wide area network (WAN)” (Rouse et al.
2015). Communicating via a common set of rules/network language, the term
‘Ethernet’ describes the protocol by which other devices can recognise, receive, and
process the information. Compared to wireless technologies, Ethernet is “typically less
vulnerable to disruptions… [and] can also offer a greater degree of network security
and control… since devices must connect using physical cabling” (Rouse et al. 2015).
• Universal Serial Bus (USB) Micro B (see Figure 28): most commonly related to the
ports that different phones offer for charging, USB Micro B is simply one type of cable
133
connection, typically with a USB connection on the other end of the cable. In the
datalogger case, micro USB ports are “primarily used for datalogger programming and
testing” (Rhodes 2018), or power.
• 802.11n: “802.11n [also known as Wi-Fi 4] is an IEEE (Institute of Electrical and
Electronics Engineers) industry standard for local Wi-Fi network communications”
(Mitchell 2020). This standard has been superseded by newer 802.11 standards,
however supports up to 300Mbps in bandwidth (depending on network conditions), and
operates in both the 2.4GHz and 5GHz bands. The benefit to using an older standard is
that the newer standards do not operate over the two bands, which may limit their
compatibility with older devices (pre-2014).
• RS-232 (Recommended Standard, see Figure 28) was, at one time in history, the most
widely used communication standard globally. The physical characteristics of the
hardware include both a 25 pin and 9 pin D sub connector. RS-232 is capable of
operating at data rates up to 20kbps and can send data about 15 metres. While this
method is cheap and easy, it is “also susceptible to noise and almost always required a
lower data rate than balanced protocols such as RS-485” (B+B SmartWorx 2018b).
• RS-485 (Recommended Standard, see Figure 28): favoured for its noise immunity and
long-distance capability, the RS-485 “can send serial data up to 1200 metres without a
repeater” (B+B SmartWorx 2018a). RS-485 allows multiple devices (up to 32) to
communicate at half-duplex on a single pair of wires, plus a ground wire. The original
standard definition states that RS-485 “can support a maximum data rate of 10Mbps at
12m distance, and 100kbps at the maximum 1200m distance” (Maxim Integrated
Products 2014), however it is suggested that advancements have been made since the
original standard and higher rates may be achievable with the right equipment.
• Data Communication Form:
• Modbus Remote Terminal Unit (RTU): “modbus RTU is an open serial protocol
derived from the Master/Slave architecture originally developed by Modicon (now
Schneider Electric)” (Real Time Automation 2020). Utilising simple 16-bit structure to
ensure reliability, data can be translated in floating point, tables, ASCII text, and
queues, amongst others. Primarily uses RS-232 or RS-485 serial interfaces, however is
well established and open standard, meaning it is free to build in, and predictably then
“is supported by almost every commercial… data acquisition software program in the
marketplace” and is “the most commonly available means of connecting industrial
134
electronic devices” (Real Time Automation 2020).
• Digital Bus Sensors: defined as the communication interface between the sensor and
the microcontroller, the bus structure is influence by many factors including cost,
reliability, and energy efficiency, amongst others. Digital bus sensors can be
incorporated into wireless or standalone data collection systems, featuring “a single
cable running the length of an entire chain of connected sensors/instruments, which are
all individually addressable” (RST Instruments n.d.).
• Radio Connection: using radio connections to connect the controller to a sensor (noting
that radio regulations specific to country of operation must be checked, in case a radio
license is required) (Trimble 2019). Either the controller’s internal radio (controller -
instrument) or an external radio (controller - external radio - instrument) can be used.
Figure 28 Different pinout (port) arrangements for transmission types described in this paper
Maximum Value 69
Average Value 17
Median Value 9
Dishwasher 1,500
Microwave 1,000
Toaster 850
Television 500
Laptop 300
Monitor 200
137
Ceiling Fan 60
Electric Shaver 15
Scanner 10
1
Note: For conversion purposes when power is presented in a different unit, the power
equation states that Power (Watts) = Voltage (Volts) x Current (Amps).
2
From Generatorist (2020)
It is important to note that some sensor networks allow idle, stand-by, and low power modes,
so as to mitigate power consumption when not actively transmitting data. For standby and low
power modes, the average power consumption is approximately 0.4W, while the average
power consumption in idle scenarios is 9W.
5.2.5.2.1.2. Operating Bands
Dataloggers can be provided with or without data transmission, which refers to the WiFi,
General Packet Radio Service (GPRS, mobile data, 2G/3G/4G etc.), and Long Range (LoRa)
technologies, that act as the medium by which information is transmitted. These technologies
all operate within distinct bands of the radiofrequency spectrum. The frequency of a radio
wave determines its characteristics, such as (Australian Government 2019):
• The distance the radio wave can travel;
• Whether it can penetrate through trees or into buildings; and
• The cost of equipment, which generally increases as the frequency increases.
Importantly, the inverse proportionality between wavelength and frequency must be tailored
to allow different services to deliver a specific purpose. Longer wavelengths (lower
frequencies) “need larger antennas but can travel longer distances than short wavelengths”
(Australian Government 2019). The balance between cost, type and number of users
impacted, and in turn high demand of desirable operating bands must be balanced, and hence
legislation (such as the Australian Radiofrequency Spectrum Plan 2017) has been developed
to manage technologies using this spectrum.
A simplified operating band designation for different technologies, against the wavelength
and frequency that applies to each is shown in Figure 29.
138
Figure 29 Simplified operating band designation for different technologies, from TeraSense Group (2019)
Advanced Instructional Systems, Inc & University of North Carolina 2011, Measurements
and Error Analysis, viewed 06 April 2020,
<https://www.webassign.net/question_assets/unccolphysmechl1/measurements/manua
l.html>
143
ALS Limited 2020, Customised monitoring instrumentation, viewed 06 April 2020, <
https://www.alsglobal.com/en-au/services-and-
products/environmental/hydrographics/customised-monitoring-instrumentation>
Australian Government 2019, What is spectrum?, viewed 06 April 2020,
<https://www.acma.gov.au/what-spectrum>
B+B SmartWorx 2018a, RS-485 Solutions, viewed 06 April 2020, <http://www.bb-
elec.com/Solutions/RS485.aspx>
B+B SmartWorx 2018b, What is RS-232, viewed 06 April 2020, <http://www.bb-
elec.com/Learning-Center/All-White-Papers/Serial/What-Is-RS-232.aspx>
Bosworth & Taylor 2016, Bandwidth Demands Place New Strains on Satellite
Communications Design, viewed 06 April 2020,
<https://www.analog.com/en/technical-articles/satellite-communications-design.html>
Campbell Scientific 2001, Voltage Measurement Accuracy, Self-Calibration, and Ratiometric
Measurements, viewed 06 April 2020,
<http://www2.ece.rochester.edu/courses/ECE111/voltaccy.pdf>
Campbell Scientific 2018, Base Station, RF500B, viewed 06 April 2020,
<https://s.campbellsci.com/documents/au/product-brochures/b_rf500b.pdf>
Campbell Scientific 2020, How does a spread-spectrum radio work?, viewed 06 April 2020,
<https://www.campbellsci.com/faqs?v=974>
Chambers, D 2016, Tailings Dam Failures 1915-2016, viewed May 11 2018,
<http://www.csp2.org/tsf-failures-1915-2017>
Clarkson, L & Williams, D 2021, Catalogue of Real-Time Instrumentation and Monitoring
Techniques for Tailings Dams, Mining Technology, vol. 130, iss. 1, pp. 52-59.
Clarkson, L & Williams, D. 2019, ‘Critical Review of Tailings Dam Monitoring Best
Practice’. International Journal of Mining, Reclamation and Environment, vol. 34, iss.
2, pp. 119-148.
Clarkson, L, Williams, D & Seppälä, J 2020, ‘Real-time monitoring of tailings dams’.
Georisk: Assessment and Management of Risk for Engineered Systems and
Geohazards.
CONTEC (n.d.), Analog I/O basic knowledge, viewed 06 April 2020,
<https://www.contec.com/support/basic-knowledge/daq-control/analog-io/>
144
Davies, M, Martin, D & Lighthall, P 2002a, 'Mine Tailings Dams - When Things go Wrong'
Proceedings of Tailings Dams 2000, Association of State Dam Safety Officials, U.S.
Committee on Large Dams, Las Vegas, Nevada: 261-273.
Electronics Notes n.d.a, LoRa Wireless for M2M & IoT, viewed 06 April 2020,
<https://www.electronics-notes.com/articles/connectivity/lora/what-is-lora-basics-
m2m-iot.php>
Electronics Notes n.d.b, What is 3G UMTS: WCDMA Tutorial, viewed 06 April 2020,
<https://www.electronics-notes.com/articles/connectivity/3g-umts/what-is-umts-
wcdma-tutorial.php>
Electronics Notes n.d.c., WiFi Standards: IEEE 802.11, viewed 06 April 2020,
<https://www.electronics-notes.com/articles/connectivity/wifi-ieee-802-
11/standards.php>
Electronics Tutorials n.d., Potentiometers, viewed 06 April 2020, <https://www.electronics-
tutorials.ws/resistor/potentiometer.html>
Engineering Institute of Technology 2019, Static and Dynamic Characteristics of Instruments,
viewed 06 April 2020, <https://www.eit.edu.au/cms/resources/technical-
resourses/static-and-dynamic-characteristics-of-instruments>
European Space Agency n.d., Satellite frequency bands, viewed 06 April 2020,
<https://www.esa.int/Applications/Telecommunications_Integrated_Applications/Sate
llite_frequency_bands>
Fell, R, MacGregor, P, Stapledon, D, Bell, G & Foster, M 2015, Geotechnical Engineering of
Dams¸ 2nd edn, CRC Press/Balkema, The Netherlands.
Generatorist 2020, Power Consumption of Household Appliances, viewed 06 April 2020,
<https://generatorist.com/power-consumption-of-household-appliances>
Hawley, M and Cunning, J 2017, Guidelines for Mine Waste Dump and Stockpile Design,
CSIRO Publishing, Australia.
Klohn Crippen Berger (KCB) 2018, Static Liquefaction and Strength Loss in Tailings Dams,
viewed 30 November 2019, <https://www.klohn.com/blog/static-liquefaction-
strength-loss-tailings-dams/>
Kuphaldt, TR 2000, ‘Electrical Instrumentation Signals – Analog and Digital Signals’, in All
About Circuits (ed.) n.d., Direct Current, viewed 06 April 2020,
<https://www.allaboutcircuits.com/textbook/direct-current/chpt-9/analog-and-digital-
signals/>
145
LePoudre, D.C. 2015, Examples, Statistics and Failure modes of tailings dams and
consequences of failure, PowerPoint slides, REMTECH.
Martindale, J 2020, What is Wi-Fi?, viewed 06 April 2020,
<https://www.digitaltrends.com/computing/what-is-wi-fi/>
Maxim Integrated Products 2014, How far and how fast can you go with RS-485?, viewed 06
April 2020, <https://pdfserv.maximintegrated.com/en/an/AN3884.pdf>
Mitchell 2020, 802.11n Wi-Fi in Computer Networking, viewed 06 April 2020,
<https://www.lifewire.com/802-11n-wireless-network-818281>
National Instruments 2019, Fundamentals, System Design, and Setup for the 4 to 20 mA
Current Loop, viewed 06 April 2020, <https://www.ni.com/en-au/innovations/white-
papers/08/fundamentals--system-design--and-setup-for-the-4-to-20-ma-curren.html>
Painter, L 2017, 4G vs LTE: What’s the difference?, viewed 06 April 2020,
<https://www.techadvisor.co.uk/feature/mobile-phone/4g-vs-lte-whats-the-difference-
3605656>
Pico Technology n.d., PT100 platinum resistance thermometers, viewed 06 April 2020,
<https://www.picotech.com/library/application-note/pt100-platinum-resistance-
thermometers>
Real Time Automation 2020, Modbus RTU Protocol Overview, viewed 06 April 2020,
<https://www.rtautomation.com/technologies/modbus-rtu/>
Rhodes 2018, Campbell Core Technology: The Datalogger, viewed 06 April 2020,
<https://www.campbellsci.com.au/blog/campbell-core-tech-datalogger>
Robertson, A.M. 2012, “FMEA Risk Analysis: Failure Modes and Effects Analysis.”
PowerPoint slides.
Rouse, M, Chai, W, Irei, A & Burke, J 2015, Ethernet, viewed 06 April 2020,
<https://searchnetworking.techtarget.com/definition/Ethernet>
RS Components n.d., The Comprehensive Guide to IP Ratings, viewed 06 April 2020,
<https://au.rs-online.com/web/generalDisplay.html?id=ideas-and-advice/ip-ratings>
RST Instruments n.d., DT2485 DT-BUS Data Logger, viewed 06 April 2020,
<https://www.rstinstruments.com/product/data-loggers/dt2485-dt-bus-data-logger>
Schneider Electric 2017, Universal Inputs/Outputs, viewed 06 April 2020,
<https://ecostruxure-building-
help.se.com/bms/topics/show.castle?id=11120&locale=en-US&productversion=1.8>
146
SensorsONE Ltd 2020, Wheatstone Bridge Strain Gauge, viewed 06 April 2020,
<https://www.sensorsone.com/wheatstone-bridge-strain-gauge/>
SignalBooster.com 2017, Differences between GSM, CDMA, TDMA, EVDO, UMTS and
HSPA+, viewed 06 April 2020,
<https://www.signalbooster.com/blogs/news/differences-between-gsm-cdma-tdma-
evdo-umts-and-hspa>
Tang, CSC, & Cheung, SPY 2011, Review of Real-time Data Transmission Systems for Slope
Instrumentation. GEO Report No. 262. Geotechnical Engineering Office, The
Government of the Hong Kong Special Administrative Region
TeraSense Group 2019, Radio Frequency Bands, viewed 06 April 2020,
<https://terasense.com/terahertz-technology/radio-frequency-bands/>
The Wireless Innovation Forum n.d., What is Software Defined Radio, viewed 06 April 2020,
<https://www.wirelessinnovation.org/assets/documents/SoftwareDefinedRadio.pdf>
Trimble 2019, Radio connections, viewed 06 April 2020,
<https://help.trimblegeospatial.com/TrimbleAccess/2019.00/en/Radio_Settings.htm>
Wilson Amplifiers 2020, The Differences Between 4G, 5G, and LTE, Explained, viewed 06
April 2020, <https://www.wilsonamplifiers.com/blog/the-difference-between-4g-lte-
and-5g/>
147
Chapter 6 The effect of different tailings dam environments and
external conditions on monitoring response
Luke Clarkson1, Todd Armstrong2 & David Williams1
1
Geotechnical Engineering Centre within the School of Civil Engineering, The University of
Queensland, Brisbane, QLD 4072 Australia
2
AECOM Australia Pty Ltd, Brisbane, QLD, Australia
This paper has been published in the Proceedings of the Tailings and Mine Waste Conference
2020, and adapted for inclusion in this thesis.
6.1. Abstract
This paper explores the influence of common external factors in the tailings dam environment
on phreatic conditions. The phreatic condition in tailings dams and deposited tailings responds
differently as a result of material permeability and changes in external conditions. It is
important for operators and designers to understand the influence of external conditions on the
phreatic condition, and how/when these changes might reflect at discrete piezometer
locations. Pore pressure changes induced at one location in the tailings dam can take a
significant amount of time to propagate to a given piezometer location and hence a delayed
effect could be expected. The paper describes a numerically modelled, observational
approach. Calibration of onsite piezometer measurements with pore water pressures modelled
in SEEP/W allowed consideration of the effect of pond level rise, lateral variation of
hydraulic conductivity, and the introduction of an external load on tailings dam pore water
pressure conditions.
6.2. Introduction
The demand for tailings storage is increasing exponentially with time, along with the
associated risks. It is estimated that the potential risk of tailings dams increases by 20-fold
approximately every 30 years (Robertson 2012). Chambers (2016) tailings dam failure
database describes an average annual rate of failure of 1 in 700 to 1 in 1,750, or two to five
annual failures for the roughly 3,500 tailings dams worldwide (LePoudre 2015 & Davies,
148
Martin & Lighthall 2002). By comparison, the estimated probability of failure for a
conventional water dam is 1 in 10,000 (Davies, Martin & Lighthall 2002). Tailings dam
failures, and the potential to forecast these failures ahead of time based on a better
understanding of existing conditions and performance is the focus of the first author’s
research. Understanding the capabilities of the installed instruments and recorded data is
clearly a critical necessity; however, the environmental conditions and an understanding of
the piezometer data are equally important.
One of the most common instruments installed in tailings dams is the piezometer, used for the
purpose of measuring phreatic surface (level where water is in equilibrium with atmospheric
pressure) and pore water pressures (void fluid pressure acting on the tailings particles) to
characterise the pore water pressure profile and select appropriate parameters and methods for
stability analyses. Typically, piezometers are installed in locations that will provide key
information to compare against design assumptions and the findings of seepage and stability
analyses, in particular comparison against allowable and safe limits. Naeini & Akhtarpour
(2018) describe the purpose of seepage analyses aligning with one of three objectives:
• “Locating the phreatic surface, managing the water resources or providing inputs for
stability analyses”;
• “Determining the safety factor under static and/or pseudo-static conditions”; or
• “Hydro-mechanical analyses to address the complex behaviour of tailings dams during
staged construction”.
This research focuses on seepage analysis for the intent of objective #1 above: understanding
the potential impact of external conditions on broad pore water pressure conditions, combined
with appropriate interrogation of the piezometer readings to better understand the
performance of any particular tailings dam. This not only enhances the ability to proactively
respond to unfavourable changes in the dam, but also informs practitioners as technology
progresses to “real-time”. Understanding that the measurement may be in real-time, but that
the condition being measured could be a delayed response to pore water pressure changes
elsewhere, is important when considering a true understanding of dam performance.
There are a multitude of parameters that might influence the flow of water in a tailings dam
environment, including those characteristic of the tailings deposit, as well as external
conditions that are a function of the climate, location, and operation of the tailings dam
(among others). Understanding the phreatic surface fluctuation and overall level in the tailings
dam is pertinent to guide operational responses to maintain the pore water levels within
149
acceptable performance limits and to take appropriate action against established trigger levels.
To further understand the effect of different tailings dam environments and external
conditions on piezometer response and reliability in terms of understanding tailings dam
phreatic conditions, this research undertakes sensitivity tests on influential parameters to
better understand their effects. This paper focuses on their effects on upstream construction.
The influential parameters assessed align with Vick’s (1990) description of the factors that
have the most significant effect on the phreatic surface location, as (in order of prevalence, for
upstream construction):
• Beach width, describing the location of ponded water with respect to the embankment
crest;
• Lateral permeability variation, as produced by grain-size segregation of the tailings
deposit during hydraulic placement; and
• Boundary flow conditions, described by starter dam permeability and foundation
permeability.
For this paper, the two most influential factors on the upstream embankment phreatic surface
are suggested to be the beach width and lateral permeability variation. These are modelled
together with the introduction of an external load, which is introduced in an attempt to
simulate objective #3 above in understanding the effect of staged construction on phreatic
conditions.
6.2.1. Hydraulic Conductivity
The first parameter focused on in this research is tailings material composition, and its
characteristic hydraulic conductivity. This parameter influences each of the modelled
scenarios when considering the naturally descending piezometer level, and specifically when
modifying the anisotropic ratio (ratio of hydraulic conductivity in the vertical direction versus
the horizontal). Hydraulic conductivity, 𝑘, determines the ability for water to flow through a
porous media: “Darcy’s Law states that the amount of water (Q) flowing through porous
media depends on the energy driving the water flow (Δh/ΔL) and the hydraulic conductivity
(𝑘) of the porous media” (Mitchell n.d.). Darcy’s Law is therefore stated as:
150
𝑄 𝑑ℎ
𝑞= = −𝑘
𝐴 𝑑𝑙
where:
𝑑ℎ
= hydraulic gradient, also referred to as the symbol 𝑖
𝑑𝑙
𝑝
ℎ = hydraulic head = +𝑧
𝜌𝑔
𝜌 = density of liquid
“More than any other engineering property of tailings, permeability [/hydraulic conductivity]
is difficult to generalise. Average permeability spans five or more orders of magnitude, from
10-2 cm/s for clean, coarse sand tailings to as low as 10-7 cm/s for well-consolidated slimes”
(Vick 1990). With tailings material often described relative to soil types, estimates of
hydraulic conductivity ranges for different generic soils compared against typical tailings
material ranges are presented in Table 28 and Figure 30.
Table 28 Typical soil hydraulic conductivity values (adapted from Domenico & Schwartz 1990 and Vick 1990)
151
Figure 30 Chart showing typical hydraulic conductivity of different soils (adapted from Domenico & Schwartz
1990)
Hydraulic conductivity should be determined through the appropriate laboratory tests on each
site’s unique tailings composition (influenced by processing method, resultant grain size,
moisture content, degree of compaction, mineralogy, and external factors). There are,
however, correlations documented in literature and traditionally accepted as reasonable
estimations of hydraulic conductivity based on different material parameters. These are
summarised in Table 29.
Table 29 Hydraulic conductivity correlations from literature (adapted from AQTESOLV 2019)
Year
(incl. Application
Name Formula1 Reference
revision Considerations1
dates)
• CH is most commonly
2 100 but published
𝑘𝐻 = 𝐶𝐻 𝐷10 Hazen 1892
1892, values range between 1
2
Hazen Formula to 1000 cm-1s-1 Hazen 1911
1911
(Length units in cm) • 0.1mm ≤ 𝐷10 ≤ 3mm
• 𝐶𝑈 ≤ 5
𝑔 𝑛3 Freeze and
1927, 𝑘𝐾𝐶 = 𝐶𝐾𝐶 𝐷2 • Valid for sediments and
Kozeny-Carmen 𝜐 (1 − 𝑛)2 10 Cherry 1979
1937, soils composed of silt,
Equation sand, and gravelly sand Rosas et al.
1956
(Length units in m) 2014
152
𝑔 500 2
𝑘𝐵 = 𝐶𝐵 (ln ) 𝐷10
𝜐 𝐷
(𝐷60 ) • 0.06mm ≤ 𝐷10 ≤ 0.6mm Beyer 1964
Beyer Formula 1964 10 • 1 ≤ 𝐶𝑈 ≤ 20 Kresic 1967
(Length units in m)
“The theoretical position of the steady-state phreatic surface in any homogeneous dam is
independent of the actual value of permeability [/hydraulic conductivity] and is dependent
only on the physical dimensions and the ratio of horizontal to vertical permeabilities”
(Cowherd, Miller, & Perlea 1993). It is noted that the modelled dam scenario in this research
is inhomogeneous, hence both factors are considered. The ratio of horizontal to vertical
permeabilities is defined as the anisotropy ratio (𝑘ℎ /𝑘𝑣 ). “Because of their layered nature,
tailings deposits exhibit considerable variation in permeability between the horizontal and
153
vertical directions” (Vick 1990). Vick (1990) continues, describing:
• 𝑘ℎ /𝑘𝑣 is “generally in the range of 2-10 for reasonably uniform beach sand deposits
and for underwater-deposited slimes zones”;
• “Transition beach zones between areas of relatively clean sands and slimes are likely
to have higher anisotropy ratios due to interlayering of finer and coarser particles”;
• Extensive sand-slime interlayering can see anisotropy ratios of greater than or equal to
100.
Naeini and Akhtarpour (2018) consolidate this research alongside more recent findings to
suggest an anisotropy ratio of 1.25 for sand tailings and 10 for slimes.
6.2.2. Piezometer Type
The two most common piezometer types are standpipe and vibrating wire. It is suggested that
these are the most common types of piezometer due to their simplicity (and hence cost
effectiveness), and suitability for a variety of applications, respectively.
Clarkson et al. (2020) describe the details of each piezometer as extracted in Table 30.
Table 30 Standpipe and vibrating wire piezometer instrumentation details (Clarkson et al. 2020)
154
• Can be subjected to rising or falling head • Very short lag time (when installed
tests to confirm function. using fully grouted method).
• Easily automated and still allow for water • Elevation of readout independent of
sampling. elevation of tips and piezometric levels.
• Used in soil and rock formations where the • Low temperature sensitivity, however,
time lag and high displacement requirements thermistor included for temperature
inherent in standpipes are not crucial, and compensation.
where the presence of standpipes will not • Frequency output signal permits
hinder construction. transmission over long distances.
• Can measure small sub-atmospheric pore
water pressures.
• Output signal independent of length of
electrical cable (signal is frequency
based).
• Can be automated with continuous
readings when connected to datalogger,
with SIM card for remote data recovery.
Limitations • Low accuracy and long lag time in • Lightning protection required.
impervious soils. • Expensive transducer and readout.
• Potential freezing problems if water near • Sensitive to temperature and barometric
surface. pressure changes.
• Porous tips can clog due to repeated inflow • Risk of zero drift (hence not appropriate
and outflow. for long-term
• Not appropriate for artesian conditions where monitoring), but some models available
phreatic surface extends significantly above with in-situ calibration check.
top of pipe. • No independent means of recording
• Interferes with material placement and position of sensor (care required for
compaction during construction. settlements).
• Can be damaged by consolidation of soil • Impossible to de-air devices that have a
around standpipe. closed reservoir, meaning the
• Readings can be affected by rainwater runoff measurement will be incorrect.
and barometric pressure. • Post data processing calculations can be
complicated if not automated.
• Wire cables can be damaged by
construction, settlement, or corrosion.
Figure 31 Cross-section of standpipe and vibrating wire piezometer installation components, respectively, from
Clarkson et al. (2020)
It is noted that an open standpipe piezometer comprises the same installation components as
the standpipe piezometer in Figure 31, however the casing remains open (no piezometer is
installed) to allow downhole standing water level measurements.
156
6.2.4. External Conditions
Vick (1990) describes the factors that have the most significant effect on the phreatic surface
location as (in order of prevalence):
For Upstream Construction:
• Beach width (operationally controlled), describing the location of ponded water with
respect to the embankment crest;
• Lateral permeability variation (operationally controlled), as produced by grain-size
segregation of the spigotted beach tailings and as measured between spigot point and
beach; and
• Boundary flow conditions (somewhat designer controlled), described by starter dike
permeability and foundation permeability.
For Downstream Construction:
• Boundary conditions as per upstream embankments, but with particular focus on the
ratio between the permeability of the embankment (including core) and the tailings
deposit;
• Anisotropy (operationally and designer controlled), produced by moderate degrees of
sand-slime interlayering; and
• Cyclone operation and construction procedure (operationally controlled), which may
introduce large quantities of water into localised portions of the embankment.
For Centreline Construction:
• Boundary conditions as per downstream embankments;
• Anisotropy as per upstream embankments and focusing on the permeability ratio
between the shell and core.
These theoretical concepts are presented in Table 31.
For this paper, which focuses on upstream construction, the two most influential factors on
the upstream embankment phreatic surface are suggested to be the beach width and lateral
permeability variation.
157
Table 31 Theoretical concepts of different external conditions on phreatic surface (adapted from Vick 1990)
Homogeneous, anisotropic
upstream embankment on
Beach Width
an impermeable
foundation.
Lateral
Nonhomogeneous
Permeability
upstream embankments
Variation
Homogeneous upstream
embankment on
impervious foundation
Anisotropy
Non-homogeneous
upstream embankment on
impervious foundation
158
6.2.5. Historical Piezometer Response Calculations
“The term slow response time is used to describe a long hydrodynamic time lag” (Dunnicliff
1993, p.140), representing the delay between total stress change in the environment, and the
delayed pore pressure measurement “since pore-water must enter the instrument” (Gibson
1963). The overall response should consider both the instrument-specific lag as well as the
time taken for pore pressure changes to transmit through the tailings material from some
arbitrary distance until it reaches the piezometer location. This section of this paper addresses
the instrument-specific lag.
“A preliminary estimate of the time lag is necessary for the design or selection of the proper
type of installation for given conditions” (Hvorslev 1951). Terzaghi and Peck (1967) describe
approximate response times for different types of piezometer, as seen in Figure 32. The 90%
response time is considered sufficient for practical application.
Figure 32 Chart showing approximate response times for various types of piezometer in homogeneous soils
(adapted from Terzaghi and Peck 1967). * with tubing 8ft long, ** without sand filter surrounding point.
159
6.2.5.1. Standpipe Piezometer
The critical factors related to instrument-specific lag are described in theory for standpipe
piezometers as:
• “The magnitude of the time lag depends on the type and dimensions of the pressure
measuring installation, and it is inversely proportional to the permeability of the soil”
(Hvorslev 1951);
• Volumetric flexibility, which can arise from two main causes (Gibson 1963):
• Compressibility of the fluid filling the device due to the presence of entrapped air or
natural gases;
• The volume expansibility of the device itself (of which Gibson 1963 assesses the
time-lag); and
• Compressibility/incompressibility of the soil/tailings type (Gibson 1963).
Penman (1960) presents an equation for estimating the response time of open standpipe
piezometers as derived from experimental data, presented as:
𝐿 𝐿
𝑑 2 ln (𝐷 + √1 + (𝐷)2 )
𝑡 = 3.3 × 10−6
𝑘𝐿
Where:
𝑡 = time required for 90% response, in days;
𝑑 = inside diameter of standpipe (cm);
𝐿 = length of intake filter (or sand zone around the filter, cm);
𝐷 = diameter of intake filter (or sand zone, cm);
𝑘 = permeability of soil (cm/sec)
On the theoretical solutions, the Highways Agency (2003) describes that “the theory for the
response of a piezometer was developed by Hvorslev (1951) for an incompressible soil, and
by Gibson (1963) for a compressible elastic soil obeying Terzaghi’s theory of consolidation”.
These theories are described below. It is noted that the equations are in terms of a parameter
nominated as the “Equalisation Ratio”. The equalisation ratio provides an indication of the
error between actual and measured pore pressure conditions (a lower equalisation ratio
reflects a higher error bar).
160
For a Standpipe Piezometer in a Compressible Elastic Soil (Hvorslev 1951)
Hvorslev’s hydrostatic response time adopts key assumptions on the soil surrounding the
piezometer, in that it is:
• Isotropic;
• Fully saturated;
• Infinite in extent;
• Incompressible; and
• No head losses in the system.
𝜇𝑔 − 𝜇𝑡 𝐹𝑘𝑡
𝜀= = exp (− )
𝜇𝑔 − 𝜇0 𝑉𝛾𝑤
where:
𝜇𝑔 = pore pressure in the soil;
𝜇𝑡 = cavity pressure at time t;
𝜇0 = initial cavity pressure (at time 0);
𝐹 = intake factor of the piezometer (cm, see below);
𝑘 = coefficient of permeability of the soil (cm/s);
𝑉 = volume factor of the piezometer system, defined as the volume flow into the piezometer
for unit pressure change in the cavity (cm3); and
𝛾𝑤 = unit weight of water (g/cm3)
However, as field piezometers are not spherical, an intake factor (F) was derived for other
geometries (such as a cylinder) as:
2𝜋𝐿
𝐹=
𝐿 𝐿 2
log 𝑒 (𝑑 + √1 + (𝑑 ) )
where:
𝐿 = length of the effective intake or well point (cm); and
161
𝑑 = diameter of the piezometer (cm).
For a Standpipe Piezometer in a Compressible Elastic Soil (Gibson 1963)
For derivation of the solution, the reader is referred to Gibson (1963). The equation presented
by Gibson (1963) considers a spherical porous element in a saturated homogeneous and
isotropic material. This equation follows Terzaghi’s consolidation theory. The function for
equalisation ratio is presented as (Gibson 1963):
1
𝜀= (𝜂 exp(𝜂12 𝑇) . erfc(𝜂1 𝑇 0.5 ) − 𝜂2 exp(𝜂22 𝑇) . erfc(𝜂2 𝑇 0.5 ))
𝜂1 − 𝜂2 1
where:
𝜂1 1
𝜂2 } = {𝜇 ± (𝜇2 − 4𝜇)0.5 }
2
and
4𝜋𝑎 3 𝑚𝛾𝑤
𝜇 = stiffness of the measuring system = 𝐴
𝑎 = piezometer radius
𝑚 = coefficient of expansibility or compressibility of the soil (assumed to be equal in
compression and expansion)
𝐴 = pore pressure coefficient = 𝑉𝛾𝑤
𝑐𝑡
𝑇 = time factor= 𝑎2
The complementary error function, erfc(𝜂1&2 𝑇 0.5 ), is defined in broad literature (Weisstein
n.d.) as:
∞
2
𝑒𝑟𝑓𝑐 (𝜂1&2 𝑇 0.5 ) = ∫ exp(𝑡 2 ) 𝑑𝑡
√𝜋 𝜂1&2 𝑇 0.5
162
𝐹 = 4𝜋𝑎
which rearranges to:
𝐹
𝑎=
4𝜋
Substituting back into Gibson’s equation for a standpipe piezometer in an incompressible soil
and expanding A presents Hvorslev’s (1951) equation:
𝐹𝑘𝑡
𝜀 = exp (− )
𝑉𝛾𝑤
It is suggested that “at high degrees of equalisation, the two theories give similar results”
(Walbancke 1975). Gibson (1963) suggests that plotting the equalisation ratio, 𝜀, against
either 𝜇𝑇 (which includes 𝜇 = 0) or 𝜇2 𝑇 (which includes 𝜇 = ∞) can present numerical
results graphically and allow results of an equalisation test to be analysed by superimposing
the results onto the charts. These charts are presented in Figure 33 and Figure 34,
respectively.
Figure 33 Chart showing variation of degree of equalisation with 𝝁𝑻, representing pore water pressure
measurement in situ (adapted from Gibson 1963)
163
Figure 34 Chart showing variation of degree of equalisation with 𝝁𝟐 𝑻, representing pore water pressure
measurement insitu (adapted from Gibson 1963)
164
explore their influence on seepage through the tailings material and subsequent influence on
pore pressure conditions at the piezometer location.
6.3.1. Calculation Scenario
Penman’s (1960) empirical equation is presented in terms of the scenario presented:
𝐿 𝐿
𝑑 2 ln (𝐷 + √1 + (𝐷)2 )
𝑡 = 3.3 × 10−6
𝑘𝐿
Where:
𝑡 = time required for 90% response, in days;
𝑑 = inside diameter of standpipe (m) = 0.016m;
𝐿 = length of intake filter (or sand zone around the filter, m) = 0.275m;
𝐷 = diameter of intake filter (or sand zone, m) = 0.0335m;
𝑘 = permeability of soil (m/sec) = 𝑉𝑎𝑟𝑖𝑎𝑏𝑙𝑒.
The piezometer dimensions are based on RST Instruments’ Casagrande Standpipe Piezometer
(RST Instruments n.d).
6.3.2. Numerical Modelling
The numerical modelling scenarios simulated hydrodynamic time lag and change in total
elevation head as a result of the time taken for pore water pressure changes to transmit
through the tailings. Different external conditions were assessed using numerical simulations
to explore their influence on seepage through the tailings and subsequent influence on pore
water pressure conditions at discrete locations, including:
• Pond level rise and beach width change. “The location of the ponded water with
respect to the embankment crest, or the width of the exposed tailings beach, is often the
most important factor influencing phreatic surface location” (Vick 1990);
• Lateral variation of hydraulic conductivity. “The variation in permeability is
characterised by the ratio of tailings permeability at the spigot point (k0) to the
permeability at the edge of the ponded water at the slimes zone (kL)” (Vick 1990); and
• External loading. While a potential function of many different influences, the excess
pore water pressure generated as a result of a theoretical load increase was checked.
6.3.2.1. Model Theory
The numerical model was generated using proprietary software, SEEP/W (GEO-SLOPE
2019). SEEP/W adopts a number of assumptions and approaches whereby expressing
165
understanding of some basic definitions in terms of the hydraulic conductivity aids
understanding of this research (US EPA 2010):
• “Homogeneous means that hydraulic conductivity (k) (or the coefficient of
permeability) in the material (natural soil or the embankment) is independent of
position”;
• “Isotropic means that hydraulic conductivity is independent of direction at the point of
measurement”;
• “If hydraulic conductivity is dependent on position then the media is heterogenous”;
and
• “If hydraulic conductivity of a media is dependent on direction at the point of
measurement then the media is anisotropic”.
This research assumes homogeneous materials; however, a balance between isotropic and
anisotropic materials is used, as described further in Section 6.3.2.2.3.
When estimating seepage direction, volume, and pore pressure through these materials,
SEEP/W computes Laplace’s equation which “represents the steady state of a field that
depends on two or more independent variables, which are typically spatial” (Pierce 2019).
The form of the equation is described as (GEOSLOPE 2019):
∂2 ℎ ∂2 ℎ ∂2 ℎ
+ + =0
∂x 2 ∂y 2 ∂z 2
SEEP/W presents this equation graphically through the use of flow nets. Laplace’s equation
describes that the flow lines must cross the equipotential lines (contours of total head) at right
angles, and the area between two adjacent flow lines “is called a flow channel and the flow in
each channel has to carry the same amount of flow” (GEOSLOPE 2019). While SEEP/W
does not create a “true” flow net, it does compute elements of flow behaviour aligned with
flow net principles to allow practical interpretation of seepage conditions. This also helps to
provide the ability to model transient flows, in which scenario flow nets are “nearly
impossible to construct” (GEOSLOPE 2019).
“In a saturated-unsaturated seepage formulation such as SEEP/W, the phreatic surface is not a
true flow line as in the context of a flow net. The phreatic surface is merely a line of zero
pressure” (GEOSLOPE 2019). Flow paths can be drawn above the phreatic surface within the
capillary rise zone, however extending too far into the unsaturated zone presents unrealistic
results. The phreatic surface represents a line of zero pressure however it is not modelled as a
distinctive zero pressure, upper boundary condition (as in traditional methods for unconfined
166
flow modelling). Hence, modern-day numerical modelling (such as SEEP/W) are suggested to
provide a more complete picture with pressure distributions and flow having the potential to
extend above the phreatic surface.
6.3.2.2. Model Setup
6.3.2.2.1. Model Geometry
The numerical model was developed based on existing conditions and dam arrangements at an
unnamed mine site in Australia. The tailings dam is constructed using the upstream method
and has been modelled in stages describing construction and deposition of the progressive
lifts. The progression of dam construction and deposition analysed is presented conceptually
in Table 32. The materials and associated colour coding used in the dam are listed in Table
33.
The initial pore water pressure conditions were defined by a steady-state condition and
dependent on the initial water table, which was modelled based on an understanding of
groundwater and tailings water level characteristics at this particular site. This initial regime
was used as the ‘parent’ analysis in the modelling approach. Each of the following stages
represent a ‘child’ analysis to the stage prior, modelled as transient seepage cases, but
dependent on the ‘parent’ analysis for initial pore water pressure conditions. The main
reasoning behind the selection of a transient seepage condition (while the authors are aware of
the limitations and sensitivity of material model parameters in this analysis type), was to
allow a time-stepped development of the phreatic surface, which in turn could be compared
against measured piezometric readings from site operations (see Section 6.3.2.2.2).
End
Stage 6 Deposition
(September 2019 to May 2020)
167
6.3.2.2.2. Model Calibration
“It is common for the rate of seepage through a tailings embankment to be estimated on the
basis of a conventional flow net analysis assuming steady seepage conditions” (Mittal and
Morgenstern 1976). However, as Stark et al. (2015) state, “there is interest in performing
transient unsaturated seepage analyses to calibrate transient seepage models with piezometric
data and investigate the level of conservatism with a design based on steady-state conditions.”
In the case of this research, the primary reason for utilising transient seepage analysis was to
leverage nearly 10 years’ worth of piezometric data and understanding. This was further
supplemented by observations during the same period, to develop a model more reflective of
actual conditions. This research focuses on transient seepage conditions as a result of
boundary conditions, as opposed to a comparison between transient seepage and steady-state
seepage behaviour.
The steps taken to calibrate the SEEP/W model against actual piezometer data were:
1. The push-in vibrating wire piezometer readings collected from site were embedded as a
“Water Total Head” Boundary Condition. This represented the baseline ‘target’ for
actual conditions:
a. Readings have been collected approximately monthly since 2011;
b. Individual ‘Water Total Head’ boundary condition functions were created for each
piezometer to model the piezometer readings over time;
c. This was then applied at the different stages of the tailings dam construction and
deposition (when the piezometers were brought online), aligning well with the
time-based transient seepage approach; and
d. The piezometer locations are identified in Figure 35.
2. The model was cloned, and the ‘Water Total Head’ boundary conditions were
removed. The material parameters were altered (hydraulic conductivity and anisotropy
ratio, using a saturated material model) to simulate the baseline model. As the transient,
saturated/unsaturated material models present a relatively complex visualisation of
phreatic behaviour, taking an initial assumption of full saturation allowed a single
phreatic surface to be the focus. This step is predominantly to ensure that the phreatic
surface is appropriately trending through different materials, in terms of relative
hydraulic conductivity and anisotropy ratios between the different material types. To
test the sensitivity of the model simulating actual conditions and response to changing
environments, the forced boundary conditions at the discrete piezometer locations
168
would need to be removed or they would bias the phreatic surface.
3. Once the phreatic surface behaved as closely as possible to that generated from actual
piezometer data, the material models were converted to saturated/unsaturated (suited to
transient seepage analysis):
a. In this research, the SEEP/W model adopted the software’s built-in estimation
methods as described in Material Parameters, using the same hydraulic
conductivity as derived for the different materials during Step 2; and
b. Modified boundary condition assumptions so that water total head outputs match
piezometer readings.
4. The sensitivity of the model was tested through application of different External
Condition Scenarios (Section 6.3.2.4).
5. By relying on the observational approach, this methodology presented a higher degree
of confidence in the model and allowed the opportunity to then apply the external
condition scenarios and simulate differences in response.
169
Table 33 Material parameters adopted in SEEP/W model
171
6.3.2.4. External Condition Scenarios
Three external condition scenarios were modelled to assess their influence on phreatic surface behaviour. These are described in Table 34.
Table 34 External condition scenarios
Pond level “The location of the The “water total head” boundary condition line that
rise and ponded water with modelled the decant water level in the dam was
beach width respect to the changed from being located 75 m from the
change. embankment crest, or the embankment upstream toe, to being located 37.5 m
width of the exposed from the beach-embankment intercept. The decant
tailings beach, is often level was raised by 2 m to simulate pond level rise.
the most important factor
influencing phreatic - Decant water level boundary condition (to right hand side of model) elevated
surface location” (Vick from Calibrated Model.
- Beach width reduced by 50% compared to Calibrated Model.
1990).
- Material parameters equal to Calibrated Model.
172
Lateral “The variation in Material changes were made to the different tailings
variation of permeability is Ky’/Kx’ ratio (also known as the anisotropic ratio).
hydraulic characterised by the ratio The sensitivity of this was checked by changing the
conductivity. of tailings permeability anisotropic ratio to 0.1 for all tailings materials.
at the spigot point (k0) to
the permeability at the
edge of the ponded water
- Decant water level boundary condition (to right hand side of model) equal to
at the slimes zone (kL)” Calibrated Model.
(Vick 1990). - Beach width equal to Calibrated Model.
- Material parameters changed to decrease anisotropic ratio.
173
6.3.2.5. Assumptions
Several key assumptions were adopted due to the limitations of a numerical model to simulate
reality or in order to be able to model the scenarios in the most practical way possible. These
included, but were not limited to:
• Instrumentation remained fully intact and operable, despite changes in external conditions;
• The piezometers modelled assume perfect installation, with no anomalies as a result of
installation methodology;
• Materials incorporated were considered homogeneous;
• Two-dimensional behaviour was adopted. No consideration was given to any three-
dimensional influence, including the likelihood for induced conditions to dissipate laterally
or take a different path of least resistance;
• The materials and phreatic conditions were developed based on the deposition and
construction phases; however, this development did not allow for modelling of progressive
filling of tailings material aside from that captured by the decant pond level rise. This may
have some influence on the magnitude of change that is observed at the transition between
deposition and construction phases, however considering the phreatic surface being strongly
influenced by the decant pond level (which was modelled as a hydraulic boundary, water
total head against time), the results are considered representative for the purpose of this
research; and
• Transient flow conditions were adopted in the numerical models, based on typical flow
conditions at tailings dam facilities. Steady state conditions are still considered relevant
(albeit conservative when determining pore water pressure) and are applicable in line with
Darcy’s equation. Vick (1990) recommends consideration of “non-steady, transient, or
unsaturated flow that occurs under capillary rather than gravity gradients” for the purpose of
determining impoundment seepage loss.
6.4. Results
The response time derived through the Penman (1960) approach (for standpipe piezometers only),
as well as the effect of different external conditions on piezometer response and reliability as
assessed through numerical modelling are presented.
6.4.1. Calculation Scenario
From the calculation and parameters described in Materials and Method, the Penman (1960)
relationship presents the empirical response time for standpipe piezometers as:
174
𝐿 𝐿
𝑑 2 ln (𝐷 + √1 + (𝐷)2 )
𝑡 = 3.3 × 10−6
𝑘𝐿
Where:
𝑡 = time required for 90% response, in days;
𝑑 = inside diameter of standpipe (m) = 0.016m;
𝐿 = length of intake filter (or sand zone around the filter, m) = 0.275m;
𝐷 = diameter of intake filter (or sand zone, m) = 0.0335m;
𝑘 = permeability of soil (m/sec) = 𝑉𝑎𝑟𝑖𝑎𝑏𝑙𝑒. This is presented as variable to align with the
Numerical Modelling approach and present indicative results for different tailings types.
0.275 0.275
0.0162 ln (0.0335 + √1 + (0.0335)2 )
𝑡 = 3.3 × 10−6
𝑘 × 0.275
−6
7.17 × 10−4
𝑡 = 3.3 × 10
0.275𝑘
−6
2.61 × 10−3
𝑡 = 3.3 × 10 ×
𝑘
8.61×10−9
𝑡= days
𝑘
By substituting various tailings hydraulic conductivity values, 𝑘, the resultant equation derives an
estimate of response time as shown in Table 35.
Table 35 Instrument response time for scenario presented (based on Penman 1960)
Tailings Hydraulic Conductivity Instrument Response Time (in days, from Instrument Response Time (in hours, from
(𝒌, m/s) Penman 1960) Penman 1960)
175
6.4.2. Numerical Modelling
The quantitative change as a result of the External Condition Scenarios was measured at the
locations of the piezometers, as described in Figure 31. These represented:
• 3A: Pore water pressure in the foundation below the initial dam;
• 3B: Pore water pressure at the base of the initial dam embankment;
• 3C: Pore water pressure within the tailings beneath the first raise; and
• 3G: Pore water pressure in the tailings below the third raise.
Within the transient analysis, particularly when considering the saturated/unsaturated material
model, the phreatic surface is much more variable than that presented in a steady-state seepage
model. Hence, the calibration focused on reaching as near as practicable the total head elevation at
different piezometer points. The results of the calibration, showing total head elevation against time,
are shown for Piezometer 3C and Piezometer 3G in Figure 36 and Figure 37, respectively. These
piezometers were selected as representative based on the quality of their installation and the
reliability of the data from onsite observations.
The ‘actual piezometer data’ plot is generated by the piezometer readings onsite, ‘calibrated’ shows
the modelled seepage profile without the piezometer-enforced boundary conditions, while the ‘pond
level rise’, ‘lateral variation’ and ‘external load’ reflect the different external condition scenarios.
It is important to recognise the limitations of the ‘calibrated’ model in reflecting the measured
conditions onsite. Evidently, the fluctuations in onsite data would be near-impossible to model
perfectly in a theoretical scenario, which assumes consistent conditions as described in Section
6.3.2.5. Hence, the difference between the ‘actual piezometer data’ and ‘calibrated’ conditions is
suggested to be due to actual external conditions that could not be appropriately
considered/accounted for in the calibrated model.
176
Piezometer 3C (Figure 36) showed a generally decreasing trend over time. This is largely attributed
to a drain that was installed between the Stage 1 and Stage 2 embankments (1222 m Australian
Height Datum (AHD)). This drain acted as a “sink” and was modelled to simulate discrete
drainpipes installed through the clay liner for control of seepage water by lowering the phreatic
surface. As the total head elevation increases, it is expected that the pore pressures recharging from
the decant pond would dissipate through this drain more readily (due to the total head difference), in
turn seeing the total head elevation readings converge over time toward the approximate elevation
of the drain and with relatively less influence from external conditions.
Piezometer 3C did not exhibit a similar, distinct response to changes in decant water level, nor
changes between construction (despite buttress placement above 3C in late 2014 and again in late
2015) and three deposition phases when compared against Piezometer 3G (Figure 37).
Piezometer 3G (Figure 37), showed the greatest effect from introduction of external condition
scenarios, anticipated due to the proximity to the decant pond, greatest influence of construction
loading and subsequent porewater pressure increase, and also the reliability of the installation
method and condition of this piezometer to represent actual conditions. The data presented some
observable trends:
177
• The magnitude of variance between the actual piezometer readings and the calibrated model
observed:
• ‘Lateral hydraulic variation’ indicated a maximum total head difference of 2.9 m
(06/03/2018);
• ‘Introduction of an external load’ presented a maximum total head difference of 2.5 m
(15/12/2016); and
• ‘Pond level rise’ presented a maximum total head difference of 3.5 m (16/10/2017).
• A distinct inflexion point could be observed between a deposition period and a construction
period. This is related to the assumption (Section 6.3.2.5) within the model that does not
allow for progressive filling of tailings. In reality, it would be expected that this change
would still occur in the phases that it does, however it would be expected gradually over time
as related to the gradual increase in pore pressure, with a peak at time of construction of the
embankment construction in this local area (for Piezometer 3G, this could be suggested to be
around 16/09/2016);
• At the earlier and lower elevation stages of the tailings dam construction and deposition, both
the actual piezometer data and modelled trends were observed to reflect the changes in
decant water level as tailings deposition typically occurred directly above the piezometers
and hence the closer proximity was anticipated to facilitate connectivity between the decant
pond and piezometer locations. However, in later deposition stages upstream, the data did not
reflect changes as readily and it could be suggested that the piezometer level was a function
of the naturally descending phreatic surface due to lateral separation of the decant pond
(recharging point) and piezometer location (discussed further in Section 6.5);
• The behaviour described in the previous point was exacerbated by the boundary conditions
modelled for ‘pond level rise’ (reduced beach width and hence increased connectivity
between decant recharge and piezometer response);
• The ‘introduction of external load’ and ‘pond level rise’ external conditions increased the
total head elevation and shortened the response time;
• Comparing the pore pressure values at different piezometer locations for the ‘introduction of
an external load’ at Stage 5 construction, it was observed that the pore pressure divergence
from the ‘calibrated’ condition was observed early in the stage for Piezometer 3G
(23/10/2015), but was delayed from this for Piezometer 3C (20/01/2016): a three month
difference; and
• The change in ‘lateral hydraulic variation’ external condition reduced the total head elevation
and showed a less significant response to changes in deposition/construction cycles.
178
The overall magnitude of the change in total head elevation was generally achieved, and the
modelled response was within 3.7 m (inclusive of data spikes) or 2.2 m (not inclusive of data
spikes), when comparing onsite piezometer readings against elevation levels generated within the
calibrated model.
6.5. Discussion
The calculated scenario suggests that, for the tailings material implemented into the modelled case
(1 × 10−5 𝑚𝑠 −1 and 1 × 10−9 𝑚𝑠 −1 ), the piezometer response time for a standpipe piezometer
would be well within a single day (1 minute to 3.5 hours, respectively). The vibrating wire
piezometer, as discussed in Section 6.2.5, was suggested to be “within 5 minutes” of a pore pressure
change, and is not anticipated to significantly influence the instrument-specific response time at the
piezometer location.
The numerical modelling calibrated total head elevations against push-in vibrating wire piezometers
readings, then applied external conditions to assess the effect on piezometer response and
reliability. In summary, the findings showed that:
• While pore pressures in Piezometer 3C were influenced by the drain between Stage 1 and
Stage 2 embankments, Piezometer 3G was observed to present the readings most influenced
by the decant water level. This influence was more prevalent in the earlier stages of the life
cycle (when the piezometer was closer to the decant pond) compared to the later stages of the
life cycle. This is suggested to indicate a few key points:
• For large dams, piezometers that are installed at the earlier stages of construction are
more readily influenced by the decant pond level and the column of saturated tailings that
was recently placed above their location;
• As the dam raises, and the decant pond (or recharge point) and tailings placement become
further from the piezometer tip location, the time taken for the piezometer to respond to a
change in phreatic condition as well as the resulting behaviour are less representative of
the change itself;
• This suggests that variables such as material parameters, boundary conditions, anisotropy
ratios, and transient flow conditions should be considered in order the understand the
overall phreatic condition of the dam as these factors complicate the phreatic behaviour
and the resultant pore pressures when compared against simplified steady-state
modelling;
• While the tailings dam and water balance were generally managed well at the case mine site,
the hypothetical pond level rise and introduction of an external load increased the total head
elevations. Piezometer 3G’s response suggested that pond level rise caused the largest rise in
179
total head elevation . This result aligns with Vick’s (1990) theory, which stated that pond
level rise “is often the most important factor influencing phreatic surface location”. The
hypothetical ‘lateral variation in hydraulic conductivity’, which reduced the anisotropic ratio,
reflected a reduction in pore pressure values;
• The different external conditions reflect different total head elevation behaviours;
• Observing the Stage 5 embankment construction as an individual event, a difference in
response time between piezometer locations was observed. Stage 5 construction phase was
modelled to simulate a 2m rise in pore pressures. The pore pressure divergence from
‘calibrated’ conditions occurred 3 months earlier for the piezometer it was constructed
directly on top of (Piezometer 3G), than for a piezometer 50m downstream and with its tip
installed at an elevation lower by 4m (Piezometer 3C). In addition, while both trends started
to re-converge with the ‘calibrated’ conditions following the event, neither piezometer
suggested that excess pore pressures had yet dissipated from the event at the end of available
readings (4.5 years later). This suggests that a change in phreatic conditions upstream is not
immediately reflected downstream, and should be considered in terms of a delayed and
prolonged response as the effect transitions through the dam; and
• The fluctuations in total head elevation measured by the piezometers during the early stages
of dam deposition/construction (start of measurements to 25/01/2017) were best simulated by
the pond level rise external condition. Between 25/01/2017 and end of measurements, the
calibrated pore water pressure appeared to best simulate the total head elevation measured
from the piezometers. This was suggested to be because, after 2017, the external condition
changes occurring upstream of the piezometers to be largely influenced by the naturally
descending phreatic surface and the installed drains, in this particular dam.
Depending on the amount and duration of data available, which can be obtained from existing and
new tailings dams, the ability to iteratively update the seepage (and in turn stability) models to
assess tailings dam performance is fundamentally achievable. It is anticipated that designers and
operators could gain value in developing seepage models that are forced to match the piezometer
readings (by utilising boundary conditions), and by comparing the phreatic surface profile with that
assumed in the initial and staged designs.
It is not anticipated that this approach would be feasible without having accurate monitoring data to
rely on. Hence, the authors recommend the use of steady-state seepage as an appropriate approach
during tailings dam design phases. The risk associated with variable tailings dam environments, the
inability to predict this prior to the dam being built and hence the assumptions made in modelling
that contradict this variability (such as assuming homogeneous units), and the implications that this
may have for the true seepage behaviour, is not considered to be a safe nor appropriate approach.
180
It is suggested that future modelling could be improved through better understanding of the initial
conditions. Ideally, the dam site should be modelled over its entire lifecycle (considering historical
land use and pre-mining state), to create a holistic model of the dam’s lifecycle. For this research;
however, the objective was mainly to observe relative change and hence this was not explored.
6.6. Conclusion
This research demonstrates a valid approach to utilising the observational approach, in combination
with a calibrated numerical model, to improve understanding of unique tailings dam responses to
external conditions. In the case presented, it was found that the response time of a standpipe
piezometer would be approximately 1 minute to 3.5 hours, based on the Penman (1960) approach.
This has the potential to range between one second to 860 days for different tailings hydraulic
conductivity values (1 × 10−3 𝑚𝑠 −1 to 1 × 10−11 𝑚𝑠 −1 , respectively). The phreatic behaviour of a
tailings dam in Australia was numerically modelled based on the measurements recorded from a
cross-section instrumented with push-in vibrating wire piezometers. By calibrating the model
against the behaviour observed from the piezometer readings, the research allowed the sensitivity of
external conditions on the tailings dam to be assessed. It was observed that external conditions more
readily influenced later stage construction and deposition. For the case study undertaken, this was
suggested to be as a result of the installation of embankment toe drains between the Stage 1 and
Stage 2 embankments (at the end of Stage 4 construction), as well as the proximity of piezometer
locations to the decant pond and recently placed, saturated tailings columns. The hypothetical pond
level rise and introduction of an external load increased the total head elevations from measured and
calibrated values, with pond level rise causing the largest magnitude of change. Decreasing the
anisotropy ratio (lateral variation of hydraulic conductivity) reduced the pore pressures. Different
external conditions were observed to influence the way that the total head elevation trended over
time, and the response time when alternating between construction and depositional phases differed
between the different external conditions.
When comparing a discrete event that influenced the phreatic condition nearby an upstream
piezometer, the downstream piezometer was also observed to respond to the same change, but with
a 3-month delay. Based on this, the author reiterates that a progression to real-time monitoring may
not necessarily reflect full understanding of the phreatic condition due to limitations of the
piezometer or the conditions that it is subject to. Significant changes to the phreatic condition could
occur upstream within the dam before they are identified by the downstream piezometer. An
interpolation between the pore pressures read on any two piezometers in the hope to understand the
phreatic condition through the dam would be a coarse assumption when considering the complexity
of transient phreatic response to external conditions, as demonstrated in this research.
181
The methodology is able to add value to an observational approach. The case for adopting steady-
state seepage during the design phase is reiterated, where this method relies on accurate monitoring
data trends that would not exist until some time after the dam is established and operating.
183
Chapter 7 Conclusions
7.1. Summary of Thesis
This thesis is concerned with the use of instrumentation and monitoring techniques that can be used
to help practitioners better understand the performance and behaviour of their tailings dam
structure. By first understanding the potential failure modes and behaviours that lead to this
(Chapter 2), practitioners can learn from previous failures (Chapter 3) and apply learnings for the
betterment of their own dam’s safety. Case studies and advancing technology allow for a more
readily available understanding of tailings dam behaviour through use of real-time monitoring
(Chapter 4, Appendix B, Appendix C and Appendix D) which can be applied to an assortment of
instrumentation and monitoring techniques (Chapter 5) connected through the appropriate sensor
networks (Chapter 5). In turn, the data collected from different instrumentation should be used to
iteratively update original design assumptions to understand any difference between predicted and
actual performance of the structure, and the effect that different external factors may have on dam
performance (Chapter 6). In hindsight, there have been no tailings dam failures that are
unexplainable using the data available. The research demonstrated the predictive ability of
instrumentation through the use of case studies (Appendix E).
The potential failure modes and behaviours were deemed a critical factor to the research; the author
did not find a single source reference that allowed the broad variety of stakeholders to pick up and
understand the fundamental mechanisms that could occur, and what the basis could be to address
these. This was developed (Chapter 2) to address four areas of interest for the fundamental tailings
dam failure modes: (1) What factors contribute to this type of failure? (2) How does the mechanism
of failure progress? (3) Are there any indicators that could have been observed, measured, or
monitored to directly measure the development of this failure? (4) How can design counter failure?
The research acknowledged that amidst several different references in literature, the understanding
of core failure modes exists, however the research adds value to the industry by synthesising this
information and providing a reliable base reference which allows practitioners a greater amount of
time to address the unique challenges of different tailings dam sites.
The state of practice for tailings dam monitoring was reviewed to understand what the legislation,
rules and regulation were in different regions (Chapter 3). This was critically reviewed through
literature and supplemented by a survey and analysis of 25 global tailings dam practitioners. It was
found that an inconsistent global approach is being taken to tailings dam monitoring. Practitioners
emphasised an interest in having more readily available information on real-time and available
instrumentation, but also reiterated that a “checklist” approach should not overrule the need for
critical thinking and analysis.
184
Following the outcomes of the survey, it was discovered that the key next step in advancing tailings
dam instrumentation and monitoring techniques was through the use of real-time (rather than
manual, incrementally read) instrumentation. Research was undertaken (Chapter 4) to establish
real-time monitoring system requirements, how this applied to recent well-known global tailings
dam failures, provide insight into the cost of real-time instrumentation, and suggest ways to extract
more value from individual instruments to full monitoring system integration. It was established
that real-time monitoring provided a higher degree of resolution on the performance of a tailings
dam structure, and works for the betterment of cost, time and resource availability onsite. This
optimisation of time and effort can instead be redirected to visual observations of the dam,
interpretation of pre-generated data representations, and updating/activation of risk management
plans, overall toward a more informed and safer operation.
Through engagement with a global network of practitioners, it was found that many did not have
the time or in-depth network to be up-to-date with advancements in instrumentation and monitoring,
amidst the many other tailings dam challenges. Hence, to best leverage the advancement of real-
time instrumentation and monitoring, practitioners require a readily available, cost-effective,
succinct guideline of monitoring and instrumentation techniques (Chapter 5). Research was
undertaken to establish a centralisation of real-time monitoring instruments suitable for tailings
dams, discussing the specifications, advantages, and disadvantages of each. An understanding of
traditional instrumentation was detailed, progressing to online monitoring systems and the value of
centralised monitoring. Collaboration with suppliers discovered advanced and innovative systems
which enable monitoring of different failure modes and mechanisms.
Importantly, and anecdotally less familiar, tailings dam practitioners will also be exposed to the
world of electrical engineering and control systems when looking toward real-time instrumentation.
This involves dataloggers and sensor nodes, sensor networks, communication systems, and other
indirect features of a monitoring system that are critical but typically not core discipline for tailings
dam practitioners. Research was undertaken (Chapter 5) to describe the components involved in
instrumentation and monitoring systems for tailings dams, translating traditionally electrical and
systems engineering terminology into a reference base suitable for the broad range of tailings dam
practitioners. This reference base contributes to advancing the industry by helping to facilitate
informed discussions, to in turn allow deployment of appropriate instrumentation systems that are
suitable to the practitioner’s short and long term intent.
When considering how to use the data collected from different instrumentation, iterative updating of
original design assumptions provides the tailored assessment required for unique environments.
With a thorough understanding of the tailings dam environment, practitioners can feel more
comfortable that the story being written by the measured data is being appropriately critiqued,
185
analysed, and utilised to compare the performance of the tailings dam against original design
assumptions. Research was undertaken to demonstrate this approach by using data sourced from a
mine site in Australia to compare piezometer installations and measurements against numerically
modelled scenarios, calibrating the model against insitu performance and in turn assessing the
sensitivity of phreatic conditions against hypothetical changes to the tailings dam environment
(Chapter 6). The research showed that understanding that the measurement may be in real-time,
but that the condition being measured is a delayed response to pore water pressure changes, is
important when considering a true understanding of dam performance. The case scenario showed
that the external conditions considered (pond level rise, introduction of an external load, and
variable lateral hydraulic conductivity) increased the total head elevation at piezometer
measurement points when compared to field data. It was suggested that different external conditions
reflect different total head elevation behaviours. The research was also presented in a way that
allowed practitioners to repeat the approach, tailored to their unique sites’ dataset.
To further support the case study demonstration of measurement trigger levels and the ability to
utilise monitoring data to forecast geotechnical deterioration was documented in Appendix E. This
research collaborated with engineers from Samarco to demonstrate the critical link between
instrumentation, data analysis, and performance of the structure being monitored. Three case studies
demonstrated the ability to forecast deterioration ahead of time (between 1.3 to 48 hours prior to the
deterioration event), dependent on the properties of the slope being monitored and considering the
appropriate data analysis to suit.
The results obtained in this thesis will serve to better understand the role of instrumentation and
monitoring techniques in safe and sustainable tailings dam practice.
186
Aligned with the release of the Global Industry Standard on Tailings Management, it is suggested
that Future Research in this sector could include:
• An agreed assessment tool which encourages ranking of tailings dams on the same scale of
consequence of failure;
• Established monitoring implications and requirements related to the varying levels of
consequence;
• Integration of monitoring data collected by remote sensing technologies;
• Data interpretation of integrated monitoring data;
• Operational training enhancements, including virtual reality systems;
• Broader sharing and disclosure of case studies and data to advance global tailings dam
practitioner access to examples of both stable and failed behaviours; and
• Continued advancement of monitoring technologies and understanding in line with tailings
failure mechanisms and behaviour.
187
Appendices
Appendix A: Tailings Dam Monitoring Survey
Africa, 2,
Asia, 0, 0% 8%
North and
Latin
Tailings Dam America, 1,
Monitoring and 4% South
America, 6,
Surveillance 24% Australia/
Oceania, 8,
32%
Survey
North
Demographic America, 8,
32%
Europe, 0,
0%
0 5 10 15 20 0 5 10
70 180,000,000
300
160,000,000
60
140,000,000 250
Years Experience
Volume (m3)
50
Height (m)
120,000,000 200
40 100,000,000
80,000,000 150
30
60,000,000 100
20
40,000,000
10 50
20,000,000
0 0 0
Tailings Dam Dam Storage Dam Height
Experience Volume
188
Tailings Dam Monitoring and Surveillance
Safety, Perception, & Education
Alarms at catastrophe/ emergency offices in towns nearby
100%
80%
60%
40%
20%
0%
Yes we have alarming Yes we have alarming No we do not have No we do not have
devices and see the devices but do not see alarming devices but alarming devices and
value the value see the value we do not see the
value
Additional : Stability analyses (including sensitivity analysis), risk assessment, and observation.
What knowledge would you share about your tailings dam monitoring strategy?
State-of-the-art technologies
Operator experience and knowledge
Compliance with acts, regs, and guidelines
Ability to predict and respond to deterioration
Targeted and strategic monitoring…
Integration of monitoring with dam safety plan
0 2 4 6 8 10 12 14 16
Additional: integrated data management system, liquefaction, live VWP predictive
monitoring, and variety of geotechnical monitoring
What do you want to learn more about?
Iteration of numerical modelling using instrumentation reads over time
Predictive technologies for slope displacement and/ or failures
Better means of establishing dam surveillance trends vs triggers
Remote, 24/7 monitoring , in addition to its influence in terms of Population of Risk
Platform for data integration
State-of-the-art in hydraulic and hydrogeologic monitoring and alarm techniques
General knowledge of materials and climate (geotechnics)
Better operator training
Standardisation of good practice in tailings dam management
Method of obtaining reasonable consequence and risk
189
Do you believe that your site represents industry best practice
for tailings dam monitoring?
Why?
Tailings Dam Site support enabling CAPEX spend on automatic monitoring,
Monitoring and as data was being collected previously but not analysed.
No
Surveillance Use of reclamation method.
6%
Variety of instruments and
projects to improve monitoring
Instrumentation Experience with dam failure.
Yes
Why Not?
and Monitoring 94%
Knowledge of operations
Instrumentation Quantity?
Installed? 300
% of sites using % of sites not using 250
Piezometers 200
150
Inclinometers 100
Extensometers 50
0
Survey points
Additional : settlement plates; decant pond water markers; freeboard markers; sonar (water
level); satellite InSAR
How do you correlate information between monitoring devices on your tailings dam?
100%
80%
60%
40%
20%
0%
I treat my instruments in I conduct minor calibration I insist on correlating data I integrate my instruments
isolation exercises between between instruments and correlate continuously
instruments through a technology
platform
What degree of historical tailings dam data exists, for you to compare against?
100%
80%
60%
40%
20%
0%
None Publicly available information Historical InSAR, maps, site A vast range of documented
only specific information publicly and privately
obtained information
How useful would a greater baseline of data be in helping to understand dam behaviour?
100%
50%
0%
Not useful Somewhat useful Very useful
190
Which guideline/s, act/s, and/ or regulation/s for tailings dam
monitoring do you currently work under in your region?
Guidelines Acts/ Regulations Other
Tailings Dam ANCOLD Department of In-house standards
Mining Association Mines and 43-101, PFS/FS,
Monitoring and of Canada Petroleum World Bank, Mine
Surveillance ICOLD State dam safety Permits through
Canadian Dam regulations POOs and NEPA,
Association Portaria 70.389 NPDS and State
Guidelines, Acts Mine Health and Department of discharge permits
and Regulations Safety Inspectorate Environment and
Guideline Heritage Protection
Mine Health Safety Act
SANS 10286:1998
Yes
Do you believe that sufficient guidance and regulation is provided 39%
through the above to ensure safety? No
61%
Points of comment:
Previous failures have been in compliance with regulations/ standards of practice.
Lack of understanding on brittle undrained failures of low plasticity tailings.
Significant variation between states, loose regulation of some existing facilities.
Incorporate fundamental geotechnical understanding into guidelines
Vague regulations.
Operators do not necessarily read the guidelines, only the engineers.
Worldwide knowledge should be shared to help the government write best regulations.
What mandatory requirements are you aware of that exist for tailings dam monitoring and
surveillance?
Real-time, telemetry
24/7 monitoring
0 2 4 6 8 10 12
What improvements would you like to see to regional and global standards of tailings dam
monitoring?
0 2 4 6 8 10 12 14
Do you believe that the frequency of reading mandated through the acts, regulations, or
guidelines, is sufficient to capture the initiation and progression of different failure types?
Identified areas for improvement: Yes
Greater understanding of contractive shearing with respect to VWP 33%
monitoring and integration
Rigid, prescriptive requirements do not have the flexibility to
adequately address the unique conditions at each site No
67%
Must be site specific, not regulation driven.
Ongoing monitoring and operator refresher/ audit to avoid complacency
Establish structured, systematic response plan to mitigate human error in stressful situations
Greater specification by regulations on the frequency of monitoring, and benchmark standards
The frequency of readings alone is not enough, there needs to be an understanding of the
response of each instrument and integration between them.
With a telemetry device, we can obtain a better understanding of the initiation and
progression of a potential failure
191
Do you have a secondary structure downstream to control
potential runout in case of a tailings dam breach? QA/QC?
100%
Surveillance
40%
20%
0%
Events We do not have a We have a secondary We have a secondary
secondary structure structure downstream and structure downstream but
downstream monitor for QA/ QC do not monitor for QA/ QC
192
Appendix B: Catalogue of Real-time Instrumentation and
Monitoring Techniques for Tailings Dams
193
Catalogue of Real-time Instrumentation and Monitoring Techniques for Tailings Dams, Clarkson et al., 2020
Instrument Float-Type Water Level Gauge Vibrating Wire Weir Monitor Ultrasonic Water Level Sensor Bubbler Pressure Transducer Water Content Reflectometer Video Camera
• Simple device.
• Can be submerged in most
• High sensitivity and stability.
• Inexpensive. canals, wells, ponds, lakes, and
• Simple device.
streams.
• Immune to zero drift. • Very supportive when used in
• Reliable.
• Inexpensive. conjunction with other
• Simple device. • 316L stainless steel case.
• Very low response to instrumentation.
• Easily automated. • A reflectometer is not required.
temperature changes. • Reliable.
• Inexpensive. • Temperature compensated.
• Captures a large area.
Advantages • Pressure sensor can be at the • Probe rods can be inserted or
• Output not affected by long • Sensor does not touch water.
• Reliable. air source as opposed to being • Vent tube to compensate for buried at any orientation to the
signal cables (frequency not • Obvious when not serviceable.
submerged. The only submerged atmospheric pressure fluctuations. surface.
affected by changes of cable • Easily automated.
• Easily automated. component is the air tube, which
resistance). • Simple to observe previous
is inexpensive to replace if • Smaller gap between the water
• Ultrasonic continuous level timestamps.
required. ports and the diaphragm so that
• Can support automated weir measurement.
less air is trapped that the user
flow measurements.
• Suitable for use with corrosive must remove during deployment.
fluids.
Visual
Catalogue of Real-time Instrumentation and Monitoring Techniques for Tailings Dams, Clarkson et al., 2020
Instrument Standpipe Piezometer Twin-Tube Hydraulic Piezometer Pneumatic Piezometer Vibrating Wire Piezometer Electrical Resistance Piezometer
Piezometer typically sealed in a borehole, embedded in
Installed in a borehole. Components typically comprise fill, or suspended in a standpipe. Twin pneumatic tubes
The piezometer tip consists of a porous filter enclosing
a filter tip and riser pipe. Zone around filter tip run from piezometer to surface. Water pressure acts Converts water pressure to a frequency signal via a
a reservoir of water. Water is separated from a
backfilled with sand, and a bentonite seal placed above on one side of the flexible diaphragm, and gas diaphragm, a tensioned steel wire, and an A deflecting diaphragm is separated from a porous
pressure gauge by flexible, water filled tubes. Tubes
sand to isolate the intake zone. The remainder of pressure on the other. When a reading is required, electromagnetic coil. Change in pressure on diaphragm ceramic filter by a water filled reservoir. Attached to
circulate water through the system, removing air. Pore
How does it work? borehole is backfilled with bentonite-cement grout. compressed nitrogen gas is fed down the piezometer causes change in tension of the connected wire. The the dry side of the diaphragm is a strain gauge, which
pressure at the piezometer tip is inferred from the
Pore water flows into standpipe until pressure until the gas pressure exceeds the water pressure and vibration caused in the wire in proximity of the coil measures deflection of the diaphragm with water
difference in hydraulic pressure between the tip and
equilibrium reached. Water level in pipe then the diaphragm is forced outward, allowing gas to generates a frequency signal which, with applied pressure, and converts measured strain to pressure.
measuring point on surface. This piezometer is widely
represents the pore water pressure in the soil around return to surface. Gas pressure is reduced until a calibration factors, provides a water pressure reading.
used in unsaturated rolled earth-fill applications.
the intake zone. neutral position where gas pressure equals water
pressure, and the water pressure can be measured.
Typical Range of 300m (water depth) 70, 175, 350, 700, 1000, 2000, 3000, 5000, 7500,
-5 to 2000kPa 0 to 690 kPa 100, 200, 500, 1000, 2000, 5000 kPa
Measurement 1500m (water level meter depth) 10000 kPa
± 0.2% full scale (for 100 & 200 kPa)
Accuracy ± 1mm ± 1.0% full scale ± 0.25% full scale ± 0.1% full scale
± 0.15% full scale (all other full scale)
Parameter Average pressure head of the gauges on each tube Electrical resistance (strain of diaphragm), usually in
Water level (m) Gas pressure (kPa) Frequency signal (Hz)
Measured (kPa) voltage or current
• Minimal zero drift makes these excellent for long
term monitoring.
• Moderately complex transducer.
• Processing of data is simple both manually and with
• Simple device. automation. • Simple to monitor.
• Simple device. • Moderately simple transducer.
• Moderately expensive. • Very short lag time (when installed using fully • Very short lag time.
• Inexpensive. • Sensor is not particularly expensive.
grouted method).
• Suitable for long-term monitoring works (simple • Elevation of readout independent of elevation of tips
• Reliable. • Reliable.
operation, no inaccessible moving parts). • Elevation of readout independent of elevation of tips and piezometric levels.
and piezometric levels.
• Simple to monitor and maintain. • Fairly long experience record.
• Short lag time. • No freezing problems.
• Low temperature sensitivity, however thermistor
Advantages • Can be subjected to rising or falling head tests to • Very short lag time.
• Minimal interference with construction operations. included for temperature compensation. • Frequency output signal permits transmission over
confirm function.
long distances.
• Elevation of readout independent of elevation of tips
• Can be used for constant or rising/ falling head • Frequency output signal permits transmission over
• Easily automated and still allow for water sampling. and piezometric levels.
permeability testing (where soil permeability is less long distances. • Easily automated.
than piezometer filter permeability).
• Used in soil and rock formations where the time lag • No freezing or elevation problems due to use of gas
• Can measure small sub-atmospheric pore water • Can measure small sub-atmospheric pore water
and high displacement requirements inherent in instead of water.
• Can measure positive and negative pore water pressures. pressures.
standpipes are not crucial, and where the presence of
pressures (modified piezometer has been used to
standpipes will not hinder construction. • Use where stability levels are critical.
measure suctions up to 90kPa at 10m depth). • Output signal independent of length of electrical • Output signal independent of length of electrical
cable (signal is frequency based). cable (signal is frequency based).
Visual
Catalogue of Real-time Instrumentation and Monitoring Techniques for Tailings Dams, Clarkson et al., 2020
Instrument Tensiometer Pressure Transducer Water Matric Potential Sensor
A pressure transducer is a measuring device which converts
Components include a hollow plastic tube with porous ceramic tip A 50mA current is applied to the sensor’s heating element.
an applied pressure into an electrical signal. Generally, a
at one end. Tube filled with water. Stress measuring gauge is The thermocouple measures temperature rise. Matric
pressure transducer consists of two parts, an elastic material
typically a vacuum gauge or electronic gauge. As the potential determined by applying a 2nd order polynomial. The
How does it work? which deforms under the application of pressure and an
surrounding soil dries, water is drawn out of the tensiometer and magnitude of temperature ride varies according to the
electrical part which detects this deformation. The electrical
a vacuum is created in the sealed tube, which is measured on amount of water in the porouis ceramic matrix, which
part can be a resistance strain gage, VW sensor, capacitive,
the gauge. changes as the surrounding soil wets and dries.
electromagnetic, etc.
Typical Range of
-160 to 0kPa -100 to 1000 kPa -2500 to -10 kPa
Measurement
Positive Pressure/
Suction Positive Pressure and Suction Suction
Suction
± 0.1% full scale (standard)
Accuracy ± 0.5 kPa ± 1kPa at matric potentials <-100kPa
± 0.05% full scale (standard)
Soil moisture tension. Pore pressures in the vadose
Parameter Measured Electrical resistance (strain of diaphragm) Soil water potential
(unsaturated) zone.
• Measures negative pore water pressures.
• Can be submerged in most canals, wells, ponds, lakes, and
• Installed in borehole. streams.
• Measurements not affected by salts in the soil.
• Simple to read. • 316L stainless steel case.
• Long lasting, with no maintenance required.
• Up to 1500kPa suction has been possible in laboratory tests. • Temperature compensated.
Advantages • Suitable year-round.
• Minimal disturbance. • Vent tube to compensate for atmospheric pressure
fluctuations. • Can be calibrated against tensiometers.
• No calibration required.
• Smaller gap between the water ports and the diaphragm so • Small size.
• Not affected by temperature of soil-water solution. that less air is trapped that the user must remove during
deployment.
• Not affected by salts in soils.
• Range of operation of all tensiometers is limited by the
formation of vapour cavities within the water reservoir (water
cavitation, which causes interruption of the measurement
through tension breakdown and requires instrument
resaturation).
• Max SDI-12 cable length ~457m. Max RS-232 cable length
• Response of miniature tensiometers sensitive to temperature. • 30s measurement time.
~60m.
• Require regular maintenance. • Users must individually calibrate each of their sensors in the
Limitations • Consumes more power than some other types of pressure
soil type.
sensor.
• Reading representative only of local conditions at tip.
• Relatively larger power requirement.
• Challenged in freezing temperatures.
• Range of measurement in vacuum gauge a function of distance
between sensor and porous filter distance. 10kPa reduction per
metre of separation.
• Not suitable for granular soils as contact with the soil is difficult
to maintain.
Visual
Catalogue of Real-time Instrumentation and Monitoring Techniques for Tailings Dams, Clarkson et al., 2020
Instrument Weirs Parshall Flumes Flow Meters (water flow variants) Self-Potential Meters Microseismic Geosynthetics
Visual
Catalogue of Real-time Instrumentation and Monitoring Techniques for Tailings Dams, Clarkson et al., 2020
Vibrating Wire Liquid Settlement
Instrument In-place Inclinometers Tiltmeters Acoustic Sensor GPS/ GNSS Shear Strips Settlement Plates
System
Internal/ Surface
Internal Internal Internal (lateral displacement) Internal/ Surface Internal/ Surface Internal/ Surface Internal/ Surface
Movement
Typical Range of
± 30° ± 5°, 15°, 30°, 90° 500mm displacement Satellite - pseudorange. ± 0.5m 7, 17, 35m
Measurement
Horizontal (<30km): 8mm Dependent on survey equipment
+ 1 ppm Root Mean to pick up vertical position of riser
Accuracy 0.005 mm/ 500 mm ± 0.002°- 0.004° ± 1mm Square (RMS) ± 50mm pipe ± 0.05% full scale
Vertical (<30km): 15mm +
1 ppm RMS
Visual
Catalogue of Real-time Instrumentation and Monitoring Techniques for Tailings Dams, Clarkson et al., 2020
Ground Based Synthetic
Instrument Robotic Total Station Real Aperture Radar Light Detection and Ranging (LiDAR) Interferometric Synthetic Aperture Radar (InSAR) Time-Domain Reflectometer
Aperture Radar
Internal/ Surface
Surface Surface Surface Surface Surface Internal
Movement
Distance: ± 1 to 10mm ±1mm, 3m, 20m (highly dependent on system and Distance: 1.35 mm
Accuracy ± 0.1mm ± 0.1mm ± 2mm + 2ppm
Angular: ± 0.5" to 1" commercial/ free acquisition) Time: <4.4 ρs
Slope Distance,
Collects and analyses a 250-point
Parameter Measured Horizontal Angle, Vertical Line of sight displacement Line of sight displacement Infrared light travel time Satellite line-of-sight displacement
waveform
Angle
• Speed of data
• Low power.
collection. • Remote (no maintenance or install).
• Wide area, fast scanning.
• Robust.
• Wide area. • Wide area (few to 100s of km2).
• Large area coverage, fast scans,
• Automatic atmospheric
long term and critical/ tactical • High sensitivity.
• Three coordinates correction. • Ground based LiDAR (terrestrial laser scanning or 3D laser • Scalable (i.e. focus on TSF of entire mine site).
monitoring.
measured (x,y,z). scanning), airborne, and satellite methods available.
• High resolution.
• Back analysis sub- • Low cost (especially if utilising open source satellite data).
• Relatively high accuracy of
• Long/ medium term sampling for ultra-slow • Quick data collection (1000km2 in 12 hours). High surface
atmospheric correction • Low noise.
monitoring. displacement analysis. density (number of surveyed points). • Can be integrated with other measurement data (for both
Advantages algorithms.
understanding and calibration).
• Advanced waveform filtering.
• Long range monitoring. • Long range capability • Can penetrate vegetation cover (if you can see light, LiDAR can
• Fixed or mobile configurations.
without significant typically reach this point). • No need for data calibration. Very high density of data
• 60 Hz frequency rejection.
• Can use real and resolution compromise. measurement.
• Less interference from mine
virtual reference points. • Can be used day and night.
infrastructure compares to other • Monitor longer continuous length than
• Corregister visual • No service interruption, with regular reporting possible.
radars. other downhole instruments.
• Integrated photo and photographs with radar
video cameras to images. • Not sensitive to horizontal north-south component of the
• May be paired with piezometers downhole
conduct remote deformation due to satellite orbit direction.
to provide pore pressure data
inspections.
• Line of sight sensitivity. • High operating costs when surveying relatively smaller areas. • Significant ground cover changes (e.g. surface excavations, • Not fully real-time (2 second analysis).
vegetation, excavation/ blasting) between each satellite image
• Accuracy of data in • Affected by refraction during heavy rain or low hanging clouds. acquisition. • Requires precise installation and
vegetated (>5cm) areas. experienced user.
• Line of sight sensitivity.
• Must have line of sight • Degraded at high sun angles and reflections. • Presence of dense vegetation limits density of measurement
to monitoring prisms. • Lower accuracy than other points (grassland negligible, forested areas challenging). • Cable abrasion causing abnormalities in
• Accuracy of data in vegetated
methods. • Unreliable for water depth and turbulent breaking waves. the outer conductor will typically result in a
(>5cm) areas.
• Point-based • Single line-of-sight. positive spike in the coefficient.
monitoring. • Fixed installation. • Very large datasets (difficult to interpret).
Limitations • Systems need to be protected
• Limited visibility in steep areas. • Connections very sensitive to water and
from environmental conditions
• Costly instrumentation. • Systems need to be • Cannot penetrate thick vegetation (where light also does not debris.
(and wet weather inhibits
protected from penetrate). • No information over water bodies (or under water), however
measurements).
• Sensitive to bad environmental conditions water edge detection is possible • Corroded cable connections (often due to
weather conditions. (and wet weather inhibits • Operating altitude limited to 2000m. splicing) will typically cause a reduction in
• Pixel size increases with range.
measurements). • Sensitivity to vertical and horizontal earth-west deformation the impedance and hence a negative spike
• Not suited to downstream tailings dams, for example, based on in the coefficient. If the damage is severe
• Near-real time data, the constant expansion of the structure outward (simulating • Frequency of scans limited to once per 12 days (6 days in enough, monitoring of the cable beyond
refreshed every 1-3 minutes outward movement unless accounted for in data analysis) some parts of the world) that location will be redundant.
Visual
Instrument Seismometer Strong-motion Accelerometer Instrument Vibrating Wire Push-in Pressure Cell Load Cells (Total Pressure Cells)
Seismic Wave Frequency (Hz), recorded on Ground acceleration through voltage conversions Frequency signal (Hz). Total Horizontal Stress
Parameter Measured Parameter Measured Frequency (Hz)
Seismograph. (voltage in V, acceleration in g). (MPa)
Visual Visual
Catalogue of Real-time Instrumentation and Monitoring Techniques for Tailings Dams, Clarkson et al., 2020
Instrument Automated Weather Station Vibrating Wire Temperature Sensor Resistance Temperature Sensor Thermistor String
Comprises five essential sensors: thermometer (using
Components: stainless steel body, tensioned wire in body,
thermocouples, thermistors, or PTD probes), barometer
electromagnetic coil, signal cable. Body of sensor expands
(finding amount of pressure applied by air on a particular Components: stainless steel body, resistance temperature
or contracts with changes in temperature, increasing or A string of thermally sensitive resistors are cabled
surface area), hygrometer (temperature difference detector (RTD), signal cable. Resistance of RTD varies
How does it work? decreasing tension of wire inside the body. Send electric together and buried in the ground. The resistance reading
between wet bulb thermometer and dry bulb with temperature, recorded, then converted to units of
pulse to coil, causing vibration at a frequency that is then changes as the temperature changes.
thermometer), anemometer (wind vane, sonic temperature.
converted to units of temperature by using conversion
anemometer, or propeller anemometer), and rain gauge
factors.
(tipping bucket principle and many other types).
Typical Range of Standard: -20 to 80°C
Various -20 to 80°C -20 to 80°C
Measurement Custom: -40 to 200°C
Standard: ± 0.5°C
Accuracy Various ± 0.5°C ± 0.1°C
Custom: ± 0.3°C
• Temperature (thermometer).
• Atmospheric pressure of air (barometer).
Parameter Measured • Humidity (hygrometer). Frequency of Coil (Hz) Resistance (ohm) Resistance (ohm)
• Speed and direction of wind (anemometer).
• Rainfall (rain gauge).
• Convenient when other types of vibrating wire sensors
are used.
• Relatively more economical than vibrating wire sensor.
• High accuracy.
• High accuracy.
• Power requirements can be offset by the use of solar.
• Temperature compensation not necessary.
• Fast response.
• Excellent long term stability.
• Conventional weather stations had to rely on
Advantages • Stability maintained over long periods.
assumptions about diurnal trends in parameters, where • Require only one cable with small diameter and only a
• High resistance to effects from water.
real-time approach (and also considering several few conductors.
• Can measure differential temperatures.
parameters) should provide more accurate estimates.
• Frequency outputs suitable for transmission over long
• Polarity insensitive.
distances (up to 3km). • Easily installed and replaced.
Visual
Catalogue of Real-time Instrumentation and Monitoring Techniques for Tailings Dams, Clarkson et al., 2020
Contributing Organisations
Instrumentation
Campbell CGG EHP Environment ESS Earth Geomotion NavStar SITECH Construction Trimble Worldsensing
Scientific Sciences Systems
Satellite
Dares
3vGeomatics Technology Inmarsat sarmap Tre Altamira
Radar
Seismic
Specialised
Geofabrics Silixa
Appendix C: Catalogue of Example Instrumentation and
Monitoring Systems for Tailings Dams in Australia
194
Catalogue of Instrumentation and Monitoring Systems for Tailings Dams, Clarkson, L & Williams, D 2020
6 Channels
16 SE Channels
Number of Channels 12 Universal Channels + 6 additional with 1 Channel 5 Channels 4 Channels 1 Channel 3 Channels 6 Channels 8 Channels Integrated sensors (tilt)
8 DIFF Channels
extension board
Current Loop
Period Average Current Loop
Current Loop Full Wheatstone Bridge
Measurements Full Wheatstone Bridge
Differential Voltage Current Potentiometer
Voltage Resistor Resistor Potentiometer AML100
Period Average Voltage Ratiometric Modbus RTU GPS
Analog Inputs Ratiometric Temperature Temperature PT100 Total Stations Vibrating Wire Piezometers
Ratiometric Bridge Thermistor Single-Ended Voltage RS485 GNSS
Static Vibrating Wire Frequency Vibrating Wire Frequency Voltage GPS
Single-Ended Voltage Pulse Thermistory
Thermistor Thermistor
Thermocouple
Thermocouple
Vibrating Wire
Chains of in-place
Requires vibrating wire Battery life decreases if Shorter radio
Some VW sensors require inclinometers limited to 32 Line of sight at 30km in
Limitations module for VW high sampling rates are communication range due
up to 6 channels sensors per data logger. distance.
measurement chosen to internal antenna
Power demanding
Catalogue of Instrumentation and Monitoring Systems for Tailings Dams, Clarkson, L & Williams, D 2020
Instrument DT2485 Datalogger DT2011B Datalogger DT2055B Datalogger DT2040 Datalogger Flexdaq Logger RStar Hub VW2106 Readout Water Level Meter Reed Switch Probe TEXCEL ETM+BTM
170 Channels (each biaxial Theoretically no limit (by Up to 255 Nodes via radio 1 Channel (manual read)
Integrated sensors
Number of Channels in-place inclinometer 1 Channel 10 Channels 40 Channels utilising a unique cascade connection. Cabled inputs Up to 6 channels with use 1 Channel (manual read) 1 Channel (manual read)
(vibration)
occupies 3 channels) feature). Modular. modular. of expansion port
Vibrating Wire
Thermistor Vibrating Wire Terminal block for
MEMS (analog and digital) Potentiometers connection to 1 VW
N/A – manual reading N/A – manual reading
Tensmeg MEMS Tilt Sensors Sensor.
1 string of up to 56 digital Vibrating Wire Vibrating Wire Vibrating Wire only. only. Vibration
Analog Inputs Linear Potentiometer Strain Gauge (full bridge) Expansion port can be
bus sensors. Thermistor Thermistors Thermistors For use with Casagrande For use with magnetic Noise
Strain Gauge Digitally Bussed used for connection for
standpipe piezometers. settlement systems.
LVDT 4-20 mA multiple sensors with the
TDR Thermistor appropriate fly lead.
etc.
Operating
-40° to 60°C -40° to 60°C -40° to 60°C -40° to 60°C -40° to 70°C -40° to 60°C -50° to 80°C -10° to 40°C -10° to 40°C -20°C to 55°C
Temperature Range
Gateway Radio
Campbell Scientific Sierra Wireless
Instrument Communication LoRa Cell WiFi Satellite RF412 EHP DL12 DT Link Rstar Hub Trimble Settop M1
Cell220 RV50X
Protocol
Carrier Approvals: LTE: AWS, 700, 850,
Telstra 1900 MHz
Cell, or Satellite (900 4G: 700, 900, 1800,
LTE: 700, 850, 1800, WCDMA: 850, 900,
Operating Band 868, 915, 915-928 MHz 900 MHz 2.4 GHz 1 GHz to 40 GHz 922 MHz MHz, or 1 GHz to 40 915 - 928 MHz 915 - 928 MHz 2100, 2300, 2600 MHz
2100, 2600 MHz 1900, 2100 MHz
GHz) WiFi: 2.4 GHz
WCDMA: 850, 2100 EV-DO/CDMA: 800,
MHz 1700, 1900 MHz
Up to 14km depending
Worldwide (as long as WiFi network, 4G
Optimal Operational Up to 15km in open Worldwide (as long as Dependent on site No limit (as long as No limit (as long as No limit (as long as Up to 800m depending on operating band, line
there is satellite 500m to 1000m coverage, or ethernet
Range space there is cell coverage) signal and hardware. there is coverage) there is coverage) there is coverage) on operating band of site, aerial types and
coverage) cable connection
mounting heights
Company Campbell Scientific DARES EHP Environment IDS Georadar NavStar Geometrics RST Instruments SITECH Vista Data Vision
Product Name Loggernet Eagle.io DARES MAPPER ehp-data.com Guardian GeoExplorer Geoviewer DT Logger Host Inclinalysis Trimble 4D Control Vista Data Vision
195
TRENDS IN REAL-TIME INSTRUMENTATION AND MONITORING
TECHNIQUES FOR TAILINGS DAMS
L.Clarkson1, D. Williams2, P. Redcliffe3 and F. van der Wal4
1
ABSTRACT
With ongoing catastrophic mine tailings dam failures, the hindsight revelation of poor safety
records, and an increasing prevalence of public scrutiny and attention of mining operations, there
is an immediate call for enhanced safety provisions of tailings dams. It is estimated that each 1/3
century the potential risk of tailings dam failure increases by 20 fold; to address increasing
demands on waste volume, tailings storage facilities must be bigger, built faster, and to be longer
lasting. Today, challenges arise in identifying and utilising the ability of monitoring systems to
understand the complex performance and rapid behaviours of these dams, in turn reflecting on the
system’s ability to be able to predict deterioration before failure occurs. New literature, mining
regulators, insurance firms, and mining practitioners are calling for increased diligence in the form
of real-time monitoring: but what can the industry offer in response?
This research presents a snapshot and insight into the trends, patterns, and common
considerations identified in the current standing of real-time tailings dam monitoring practice.
Based primarily on information derived from concurrent research by the author (in form of a
Catalogue of Real-time Instrumentation and Monitoring Techniques for Tailings Dams), challenges
and opportunities when transitioning to a real-time monitoring system are discussed. Learnings
and key considerations between different monitoring approaches are explored, including themes of
limitations and advantages that were observed consistently through global supplier feedback.
INTRODUCTION
Real-time monitoring provides the opportunity to have continuous readings of different
instrumentation that are installed within the tailings dam structure, automatically presented in an
interpretable format and notifying key personnel of issues before they have even considered
scheduling a time to analyse the data. This is an advancement on traditional approaches, whereby
in order to collect and interpret the data, mine personnel are required to:
1. Visit the instrumentation site;
2. Download the data;
3. Return to the office;
4. Transfer/ input the data into the system;
5. Manipulate the data into an interpretable form;
6. Interpret the data (or communicate the data to the interpreting party, which may then take in
the order of weeks to months to receive feedback); then
7. Respond as appropriate.
There are a number of inherent risks associated with this process. The three most significant risks
are the time between readings, potential for human error, and hazardous access. The time
between collection and reporting can vary from anywhere between a day to a month, giving rise to
the question of how truly representative the data is of dam performance by the time it is analysed.
This process is also challenged by the speed at which deterioration of tailings dams progresses to
failure; the monitoring process is near redundant if it is not appropriately frequent to catch
deterioration and respond while the opportunity exists. Human error with data collection, input,
transfer, manipulation also exists: human error can be caused by time pressures, complacency, or
for the simple fact that a mistake was made. By automating the data handling, this risk is mitigated
and also allows mine personnel to reallocate this time to critical interpretation of the data.
Hazardous access is particularly prevalent during times of deterioration. Should an area be
deemed unstable, it is important to understand its behaviour but not at the risk of sending a human
to collect a measurement. Real-time, remote monitoring mitigates this exposure.
Real-time instrumentation works to counter these (and other) limitations by:
• Automating data handling to reduce the risk of human error and allowing mine personnel to
reallocate the time to interpretation of the data;
2
• Centralising monitoring data that can be queried by authorised users at any time, from
anywhere in the world;
• Exhibiting increased monitoring program reliability while reducing data acquisition and
processing costs; and
• Allowing current and historical data interpretation with interaction of any data along the
timescale.
Literature has documented a considerable overlap between the instrumentation recommended for
embankment dams when compared to concrete dams (Avella 1993). However, the embedment of
real-time capabilities appears to be on a case-by-case basis (aside from seismic monitoring), and
as such is not readily documented. Comparing the mandated read frequency between
embankment dams (Clarkson 2019) and concrete dams (Avella 1993), the following key
observations are noted:
• The minimum recommended inspection frequency for seepage measurements and pore
pressures is the same;
• For piezometers, observations wells, foundation deformation, and extensometers, concrete
dam readings were recommended at a frequency of two to six times higher than
embankment dams; and
• For geodetic surveys (EDM, theodolites, etc.), embankment settlement points, and total
pressure cells, concrete dam readings were recommended at a frequency of 10 to 12 times
higher than embankment dams.
ICOLD (2001) reported that in the 18,000 mines around the world, the failure rate of tailings dams
in the past 100 years was estimated at 1.2%, while the failure rate of the traditional water storage
dam was 0.01%.
There has been speculation around recent, catastrophic tailings dam failures, that the monitoring
would not have been able to predict or foresee the failure. Yet, the full opportunity presented by
instrumentation and monitoring systems has not yet been realised or implemented. Until the global
practice of tailings dams monitoring is improved, and opportunities are utilised in practice (hence
proving whether or not these systems are able to predict tailings dam failure), it is inappropriate to
deem this an insufficient method for identifying failure ahead of time.
METHOD
This paper forms one part of a wider research project to develop a comprehensive monitoring
strategy for tailings dams. This research is led by Luke Clarkson and supervised by Professor
David Williams. As part of this particular project, three complementary research papers addressed
specific elements of monitoring and are referenced as:
• Clarkson, L, Williams, D, Redcliffe, P, and van der Wal, F, in press (under journal review).
Catalogue of Real-time Instrumentation and Monitoring Techniques for Tailings Dams;
• Redcliffe, P, 2019. Operating Requirements and Conditions Influencing Real-Time
Piezometer Monitoring of Tailings Dams, Undergraduate Thesis, The University of
Queensland, Brisbane; and
• van der Wal, F, 2019. Satellite Technology for Advanced Monitoring of Tailings Dams,
Undergraduate Thesis, The University of Queensland, Brisbane.
The broader project involves two main phases, broadly listed as:
1. Identify and catalogue the present-day range of tailings dam monitoring instrumentation,
systems, and technologies (this research paper);
2. Developing guidance on improved monitoring of tailings dams.
This conference paper presents a snapshot and insight into the trends, patterns, and common
considerations identified in the current standing of real-time tailings dam monitoring practice. The
primary source of information by which this paper is derived is the Catalogue of Real-time
3
Instrumentation and Monitoring Techniques for Tailings Dams. This catalogue established a
centralisation of real-time monitoring instruments suitable for tailings dams, discussing the
specifications, advantages, and disadvantages of each. An understanding of traditional
instrumentation was detailed, progressing to online monitoring systems and the value of
centralised monitoring. Collaboration with suppliers discovered advanced and innovative systems
which enable monitoring of different failure modes and mechanisms.
While static liquefaction has become a topic of interest in recent years, the focus in mitigating the
risk associated with this has been on design considerations of potential strength reductions,
operational control and planning, and material characterisation for susceptibility to liquefaction and
to understand pre-consolidation stresses. KCB (2018) stated that “much of the risk depends on the
in-situ stress regime, which is difficult to measure and monitor.” Considering the speed of static
liquefaction occurrence, and the current inability of monitoring systems to identify conditions
preceding static liquefaction, real-time monitoring is anticipated to be a beneficial step in the right
direction to significantly increase the frequency of monitoring, more readily understand developing
conditions in the tailings dam structure, and in turn develop further knowledge on the phenomena
and ways that the risk can be mitigated.
Instrument Applicability
A qualitative assessment of the ability and usefulness of different parameters to provide an
indication of potential instability was undertaken. This was based on the instruments documented
in the Catalogue of Real-time Instrumentation and Monitoring Techniques for Tailings Dams, and
considered the following:
• Direct Indicators: Provide a direct measurement of parameters that would be expected to
change in the event of instability (e.g. deformation of the dam face downward and outward
could indicate slope instability);
• Supporting Indicators: Provide a secondary measurement of parameters that could be
influenced through the initiation and/ or progression of instability (e.g. settlement of the dam
crest, while primarily considered as deformation behaviour, can also reduce freeboard and
encourage overtopping); and
• Indicator Not Recommended: It would be recommended to use different parameters to
measure for this particular key performance metric.
A summary of this assessment is presented in Table 1. Note that NR = Not Recommended.
Internal
Foundation Slope
Erosion and Overtopping Seepage Seismicity
Failure Stability
Piping
Phreatic
Surface/ Water Support Support Direct Direct Support Support
Level
Pore Water
Direct Support NR Direct Support Direct
Pressure
Seepage Flow Direct Direct NR Direct Support Direct
Deformation and
Support Support Support Support Support Direct
Movement
Seismicity Support Direct Support NR Direct Direct
Earth Pressures Direct NR NR NR NR Direct
Climate Support Support Support Support Support Support
TABLE 1 – Qualitative summary of instrument applicability in measuring different key performance
metrics
4
Parameter Measured
Whether the final measurement is derived through correlation or direct measurement, different
instruments sometimes rely on the measurement of similar parameters to understand dam
conditions. For example, the widest variety of key performance metrics is assessed through
measurement of frequency signals, in Hertz. This measurement was applied to several different
scenarios including for measurement of:
• Phreatic surface and water level (vibrating wire weir monitor);
• Pore pressure (vibrating wire piezometer);
• Deformation and movement (liquid settlement system);
• Seismicity (seismometer);
• Earth pressures (vibrating wire push-in pressure cell and total pressure cells); and
• Climate (vibrating wire temperature sensor).
It is clear in this case that many of the instruments measuring frequency adopt the vibrating wire
concept, contributing to four of the six key performance metrics that can rely on the measurement
of frequency. Other parameters that presented a multitude of applications include the
measurement of electrical resistance, across six key performance metrics, and pressure, across
four key performance metrics. Technologies with such widespread application could be identified
for research development, where further investment in the concept could provide a multitude of
benefits.
A summary chart detailing the different parameters measured, alongside the key performance
metric that this parameter contributes to, is presented in Figure 1.
FIG 1 – Chart showing parameter measured versus number of instruments measuring the
parameter, categorised by key performance metrics
Accuracy of Measurement
The accuracy of instrumentation in the geotechnical monitoring context is often measured as a
percentage of full scale (% FS). This means that the percentage of variation is fully dependent on
the maximum reading that can be achieved using the device. The distribution of accuracy of
different measured parameters is presented in Figure 2. Seismicity has not been included as the %
of full scale accuracy, considering both the increased magnitude of events recorded and sheer
5
distance over which readings are taken (unlike other instruments which measure conditions
relatively local to the sensor) presents ±% of full scale accuracy between 75% and 2000%.
FIG 2 – Chart showing parameter measured versus number of instruments measuring the
parameter, categorised by key performance metrics
Generally, the trend shows that instruments used for monitoring deformation and movement exhibit
the relative highest accuracy as ±% full scale. This reflects accuracies including ±0.1mm at
distances of up to 5000m (ground based synthetic aperture radar) for surface monitoring, and
0.004° for a tilt of 90° (for example) for internal structure monitoring. This is considered critical due
to the sensitivity of structures to discrete movements, for example if surface monitoring is to
identify the progression of failure, trigger levels integrated into the Trigger Action Response Plan
(TARP) are typically in the order of 1 – 10mm/hr, making the accuracy at long distances an
important criteria.
Seepage flow, on the other hand, may not need to be as accurate as deformation monitoring. The
skew in a higher ±% full scale seen in Figure 2 is predominantly through the consideration of weirs
and parshall flumes. At the maximum range of measurement, flow rates of ±0.4m 3/s and ±0.5m3/s,
respectively in these structures, can still be measured. This could be considered suitable for the
application, where the opportunity still exists to refine seepage measurements using the likes of
flow metres or microseismic monitoring in critical areas.
Complexity to Monitor
A number of instruments comprise complex systems that can be difficult to install, monitor, or
automate. These will typically require greater involvement from instrument specialists (typically
from the supplier side) at the different stages of complexity, and can make it challenging to
troubleshoot onsite should an error or anomalous reading arise. On the contrary to the complexity
of new technologies, it is also observed that there are few familiar operators that can readily
interrogate instruments that have existed for a long time and are complex to read, such as the twin-
tube hydraulic piezometer. Instruments that were specifically identified for their simplicity or
complexity of monitoring and maintenance include:
Simple Devices
• Phreatic Surface and Water Level
o Float-type water level gauge (simple device, easily automated);
6
o Ultrasonic water level sensor (simple device, easily automated);
o Bubbler (simple device, easily automated); and
o Video camera (simple, familiar, and easy to observe previous timestamps).
• Phreatic Surface and Pore Pressures
o Standpipe piezometer (simple device, simple to monitor and maintain);
o Vibrating Wire piezometer (simple to monitor, easily automated); and
o Electrical resistance piezometer (simple to monitor, easily automated).
• Pore Pressures
o Tensiometer (simple to read).
• Seepage Flow
o Weirs (simple, common multi-industry device); and
o Parshall Flumes (simple, common multi-industry device).
• Deformation and Movement
o Acoustic sensor (simple to install); and
o Settlement plates (simply to install, minimal system components).
• Climate
o Resistance temperature sensor (easily installed and replaced).
Complex Devices
• Phreatic Surface and Pore Pressures
o Twin-tube hydraulic piezometer (moderately complex monitoring and maintenance,
moderately complex automation); and
o Pneumatic piezometer (moderately complex monitoring and maintenance, complex
automation).
• Seepage Flow
o Self-potential meters (Several electrochemical mechanisms (natural and manmade)
contribute to the self-potential phenomena, potentially clouding interpretation).
• Deformation and Movement
o Radar/ LiDAR Devices (require training and/ or supplier involvement to set up stations,
undertake survey, and sometimes interpret large datasets); and
o Time-Domain Reflectometer (requires precise installation and experienced users).
• Seismic
o Complex systems requiring training and experience.
• Earth Pressures
o Vibrating wire push-in pressure cell (technically complex installation).
7
Instrument Type
The category of Instrument Type describes the elements which were directly related to the physical
properties of the instrument, and prevalent external influences in the tailings dam environment. The
significance of this is reiterated by the inevitable embedment of instruments in harsh tailings
environments, typically in climatically challenging locations, and subject to ongoing nearby
operations.
Durability
Durability determines an instrument’s position on the scale between immunity to shock loading,
and fragility (being affected by shock and vibration). Some instruments are susceptible to shock
loading based on their location of install (such as settlement plates, particularly in areas of
progressive construction), while others are susceptible based on their design (such as a thermistor
string which comprises fragile buried cables, or automated weather stations, which are designed to
typically operate in remote, low-interaction locations).
By understanding the durability of an instrument (or lack thereof), practitioners have the ability to
strategically locate the instrument to mitigate the risk of interruption during operation. Alternatively,
should an instrument need to be installed in a specific area, an appropriately durable instrument
can be selected for the application (whether immune if interaction is expected or allowably fragile if
interaction is not expected).
Temperature Influence
The influence of temperature on the operation of an instrument can, in turn, influence the
measurement returned. This can be through:
• Slow responses to rapidly changing temperatures (vibrating wire temperature sensor),
• Causing internal fluids to expand at different rates than the surrounding soils (total pressure
cells),
• Diurnal movement of surface features (expansion/ contraction with change in temperature)
altering the surface movement measurements (radar measuring a slope with draped mesh),
• Excitation and hence volumetric change of fluids , pushing typical correlations outside of the
ranges that they were derived for (weirs);
• Sensitivity of specific instrument types to temperature change (vibrating wire piezometer,
electrical resistance piezometer, tensiometer, in-place inclinometers); or
• Required correction for air temperatures considering the method of measurement (ultrasonic
water level sensor).
Some instruments mitigate the potential influence of temperature through the inclusion of
thermistors for temperature compensation (such as in the vibrating wire liquid settlement system
and total pressure cells), or by harnessing techniques that do not require temperature
compensation at all. Alternatively, changing instrumentation types could be a risk mitigation,
whether through substitution the instrument type itself, or selection of a variant to the particular
type of instrument (such as flow meters, where differential pressure flowmeters are suitable for
extreme temperatures and pressures).
Environmental Susceptibility
It appears that the chemical influence traditionally needed to be considered in different tailings
environments (such as water and soil corrosivity, aggressivity, etc.) has been appropriately
accounted for in modern instrumentation design. However, the influence of climatic, environmental
susceptibility remains important. Particularly in freezing conditions, instruments such as float-type
water level gauges, vibrating wire weir monitors, ultrasonic water level sensors, pressure
transducers, standpipe piezometers, and instrument readout locations are challenged in operation.
This is typically either as a result of a physical interaction between the instrument and water,
reliability of water in cabled systems, the presence of ice creating false readings, or climatic
sensitivity of readout units.
8
Some instruments that incorporate gas/ electrical currents (pneumatic piezometers/ electrical
resistance) or operate remotely (such as radar/ LiDAR), mitigate the risk of freezing conditions and
can operate effectively, regardless.
Weather, in particular wet weather, is also an important consideration when installing and
monitoring different instruments. A number of different instruments are documented to be
influenced by atmospheric change, including:
• Real Aperture Radar/ Ground Based Synthetic Aperture Radar (wet weather inhibits
measurements, and can create a false surface that the radar reads instead of the slope
surface);
• Robotic Total Station (sensitive to bad weather conditions);
• Standpipe piezometer (readings can be affected by rainwater runoff); and
• Light Detection and Ranging, LiDAR (affected by refraction during heavy rain or low hanging
clouds).
The weather conditions, often conveniently gathered from an onsite automated weather station,
can be appropriately accounted for by ensuring appropriate consideration when interpreting the
measurements of susceptible instruments.
Integration versus Isolation
As instrumentation and monitoring practices progress toward automation, the opportunity to better
integrate different instruments for an improved overall understanding of dam behaviour is
presented. This is as opposed to treating the instruments in isolation. However, some noise
sensitivity (discussed further in Instrument Autonomy) can limit the opportunity to integrate nearby
instruments. For example, self-potential meters measure the electrical potential between two
stations, however are sensitive to noise from nearby electrical installations. Other instruments and
techniques susceptible to noise include microseismic monitoring and thermistor strings.
Other instruments present opportunity for integration through complementary parameters.
Integration can also complement whole-of-system redundancy by presenting a degree of
contingency should one instrument fail, yet was previously observed to mirror a change in
parameters elsewhere (for example pore pressures measured in a piezometer could be correlated
with measured flow rates in a downstream weir, or increased earth pressures could be correlated
with vertical displacement of foundation materials).
Instrument Handling
The category of Instrument Handling describes the elements which were directly related to the
touch points at which the operators might be involved. The significance of this is reiterated when
considering the applicability of research solutions to practice, and in working toward cost
optimisation in instrumentation capital and operating expenditures.
Complexity of Installation
As described as part of an earlier section, the complexity of installation can challenge an
instrument’s suitability for a certain application, considering:
• Turnaround time and urgency of instrument requirement onsite;
• Ability to configure/ troubleshoot the installation onsite if the instrument is interrupted or
presenting anomalous readings;
• Autonomy of mine site in being able to purchase and repeatedly install instruments of a
certain type; and
• Cost implications when considering instrument cost and additional labour/ support required.
Simple devices that readily provide valuable indicators of performance are often favoured in an
ongoing operational environment, over complex yet specialist equipment which may instead be
reserved for complex and specialist scenarios. Hence, from the supplier perspective there is value
in simplifying the installation process or where this is unavoidable, providing the right knowledge,
9
understanding, and support to allow operators control with the right level of verification to ensure
quality monitoring practices.
Operator Familiarity
From a practical perspective, the familiarity of operators with any particular instrument (or at the
least, with any particular concept, such as vibrating wires) can hold a multitude of benefits. An
ability to troubleshoot, repair, install, and importantly understand the purpose and limitations of any
instrument drastically improves the likelihood of valuable measurements and understandings being
returned. While new and innovative instruments are typically quoted to require some sort of
external support, allowing the ability for any mine site to operate autonomously with the right
understanding is encouraged at the forefront.
On the contrary, it is noted that there are few operators with historical knowledge of how to
monitor, de-air, and manage instruments like the twin-tube hydraulic piezometer.
Common, and multi-industry instruments, such as standpipe/ vibrating wire piezometers, weirs/
flumes, video cameras, inclinometers, GPS/ GNSS, seismic monitoring, and total pressure cells
will always benefit from the inherent familiarity, and also the simplicity of monitoring that comes
from relatively increased repetition of installation and monitoring of these types of instrument.
Frequency and Difficulty of Maintenance
Along the same theme of improving the useability, interrogation, and operation phase of different
instruments is exploring the complexity of, and frequency required to access instruments to
undertake maintenance. In some cases, no maintenance of instrumentation is required (such as in
InSAR due to remote monitoring, and water matric potential sensors). In other cases from the
research, maintenance requirements were specifically noted for some instruments, including:
Minor Maintenance Required
• Phreatic Surface and Pore Pressures
o Standpipe piezometer (simple to maintain).
• Seepage Flow
o Weirs (predominantly clearing/ repairing damage or clogging by debris);
o Parshall flumes (debris passes more readily hence less rigorous maintenance than weirs);
and
o Flow meters (sufficient variants to tailor to the specific measuring environment, hence
instrument suitability mitigates risk of instrument failure).
Moderate Maintenance Required
• Phreatic Surface and Water Level
o Vibrating wire weir monitor (periodic maintenance required to replace desiccant in vent
line).
• Phreatic Surface and Pore Pressures
o Twin-tube hydraulic piezometer (moderately complex maintenance, and requires periodic
de-airing which can take between 10-40 minutes); and
o Pneumatic piezometer (moderately complex maintenance, system cannot be de-aired).
High Maintenance Required
• Phreatic Surface and Water Level
o Bubbler (due to mechanical wear).
• Pore Pressure
o Tensiometer (regular maintenance required).
10
Instrument Autonomy
The category of Instrument Autonomy describes the elements which were directly related to the
ability to rely on instruments to report accurate and reliable information with minimal user
involvement. The significance of this is reiterated in the innovative advancements of the industry,
and an inherent necessity to be able to trust that the information provided reflects the true health of
the dam.
Communication
The ability to transmit and communicate information over long distances is inherently dependent on
the measurement devices (e.g. dataloggers) and transmitters (e.g. WiFi, Bluetooth) linked to the
core instrument. Hence, when considering the instrumentation itself, the ability to communicate the
information being measured depends on the output type.
For example, in vibrating wire and electrical resistance instruments, the signal being frequency
based means that the output signal is independent of the length of electrical cable (frequency is not
affected by changes of cable resistance). Feedback on the vibrating wire temperature sensor
indicated that these outputs are suitable for transmission over long distances (up to 3km).
Tiltmeters, on the other hand, which measure the AC resistance of excitation electrodes, exhibit
long-range communications up to 15km.
Communications in strong-motion accelerometers (measuring seismicity) can be interrupted during
an earthquake, resulting in near-real time measurements instead of real-time.
The ability to measure critical dam performance parameters is redundant if this information is not
communicated to the appropriate receiver. Hence, it is recommended that additional research is
undertaken into the suitability of different measurement devices and transmitters suitable to remote
tailings dam environments.
Stability and Reliability
An instrument’s long term stability provides an indication of whether any decalibration (and hence
repair, or replacement) or zero drift may occur. Zero drift is when the zero reading is modified by a
change in ambient conditions, hence presenting readings not representative of true measurement
and not appropriate for long-term monitoring. Reliability, on the other hand, describes the likelihood
that failure in one area of the instrument can make the whole instrument redundant. For example,
should instrumentation with several nodes operate through a daisy-chain (a single cable
progressing through all sensors to the eventual datalogger), failure or shearing of a single location
along the daisy-chain has the potential to deem the entire instrument failed.
In the assessment, long term stability and reliability was advertised as an advantage for different
instruments, and was highlighted for:
• Phreatic Surface and Water Level
o Float-type water level gauge;
o Vibrating wire weir monitor;
o Ultrasonic water level sensor; and
o Bubbler.
• Phreatic Surface and Pore Pressures
o Standipe piezometer; and
o Twin-tube hydraulic piezometer.
• Seepage Flow
o Weirs;
o Parshall flumes; and
o Self-potential meters.
• Deformation and Movement
11
o In-place inclinometers; and
o Vibrating wire liquid settlement system.
• Earth Pressures
o Total pressure cells.
• Climate
o Vibrating wire temperature sensor; and
o Resistance temperature sensor.
On the other hand, some instruments were flagged to be potentially unreliable in the long term,
including:
• Phreatic Surface and Pore Pressure
o Pneumatic piezometer
o Vibrating wire piezometer (risk of zero drift, but some models available with in-situ
calibration check); and
o Electrical resistance piezometer (risk of zero drift, but some models available with in-situ
calibration check).
• Seismicity
o Strong-motion accelerometer (difficulty in maintaining accurate absolute timing, resulting
in time drift).
• Climate
o Thermistor string.
Power Demand
Power requirements of different instrumentation, particularly in remote areas, is significant in terms
of ensuring that the instrument remains operable and reporting the measured parameters. If
required power is not maintained to the instrument, the instrument’s purpose (and associated
understanding of dam performance at that location) could be redundant. A select set of instruments
reported power consumption as either an advantage or limitation, with time-domain reflectometers
(measuring for Deformation and Movement) the only instrument to advertise low power as an
advantage. For those that consume a relatively large amount of power, an assessment would be
required of the reliability of the power source, or whether there is an opportunity to offset power
requirements through the use of different sources, such as solar. Instruments noted to be
challenged by relatively higher power consumption include:
• Phreatic Surface and Water Level
o Pressure transducer.
• Pore Pressures
o Water matric potential sensor.
• Seismicity
o Strong-motion accelerometer (relatively high standby power requirement for digital
systems).
• Climate
o Automated weather station.
Noise
In the interpretation of raw measurements, it is critical to identify what is potentially data “noise”
and what is true change in readings. Data “noise” represents an interruption of the direct
measurement, whether due to physical, atmospheric, or electrical disturbance. In the physical
12
case, debris, vegetation, and obstacles not related directly to the target structure can cause false
readings of remote (and line of sight) instrumentation. For example, LiDAR cannot penetrate thick
vegetation (where light does not penetrate), InSAR is challenged by dense vegetation (grasslands
present negligible noise, forested areas are challenging), and radar also exhibits a reduced
accuracy of data in vegetated areas. In the atmospheric case, noise and false readings can be
introduced by diurnal movement of structures through temperature variability (in particular,
expansion/ contraction of steel), as well as false surfaces (surface monitoring) and water levels
(standpipe piezometer) generated by rainfall runoff. In the electrical case, instruments measuring
voltage differences and electrical properties can experience false readings generated by nearby
electrical installations. Microseismic monitoring is also susceptible to noise as the seismic signals
generated during monitoring have extremely low amplitude and hence are sensitive to background
mine and processing noise.
RECOMMENDATIONS
In both development of the Catalogue of Real-time Instrumentation and Monitoring Techniques for
Tailings Dams, and interpretative analysis of the results and trends in this research, a number of
recommendations have been identified in two primary themes: For Industry Consideration and
Further Research Development.
13
CONCLUSIONS
This research analysed a catalogue of real-time instrumentation and monitoring techniques for
tailings dams, focusing on the key performance parameters for identification of prevalent failure
modes and mechanisms. Through due consideration of the advantages, disadvantages,
specifications, and considerations that are pertinent to each tailored instrument type, practitioners
have the opportunity to be better informed when engaging with consultants and suppliers, but most
importantly are presented with the information needed to help toward a better understanding and
management of their tailings dam.
Identifying trends in real-time instrumentation and monitoring techniques for tailings dams allows
the industry to identify and address critical limitations, build on concepts that have widespread
potential (such as vibrating wire), and also reference transparent information so that safety can
remain at the forefront. This research has built on the concurrent development, Catalogue of Real-
time Instrumentation and Monitoring Techniques for Tailings Dams, with the intent to provide
commentary and analysis around otherwise relatively factual information.
ACKNOWLEDGEMENTS
The primary author would like to acknowledge the contributing authors for their support in different
elements of this work: David Williams for sponsoring/ supervising the PhD research through The
University of Queensland; and Pippa Redcliffe and Freya van der Wal for their enthusiasm and
deep-dive final year theses into two topics of the broader project: piezometer monitoring and
InSAR monitoring, respectively. The author would also like to acknowledge the extensive list of
suppliers that provided content, support, and insight into development of the Catalogue of Real-
time Instrumentation and Monitoring Techniques for Tailings Dams. Two primary contributors to
this are RST Instruments and Canary Systems, who embody the transparent and supportive
approach required to accelerate advancement of the industry toward safer mining.
REFERENCES
Avella, S 1993. An analysis of a worldwide status for monitoring and analysis of dam deformation, Masters Thesis,
University of New Brunswick, Fredericton, New Brunswick, Canada.
Clarkson, L & Williams, D., 2019. Critical Review of Tailings Dam Monitoring Best Practice. International Journal of
Mining, Reclamation and Environment. https://doi.org/10.1080/17480930.2019.1625172
Clarkson, L, Williams, D, Redcliffe, P, and van der Wal, F, in press. Catalogue of Real-time Instrumentation and
Monitoring Techniques for Tailings Dams.
ICOLD 2001, Tailings Dams—risk of dangerous occurrences, lessons learnt from practical experiences, United Nations
Environmental Programme (UNEP), Division of Technology, Industry and Economics (DTIE) and International
Commission on Large Dams (ICOLD), Paris, France, Bulletin 121
Klohn Crippen Berger (KCB) 2018. Static Liquefaction and Strength Loss in Tailings Dams [online]. Available from:
<https://www.klohn.com/blog/static-liquefaction-strength-loss-tailings-dams/> [Accessed: 30 November 2019].
Redcliffe, P, 2019. Operating Requirements and Conditions Influencing Real-Time Piezometer Monitoring of Tailings
Dams, Undergraduate Thesis, The University of Queensland, Brisbane.
van der Wal, F, 2019. Satellite Technology for Advanced Monitoring of Tailings Dams, Undergraduate Thesis, The
University of Queensland, Brisbane.
14
Appendix E: State-of-the-art Monitoring Techniques for
Tailings Dams
196
STATE-OF-THE-ART MONITORING TECHNIQUES FOR SAMARCO TAILINGS DAMS
LUKE CLARKSON1
Samuel Carneiro2; Breno de Matos Castilho2; David Williams1; Marc Ruest1; Marcelo Llano1; and David Noon3
1
The University of Queensland Geotechnical Engineering Centre, School of Civil Engineering, Brisbane, Queens-
land, Australia.
2
Samarco Mineração, Mariana, Minas Gerais, Brazil.
3
GroundProbe, Brisbane, Queensland, Australia.
ABSTRACT
*RRGWDLOLQJVGDPPDQDJHPHQWEHFRPHVFULWLFDODVWKHFRPSOH[LW\RIPLQLQJSURMHFWVLQFUHDVHVDORQJZLWKWKHLQ
creasing global focus on safe and sustainable development. The industry standard, and fundamental benchmark for a
tailings storage facility, is to provide ‘safe, stable, and economical storage of tailings presenting negligible public health
and safety risks and acceptably low social and environmental impacts during operation and post-closure’.
Geotechnical stability is a key consideration in many tailings dam failures, and monitoring and instrumentation of tai-
OLQJVGDPVSURYLGHVYDOXDEOHLQVLJKWVLQWRSRWHQWLDOWDLOLQJVGDPIDLOXUHPHFKDQLVPV*LYHQWKHH[WUHPHO\KLJKFRVWV
RIWDLOLQJVGDPIDLOXUHVVXFKPRQLWRULQJDQGLQVWUXPHQWDWLRQLVDFRVWHႇHFWLYHPHDQVRIGHWHUPLQLQJWKHFULWLFDOSDUD
meters that may mitigate the risks and potential consequences of tailings dam failures. This paper discusses the out-
comes of a back-analysis of state-of-the art slope monitoring techniques undertaken at a number of sites at Samarco
Mine in Brazil. The data are assessed in terms of geotechnical stability and the associated alarm levels. The ability to
SUHHPSWLYHO\LGHQWLI\DQGUHVSRQGDSSURSULDWHO\WRVORSHLQVWDELOLWLHVLVH[SORUHG
1. INTRODUCTION
7KHGHPDQGIRUWDLOLQJVGDPJURZWKDQGLQWXUQWKHULVNDVVRFLDWHGZLWKWKHVHVWUXFWXUHVLVLQFUHDVLQJH[SRQHQWLDOO\
ZLWKWLPH7RDGGUHVVZDVWHYROXPHUHTXLUHPHQWVWDLOLQJVVWRUDJHIDFLOLWLHVPXVWEHEXLOW5REHUWVRQ
FASTER: higher pore pressures, static liquefaction, rushed constructed, less observed time; and
LONGER LASTING: Mine closure halts operation, but the waste facility still stands. Time-dependent deterioration
PXVWQRWEHH[FOXGHGIURPFRQVLGHUDWLRQ
6WDWLVWLFV KDYH UHSRUWHG WZR WR ¿YH WDLOLQJV GDP IDLOXUHV DQQXDOO\ IRU WKH URXJKO\ WDLOLQJV GDPV ZRUOGZLGH D
probability of 1 in 700 to 1 in 1,750) (LeProude, 2015; Davies et al., 2002a). Comparatively, the estimated annual pro-
bability of failure for a conventional dam is 1 in 10,000 (Davies et al., 2002a).
7KHWUHQGRIUHSRUWHGGHDWKVKDVDOVRGHFUHDVHGVLJQL¿FDQWO\DOWKRXJKWKHVDPHFDQQRWEHVDLGIRUDPRXQWRIWDLOLQJV
released, reiterating that there is still work to be done. The value of education and resilience in Chile has seen the
IDLOXUHRIWDLOLQJVGDPVGXHWRVHLVPLFOLTXHIDFWLRQUHGXFHG³IURPLQSUHFDVHVWR]HURLQSRVWFDVHV
WKH&KLOHDQHDUWKTXDNHRIPDJQLWXGHGLGQRWFDXVHDQ\IDLOXUH´$]DPDQG/L,QSHUVSHFWLYHWKH/D
/LJXDHDUWKTXDNHRIPDJQLWXGHRQ0DUFKH[KLELWHGWKHPRVWZLGHVSUHDGLQÀXHQFHZLWKUHSRUWHGGDP
failures, 200 reported deaths, and 2.5 million cubic metres of reported release as a result.
6LQFHWKHVWDWHRI0LQDV*HUDLVLQ%UD]LOKDVKDGHLJKWWDLOLQJVGDPIDLOXUHVVL[RIZKLFKDUHFRQVLGHUHGPDMRU
This state, roughly the size of France, is responsible for most of the country’s iron ore production, and thus has the
biggest concentration of tailings dams in Brazil. Most of the failures were found to be related to liquefaction as the
FRDUVHVDQG\SURSHUWLHVRILURQRUHWDLOLQJVH[KLELWFKDUDFWHULVWLFDOO\KLJKVXVFHSWLELOLW\WROLTXHIDFWLRQ7KHXSVWUHDP
method is the predominant method for dam construction in Brazil. Also, as Brazil was mostly considered non-seismic
until recently, dynamic analyses are still incipient in the country and there are no acts or regulations that require dyna-
mic loadings to be taken into account in dam design.
In recognising cases such as that where Brazil is unfamiliar to design considerations to which Chile has advanced,
geotechnical information pertinent to increasing the knowledge and understanding of in-situ failure risk analysis should
be shared in consideration of safety, livelihood, reputation, and best standard practice. This is anticipated to allow a
global reach of geotechnical engineers to design, construct, operate, and maintain these structures responsibly. A
communal approach from all tiers of geotechnical personnel for the betterment of stakeholder safety embraces mul-
WLSOHEHQH¿WV$FWLQJLQDFFRUGDQFHZLWKWKLVUHFRPPHQGDWLRQWKLVUHVHDUFKSDSHUGLVFXVVHVORFDOLVHGWDLOLQJVGDP
failure case studies from Samarco Mine in Minas Gerais, Brazil. With a focus on state-of-the-art monitoring techniques,
LQSDUWLFXODUUDGDUGHIRUPDWLRQPRQLWRULQJWKHLQWHQWLVWRGHWDLOKRZEDFNDQDO\VHGLQWHUSUHWDWLRQRIH[LVWLQJPRQLWRULQJ
DQGLQVWUXPHQWDWLRQFDQEHQH¿WIXWXUHGDPPRQLWRULQJ,PSURYHGVWUDWHJLHVLQWHUPVRIFULWLFDOSDUDPHWHUWULJJHUOHYHOV
and ultimately the ability to anticipate and respond promptly to tailings dam instabilities are anticipated.
MINE BACKGROUND
)RXQGHGLQ6DPDUFRPLQHLURQRUHLQ0LQDV*HUDLV%UD]LOWRSURGXFHDQGH[SRUWSHOOHWVWRFRXQWULHVLQWKH
Americas, the Middle East, Asia, and Europe. Samarco currently operates under a 50-50 joint venture arrangement
between Vale S.A. and BHP Billiton Brasil Ltda.
At the time of this paper, Samarco was embedded in, and committed to the social, environmental, and economic
recovery of regions impacted by failure of the Fundão dam in late 2015. As part of this recovery, and in line with the
recommendations introduced, Samarco bears a self-induced responsibility and obligation to share learnings. An open
discussion with both national and international peers increases awareness of the risks, and how interpretation of pre-
vious events at this mine can contribute to a safer environment.
SANTARÉM DAM
Santarém Dam is located 3 km downstream of Fundão dam (Figure 1). Santarém Dam is designed as a civil gravity
GDPDFRQFUHWHVWUXFWXUHUHWDLQLQJERWKZDWHUDQGWDLOLQJV2QIDLOXUHRI)XQGmRGDP6DQWDUpP'DPRYHUWRSSHG
and sustained structural damage, but did not fail. To ensure future integrity of the dam, Samarco have constructed an
HDUWKHQEXWWUHVVWRUHLQIRUFHWKHGRZQVWUHDPVLGH5DGDUPRQLWRULQJGHWHFWHGDPLQRUORFDOLVHGIDLOXUHGXULQJFRQV
truction.
Dike 2
Dike 2 is located 500 m upstream of Dike 1 of the Fundão dam (Figure 1). Internal to the Fundão reservoir, Dike 2 was
designed to retain the slimes component of the tailings, while sands were retained by Dike 1, both of which failed in
WKHHYHQW2IWKHP3 of material remaining within Fundão after the failure, the majority were slimes
situated upstream of the breached Dike 2. In January 2016, 1,000,000 m3 of this material was mobilised from the
reservoir as a result of heavy rainfall, which carried downstream uninterrupted to overtop Santarém Dam. With repair
works underway on the downstream Dike 1, the risk of failure of the adjacent Selinha Dike, surrounding natural slopes,
RUSURJUHVVLRQRIIDLOXUHRI'LNHZDVFRQVLGHUHGPDMRUZLWKORVVRIOLIHDSRWHQWLDOFRQVHTXHQFH5DGDUPRQLWRULQJ
detected several minor events, of which one particular slimes collapse will be assessed.
Natural Slope
7KHQDWXUDOVORSHLVORFDWHGPGRZQVWUHDPRI)XQGmRGDP)LJXUH2QUHFRYHU\RIWKH)XQGmRGDPIDLOXUH
DFRႇHUGDPZDVEHLQJFRQVWUXFWHGZLWKWKHLQWHQWWRGUDLQDQGDOORZUHSDLUZRUNVWRWDNHSODFHGRZQVWUHDP9LVXDO
LQVSHFWLRQRIWKHZRUNVLWHLGHQWL¿HGDSRWHQWLDOFLUFXODUVKHDUIDLOXUHIRUPLQJRQWKHDGMDFHQWQDWXUDOURFNVORSH:LWK
SHUVRQQHOZRUNLQJEHORZWKHDUHDULVNDVVHVVPHQWLGHQWL¿HGORVVRIOLIHDVDSRWHQWLDOFRQVHTXHQFHRIIDLOXUH1RGDWD
ZDVDYDLODEOHRQWKHPDWHULDOFKDUDFWHULVWLFVRIWKHURFN5DGDUPRQLWRULQJGHWHFWHGDQGSURYLGHGVXႈFLHQWZDUQLQJ
for failure of the circular shear failure during monitoring.
Back-analysis of in-situ behaviours represents one of the most valuable approaches in establishing movement trig-
gers. For slope stability assessment, in-situ movement data may be collected by instruments embedded in the sub-
VXUIDFHVXFKDVLQFOLQRPHWHUVRUIURPWKHVXUIDFHE\H[WHQVRPHWHUVVXUYH\SRLQWVVORSHVWDELOLW\UDGDURU/L'$5
6XU¿FLDOUHDGLQJVFDQEHFRQYHQLHQWDQGUHDGLO\FROOHFWHGLQGLFDWRUVRILQWHUQDOGHIRUPDWLRQ7KH\PD\DOVREHXVHGWR
FRQ¿UPDQGRURSWLPLVHFRPSXWHUPRGHOOLQJRIDQWLFLSDWHGEHKDYLRXUVLQFOXVLYHRIVLWHJHRWHFKQLFDOYDULDELOLW\
([LVWLQJUDGDUPRYHPHQWWKUHVKROGVDW6DPDUFRDUHEDVHGRQEDFNDQDO\VLVRIDPLQRUIDLOXUHWKDWZDVFDSWXUHGLQ
Santarém Dam. The basis of selection was those thresholds that allowed for greater warning prior to the event occu-
rring. The threshold values for radar monitoring, and responses for each threshold level are summarised in Table I. It
is important to acknowledge that movement velocities are reported as opposed to magnitude, as it is the accelerated
change of movement that is the critical indicator of failure in the short term, with large movements over a longer dura-
WLRQRIWLPHDEOHWREHSUHGLFWHGUHPHGLDWHGRURWKHUZLVHUHPDLQLQJJOREDOO\VWDEOH5HJDUGOHVVRIWKHWULJJHUOHYHO
WKH*HRWHFKQLFDO(QJLQHHUFDQHYDFXDWHWKHVWUXFWXUHVDQGRUWULJJHU6DPDUFR¶V(PHUJHQF\$FWLRQ3ODQDWDQ\WLPH
7DEOH,5HVSRQVHSODQIRUWULJJHUOHYHOVLQFOXVLYHRIQRPLQDWHGUDGDUPRYHPHQWWKUHVKROGV
Yellow 2UDQJH 5HG
5DGDU PRYH
Trigger
*At Samarco, contingency is applied in the monitoring and instrumentation strategy in order to allow for this vali-
GDWLRQSURFHVV)RUH[DPSOHDXWRPDWHGUHDGLQJVIURPWHOHPHWU\LQVWDOOHGRQSLH]RPHWHUVZRXOGEHYDOLGDWHGE\
PDQXDOLQVSHFWLRQDQG6ORSH6WDELOLW\5DGDU665PRQLWRULQJZRXOGEHYDOLGDWHGDJDLQVW,QWHUIHURPHWULF6\QWKHWLF
$SHUWXUH5DGDU,Q6$5
7KHPRVWUHFHQWUHOHDVHRI3RUWDULDDWWLPHRIWKLVSDSHUZDV3RUWDULDSXEOLVKHGRQ0D\7KLVUHOHDVH
ZDVSDVVHGLQFROODERUDWLRQZLWK6DPDUFRZKHUHGH¿QLWLRQRIWKHQHZLQGXVWU\VWDQGDUGIRUPRQLWRULQJLVFRPSOHPHQ
WHGE\WKHRXWFRPHVDQGOHDUQLQJVIURPWKHIDLOXUHWKHPRVWVLJQL¿FDQWHYHQWIRU%UD]LOLQUHFRUGHGKLVWRU\:KLOH
the monitoring requirements from Portaria 416 remain, the following requirements have been added:
• Every dam must have a monitoring system that allows the company to assess dam safety;
7KHFRPSOH[LW\RIWKLVV\VWHPLVGHSHQGHQWRQGDPIDLOXUHFRQVHTXHQFH
• For dams deemed of high consequence, the monitoring system must be real time, and must have video came-
ras, both monitored 24 hours a day, seven days a week; and
7KLVPRQLWRULQJV\VWHPDPRQJVWRWKHUHOHPHQWVPXVWEHDVVHVVHGIRUDOOVLWHGDPVWZLFHSHU\HDUE\DQH[
ternal consultant.
3. MONITORING TECHNIQUE AND DATA TREND ANALYSIS
CALCULATION PERIOD
,QWHUPVRIPRQLWRULQJDQGLQVWUXPHQWDWLRQFDOFXODWLRQSHULRGVGH¿QHWKHWLPHGXUDWLRQRYHUZKLFKWKHGHIRUPDWLRQ
GDWDLVDYHUDJHGLHWKHWHPSRUDOUHVROXWLRQRIWKHGDWD7KHUHLVRIWHQDQH[FKDQJHEHWZHHQVSDWLDODQGWHPSRUDO
resolutions; a lesser temporal resolution (considering data averaged over longer time periods) will give smooth data,
but less accurate trending, while a greater temporal resolution has the potential to be distorted by data ‘noise’ in the
V\VWHPZKLFKDOVRKDVWKHSRWHQWLDOWRWULJJHUIDOVHWKUHVKROGDODUPVRQLVRODWHGGDWDVSLNHV)RUWKLVUHDVRQGLႇHUHQW
FDOFXODWLRQSHULRGVZHUHDVVHVVHGZLWKWKHLQWHQWWRHVWDEOLVKVHQVLWLYLW\RIWKHH[LVWLQJWKUHVKROGOHYHOVDW6DPDUFR
Mine, and the applicability of these in anticipating the failure event. These are represented by “Velocity over calculation
period” parameters in this report. These are compared against cumulative displacement of the scanned target: a direct
output of the radar technology.
Santarém Dam
A minor local failure was observed on the spillway area at Santarém Dam on 12 June, 2016, at which time the slope
was being monitored by slope stability radar (Figure 2). The material was displaced along a slip plane that formed two
thirds of the way up the second bench. In the months preceding, GroundProbe radar surveillance reports documen-
WHGQHJOLJLEOHGHIRUPDWLRQUDWHVWKDWZHUHQRWDVVRFLDWHGZLWKRSHUDWLRQVWUDႈ FRUZKHQFRUUHODWHGDJDLQVWUDLQIDOO
periods.
Figure 2 - Slope stability radar monitoring for Santarem Dam, showing location of local failure
7KHH[LVWLQJVLWHWKUHVKROGVIRUYHORFLW\SURYHGHႇHFWLYHLQWKHLUDELOLW\WRSUHGLFWORFDOIDLOXUHLQSDUWLFXODUZKHQXVLQJ
WKH³9HORFLW\RYHU´FDOFXODWLRQSHULRG)LJXUH7KLVSURYLGHGZDUQLQJDW¿YHKRXUVWKUHHKRXUVDQGRQHKRXUIRU
yellow, orange, and red alarms, respectively (Table II). In consideration of the responses designated for each alarm as
seen in Table I, this degree of notice presents a safe and feasible response time.
7KH UHJUHVVLYH SURJUHVVLYH WUHQGV H[KLELWHG PRYHPHQW IRU DOPRVW WZR GD\V SULRU WR IDLOXUH 'XULQJ WKH UHJUHVVLYH
WUHQGLQJVWDJHD\HOORZDODUPZDVWULJJHUHGIRUDVLQJOHKRXU,QFRPSDULVRQWRWULJJHUVRIPPKDQGPPKZKLFK
ZRXOGKDYHEHHQFDXWLRQLQJIRU¿YHKRXUVDQGHLJKWKRXUVUHVSHFWLYHO\WKLVDODUPSURYHGHႇHFWLYHLQDFNQRZOHGJLQJ
the increase yet not escalating the alarm beyond what a regressive trend warrants.
)XUWKHU WKH ULVN RI XQHYHQWIXO UHDGLQJV LQ VWDEOH RSHUDWLQJ FRQGLWLRQV H[LVWV ZLWK ORZHU DODUP WKUHVKROGV ,Q FRQVL
deration of all velocity calculation periods over the entire duration that the same radar that captured that failure was
RSHUDWLQJPPKZRXOGEHDODUPLQJRQDYHUDJHRIVWDEOHRSHUDWLQJWLPHDQGPPKIRU
Figure 3 - Velocity and deformation data trending for Santarém Dam, velocity alarm thresholds shown
5HDGLQJVGHVLJQDWHGDV³³ZHUHQRWLGHQWL¿HGSULRUWRWKHHYHQW
Dike 2
$FROODSVHRIGHEULVZDVREVHUYHGWRSURJUHVVDWWKHWRHRI'LNHRYHUWKHSHULRGRIVL[GD\VIURP$XJXVWWR
$XJXVW)LJXUH'HEULVÀRZZDVGLVSODFHGGRZQVWUHDPLQDSURJUHVVLYHIDLOXUHWUHQGZKHUHWKHUDWHRI
GLVSODFHPHQWFRQWLQXHGWRLQFUHDVHXQWLOFROODSVHDSSUR[LPDWHO\RIWKHWRWDOPDWHULDOGHIRUPDWLRQRFFXUUHGRQWKH
VL[WKGD\1RVLJQL¿FDQWUDLQIDOORFFXUUHGGXULQJRUSULRUWRWKLVHYHQW
It is anticipated that loose, saturated slimes comprising high proportions of clay- and silt-sized particles remained after
WKHFROODSVH7KHPRLVWXUHFRQWHQWRIWKHVOLPHVLVW\SLFDOO\DURXQGZKLOHWKHOLTXLGOLPLWLVDSSUR[LPDWHO\
LQGLFDWLQJORZFRQVLVWHQF\%DVHGRQWKHUDGDUGHIRUPDWLRQWUHQGVDQGYLVXDODVVHVVPHQWRIWKHVXUURXQGLQJ
DUHDLWLVWKHDXWKRU¶VRSLQLRQWKDWK\GUDXOLFIRUFHVLQGXFHGE\XSVWUHDPUXQRႇÀRZVSUHIHUHQWLDOO\GHJUDGHGDSDWKZD\
through the material; the same forces that failed an upstream earthen buttress by erosion piping two months prior.
In the author’s assessment, mass collapse of the material may have been observed as a result of progressive erosion
RIWKHVXUURXQGLQJEDQNVGULYHQE\GHFUHDVLQJFRQ¿QLQJVWUHVVHVLQFRPELQDWLRQZLWKZDWHUUHFKDUJHHQFRXUDJLQJ
an elevated moisture content and the associated decrease in shear strength of the slimes. In terms of the data, this
mechanism could be represented by the process of shear stress gradually reducing the shear strength until the point
DWZKLFKWKHVOLSWULJJHUVDVUHÀHFWHGE\FRPSDUDWLYHO\HOHYDWHGUDWHVRIGHIRUPDWLRQ
1RWDEO\GHIRUPDWLRQWUHQGLQJLQWKHVXUURXQGLQJDUHDVVXJJHVWVVWHDG\FUHHSZKLFKLVDQWLFLSDWHGWREHUHÀHFWLYHRI
WKHVDPHLQÀXHQFHVWKDWIDLOHGWKHDGMDFHQWPDWHULDODQGPD\EHHDUO\VWDJHLQGLFDWRUVRIDVLPLODUHYHQW
,QWKHPRQWKSUHFHGLQJ*URXQG3UREHUDGDUVXUYHLOODQFHUHSRUWVGRFXPHQWHGQRHYLGHQFHRIVLJQL¿FDQWGHIRUPDWLRQ
SURFHVVHVZLWKGHIRUPDWLRQUDWHVFRQVLVWHQWO\EHORZPPGD\
Figure 4 - Slope stability radar monitoring for Dike 2, showing location of slimes collapse
7KHH[LVWLQJVLWHWKUHVKROGVIRUYHORFLW\SURYHGHႇHFWLYHLQWKHLUDELOLW\WRSUHGLFWFROODSVHRIWKHVOLPHVDW'LNHSURYL
GLQJZDUQLQJDWXSWRKRXUVKRXUVDQGKRXUVIRU\HOORZRUDQJHDQGUHGDODUPVUHVSHFWLYHO\)LJXUHDQG
Table V). This degree of notice presents a safe and feasible response time. Unlike Santarém Dam, this failure was
VORZDQGGXFWLOHH[KLELWLQJPRYHPHQWIRUVL[DQGDKDOIGD\VSULRUWRIDLOXUHDGXUDWLRQWKDWVDZWKHODUJHUFDOFXODWLRQ
SHULRGVVXFKDV9HORFLW\RYHUDQG9HORFLW\RYHUSURYHPRVWHႇHFWLYHLQIRUHFDVWLQJWKHIDLOXUH
Figure 5 – Velocity and deformation data trending for Dike 2, velocity alarm thresholds shown
The slope material comprised large quantities of loose rock, initially intact but anticipated to have weathered and de-
graded over time. Two primary mechanisms are anticipated to have driven the regressive-progressive failure of the
QDWXUDOVORSHEUHDNXSRIURFNDWWKHWRHRIWKHVORSHLQDGGLWLRQWRDQLQFUHDVHLQZDWHUSUHVVXUH:LWKFRPSOH[LW\
DGGHGDVDUHVXOWRILQWHUDFWLRQEHWZHHQWKHVHWZRGULYHUVLWLVREVHUYHGWKDWWKHIDLOHGDQGH[SRVHGVHFWLRQVRIWKH
slope became susceptible to further degradation from weathering, in turn entering a cycle of progressive instability.
This is evidenced by the successive recorded events, all of similar behaviour, in addition to the incremental trend of
movement along a preferential deformation path down the side of the slope (Figure 6).
)LJXUH6ORSHVWDELOLW\UDGDUPRQLWRULQJIRU(L[RLQ1RYHPEHUVKRZLQJORFDWLRQRIFLUFXODUVOLSIDLOXUH
7KHH[LVWLQJVLWHWKUHVKROGVIRUYHORFLW\SURYHGLQHႇHFWLYHLQWKHLUDELOLW\WRSUHGLFWORFDOIDLOXUHRI(L[R9HORFLW\RYHU
60 triggered a yellow alarm half an hour prior to failure, however could easily be missed if read without the knowledge
that a failure were about to occur. No orange alarms were triggered.
Alternate metrics are recommended as additional control measures. Inverse velocity was investigated to also be in-
HႇHFWLYHVRDQDVVHVVPHQWRIGHIRUPDWLRQWULJJHUVZDVFRQGXFWHG)LJXUHDQG7DEOH,,,'HIRUPDWLRQWULJJHUVDUH
XVHIXOLQGLVDJJUHJDWLQJGDWDQRLVHIURPVPDOOGHIRUPDWLRQGDWDDQGKHQFHZLOOSUHVHQWVOLJKWO\GLႇHUHQWUHVXOWVWKDQ
the velocity analysis.
)LJXUH'HIRUPDWLRQGDWDWUHQGLQJIRU(L[R
7DEOH,,,'HIRUPDWLRQDODUPDVVHVVPHQWIRU(L[R
Time before event (hours)
1RWH*UH\HGRXWFHOOVLQGLFDWHOHVVXVHIXOUHVXOWVZKHUHWKHVHWULJJHUVZHUHH[FHHGHGRQLQLWLDWLRQRIWKHUHJUHVVLRQ
trend, and remained above the trigger level until failure.
7KHDELOLW\WRFRQ¿GHQWO\IRUHFDVWGHYHORSLQJPRYHPHQWLQWKHSURJUHVVLYHWUHQGXVLQJWKHDSSURDFKGHVFULEHGLQ)L
JXUHZDVQRWSRVVLEOHXQWLODWOHDVW¿YHVFDQVLQWRWKHWUHQG7KLVZDVEHFDXVHDGHIRUPDWLRQWUHQGWKDWLVDYHUDJLQJ
movement over previous hours will capture the regressive trend prior, lagging current movement trends by a number
of scans. To eliminate this, the deformation data would need to be averaged over 24 hours, which is less practical for
IRUHFDVWLQJVKRUWWHUPHYHQWVVXFKDVWKH(L[RHYHQWDVWKHGDWDZLOOVLPSO\UHÀHFWWKHGHIRUPDWLRQWUHQGLWVHOI
&DOFXODWHGUDWHVIRUPPKRXUVSUHVHQWHGWKUHHRIWKHPRUHGHVLUDEOHZDUQLQJVDWDQGPLQXWHVSULRUWR
IDLOXUH7KHULVNH[LVWVWKDWXQZDQWHGDODUPVZLOOWULJJHUDWVXFKORZUDWHVKRZHYHUDVVHVVPHQWUHSRUWHGWKDWIRUWKH
PRQWKSULRUWRLQLWLDWLRQRIWKLVHYHQWRIRSHUDWLQJWLPHZRXOGEHDODUPLQJIRUPPKRXUVDQGIRUDQG
PPKRXUV
It is anticipated that geotechnical engineering judgement will be applied to the deformation assessment approach,
both on setup for conditional suitability of the alarm, and on response should it trigger. While this deformation trigger
PD\EHDSSURSULDWHIRU(L[RLWPD\DODUPPRUHIUHTXHQWO\ZKHQPRQLWRULQJVORSHVRIGLႇHUHQWPDWHULDOSURSHUWLHV,Q
contrast, when the deformation alarm is applied to stable operating conditions for Santarém Dam and Dike 2 (analysed
RYHUWKHHQWLUHGXUDWLRQRIGDWDFDSWXUHDWWKHVHVLWHVXQHYHQWIXODODUPVIRUPPKRXUVDUHSUHVHQWHGDQG
RIWKHWLPHEXWZRXOGWULJJHUDSSUR[LPDWHO\KRXUVDQGKRXUVSULRUWRIDLOXUHUHVSHFWLYHO\
4. DISCUSSION
Opportunities
Pertinent opportunities that can be derived from the data from the three events comprise:
([LVWLQJVLWHWKUHVKROGVIRUGHIRUPDWLRQYHORFLWLHVZHUHHႇHFWLYHLQSUHGLFWLQJWZRRIWKHWKUHHHYHQWVXSWR
hours prior to failure. The triggers allowed for, at minimum, one hours’ warning on red alarm prior to failure;
)RUWKH(L[RIDLOXUHGHIRUPDWLRQVZHUHQRWRIJUHDWHQRXJKPDJQLWXGHWRDVVXUHWKDWWKHYHORFLW\WULJJHUV
were progressing toward a failure event. In this scenario, a second deformation metric was recommended in
order to validate the velocity triggers which allowed 1.4 and 1 hours’ warning on yellow and orange alarms,
respectively;
)RUHFDVWLQJRIVORSHLQVWDELOLWLHVLQVXႈFLHQWWLPHDOORZVWKHJHRWHFKQLFDOHQJLQHHURURWKHUZLVHWRSUHHPSWL
vely identify and respond appropriately to slope instability;
• The value in real-time monitoring techniques is realised where failure progression over a period of two through
WRHLJKWGD\VDVLQWKHFDVHVDQDO\VHGPD\QRWEHLGHQWL¿HGRQDELPRQWKO\LQVSHFWLRQUHJLPHIRUH[DPSOH
and
,WLVQRWFRVWHႇHFWLYHWRKDYHVRPHUHDOWLPHVROXWLRQVRSHUDWLQJLQVWDEOHFRQGLWLRQV9LVXDOREVHUYDWLRQV
DUHUHFRPPHQGHGDVWKH¿UVWDQGIRUHPRVWFRQWUROPHDVXUHWREHLPSOHPHQWHG7KHUHLVLPPHQVHYDOXHLQ
training operational personnel on geotechnical hazards; hundreds of eyes on the ground that can identify and
DGYHUWLVHGHWHULRUDWLQJFRQGLWLRQVDUHLQYDOXDEOHDQGH[LVWLQJDVVHWV
Limitations
7KHJUHDWHVWYDOXHLQPRQLWRULQJDQGLQVWUXPHQWDWLRQGDWDLVLQWKHWHFKQLTXHRIDQDO\VLVGULYHQE\RSHUDWRUH[SHULHQ
ce, understanding of site conditions, and how these relate to an understanding of the plausible failure mechanisms
DQGWULJJHUVIRUGLႇHUHQWDUHDVRIWKHPLQH7KLVH[SHULHQFHPXVWEHOHYHUDJHGLQRUGHUWRDFFRXQWIRUJHRWHFKQLFDO
variability, making appropriate considerations on the prediction of failure for:
2QHFRQVHUYDWLYHWULJJHUYDOXHIRUHDFKWKUHVKROGDSSOLHGDFURVVWKHHQWLUHVLWHPD\QRWEHHႇHFWLYHIRUGLႇH
UHQWW\SHVRIVORSHVLHDEULWWOHURFNIDLOXUHIURPDVXEYHUWLFDOZDOOZLOOGHIRUPDQGIDLOGLႇHUHQWO\WKDQGXFWLOH
ÀRZRIGHEULVRQDJHQWO\JUDGLQJVORSHDQG
• Wet weather, although often associated with geotechnical instabilities, can also produce ‘noisy’ radar data on
VPDOOHUFDOFXODWLRQSHULRGVLQDGGLWLRQWRIDOVLI\LQJKRZFORVHWKHPDWHULDOLVWRWKHUDGDUDVVXU¿FLDOZDWHUUXQRႇ
is recorded as movement during the scan.
Understandably, monitoring with no history of collapse or benchmarking data is challenging, however not redundant.
7KHH[LVWLQJGDWDDWDQ\VLWHLVRILPPHQVHYDOXHHYHQLIQRIDLOXUHKDVRFFXUUHGWKHGDWDSUHVHQWVDUDQJHRIFDVHV
ZLWKLQWKHIDLOXUHHQYHORSHZKHUHWKHWKUHVKROGYDOXHFDQEHVHWDVWKHKLJKHVWH[LVWLQJREVHUYHGUDWHDQGPRYHPHQW
EH\RQGWKLVUHTXLUHVVXႈFLHQWDWWHQWLRQXQWLOWKHPHFKDQLVPVDQGHႇHFWVDUHDSSURSULDWHO\XQGHUVWRRG
Developing a database of site events, their trending, and the associated implications is a tool readily accessible to
civil, mine, and geotechnical operators, globally. The more back-analysis that is completed on these events, the more
HPSLULFDOFRQ¿GHQFHFDQEHKHOGDURXQGXQGHUVWDQGLQJGLႇHUHQWW\SHVRIVLWHIDLOXUHDQGHQVXULQJWKDWDSSURSULDWH
thresholds are in place to anticipate and respond to events appropriately.
5. CONCLUSION
7KLVSDSHUGHVFULEHVDEDFNDQDO\VLVRIWKUHHGLႇHUHQWVORSHLQVWDELOLWLHVDW6DPDUFR0LQH%UD]LO7KHDQDO\VLVDV
VHVVHGVXLWDELOLW\RIH[LVWLQJWULJJHUVPDGHUHFRPPHQGDWLRQVIRUDGGLWLRQDOPHWULFVDQGFRQVHTXHQWO\DGGHGWRWKH
database of failure analyses.
7KH\HOORZRUDQJHDQGUHGDODUPWULJJHUVIRUUDGDUPRQLWRUHGVORSHGHIRUPDWLRQYHORFLWLHVDUHVHWWRPPKPPK
DQGPPKDW6DPDUFR0LQH)RUWKHORQJHUGXUDWLRQHYHQWVWKHVHDODUPVSURYLGHGVXႈ FLHQWZDUQLQJWLPHWRDOORZ
appropriate responses to take place, with evacuation of the slope area prompted at least 1.3 hours prior to failure. For
WKHVKRUWHUGXUDWLRQHYHQWZKLFKLQWKLVFDVHZDVDOVRRIORZHUPDJQLWXGHWKHWULJJHUVZHUHLGHQWL¿HGWREHSRWHQWLDOO\
missed. A second level of check was introduced in form of a deformation alarm, which aids to validate the velocity
reading when used under the correct application.
7KHDQDO\VHVSURYLGHGDGGLWLRQDOFRQ¿GHQFHWKDWUHDOWLPHPRQLWRULQJRIVORSHVWDELOLW\FDQDLGLQSUHGLFWLQJIDLOXUH
up to 42 hours prior to the event. The recommendation is made that applied understanding of failure mechanisms,
back-analysis of monitoring and instrumentation data in terms of this, and ultimately the development of threshold
triggers to prompt appropriate response should be readily integrated with the dam monitoring strategy.
In developing the dam monitoring strategy, case studies shared by other operators should be considered. With a safety
IRFXVVLPLODULWLHVEHWZHHQVLWHFRQGLWLRQVQHZOHDUQLQJVRUXQH[SHFWHGEHKDYLRXUVDUHRIYDOXHWROHYHUDJH:KHUH
previous commercial reluctance has inhibited the value of knowledge sharing, greater geotechnical community colla-
ERUDWLRQLVUHFRPPHQGHGWRDLGLQPLQLPLVLQJUHSHDWHUURUVLQIDYRXURIPD[LPLVHGGLOLJHQFH
6. REFERENCES
$XVWUDOLDQ*RYHUQPHQW'HSDUWPHQWRI,QGXVWU\7RXULVPDQG5HVRXUFHV7DLOLQJV0DQDJHPHQW&DQEHUUD
$]DP6 /L4µ7DLOLQJV'DP)DLOXUHV$5HYLHZRIWKH/DVW<HDUV¶*HRWHFKQLFDO1HZV:DVWH*(2
technics, p.50-53
• Davies, M, Martin, D & Lighthall, P 2002a, ‘Mine Tailings Dams - When Things go Wrong’
/H3RXGUH'&([DPSOHV6WDWLVWLFVDQG)DLOXUHPRGHVRIWDLOLQJVGDPVDQGFRQVHTXHQFHVRIIDLOXUH3RZHU
3RLQWVOLGHV5(07(&+
5REHUWVRQ$0)0($5LVN$QDO\VLV)DLOXUH0RGHVDQG(ႇHFWV$QDO\VLV3RZHU3RLQWVOLGHV
• National Department of Mineral Production 2012, ‘Portaria 416’, Ministry of Mines and Energy, Brazil.
1DWLRQDO'HSDUWPHQWRI0LQHUDO3URGXFWLRQµ3RUWDULD¶0LQLVWU\RI0LQHVDQG(QHUJ\%UD]LO
• Darling 2011, SME Mining Engineering Handbook…
)HOO50DF*UHJRU36WDSOHGRQ'%HOO* )RVWHU0*HRWHFKQLFDO(QJLQHHULQJRI'DPVÜQGHGQ&5&
3UHVV%DONHPD7KH1HWKHUODQGV